fs/fscache/stats.c: fix memory leak
[linux-2.6.git] / fs / namespace.c
index 3dbfc07..4e46539 100644 (file)
@@ -9,30 +9,17 @@
  */
 
 #include <linux/syscalls.h>
-#include <linux/slab.h>
-#include <linux/sched.h>
-#include <linux/spinlock.h>
-#include <linux/percpu.h>
-#include <linux/init.h>
-#include <linux/kernel.h>
-#include <linux/acct.h>
+#include <linux/export.h>
 #include <linux/capability.h>
-#include <linux/cpumask.h>
-#include <linux/module.h>
-#include <linux/sysfs.h>
-#include <linux/seq_file.h>
 #include <linux/mnt_namespace.h>
 #include <linux/namei.h>
-#include <linux/nsproxy.h>
 #include <linux/security.h>
-#include <linux/mount.h>
-#include <linux/ramfs.h>
-#include <linux/log2.h>
 #include <linux/idr.h>
-#include <linux/fs_struct.h>
-#include <linux/fsnotify.h>
-#include <asm/uaccess.h>
-#include <asm/unistd.h>
+#include <linux/acct.h>                /* acct_auto_close_mnt */
+#include <linux/ramfs.h>       /* init_rootfs */
+#include <linux/fs_struct.h>   /* get_fs_root et.al. */
+#include <linux/fsnotify.h>    /* fsnotify_vfsmount_delete */
+#include <linux/uaccess.h>
 #include "pnode.h"
 #include "internal.h"
 
@@ -78,7 +65,7 @@ static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry)
  * allocation is serialized by namespace_sem, but we need the spinlock to
  * serialize with freeing.
  */
-static int mnt_alloc_id(struct vfsmount *mnt)
+static int mnt_alloc_id(struct mount *mnt)
 {
        int res;
 
@@ -95,7 +82,7 @@ retry:
        return res;
 }
 
-static void mnt_free_id(struct vfsmount *mnt)
+static void mnt_free_id(struct mount *mnt)
 {
        int id = mnt->mnt_id;
        spin_lock(&mnt_id_lock);
@@ -110,7 +97,7 @@ static void mnt_free_id(struct vfsmount *mnt)
  *
  * mnt_group_ida is protected by namespace_sem
  */
-static int mnt_alloc_group_id(struct vfsmount *mnt)
+static int mnt_alloc_group_id(struct mount *mnt)
 {
        int res;
 
@@ -129,7 +116,7 @@ static int mnt_alloc_group_id(struct vfsmount *mnt)
 /*
  * Release a peer group ID
  */
-void mnt_release_group_id(struct vfsmount *mnt)
+void mnt_release_group_id(struct mount *mnt)
 {
        int id = mnt->mnt_group_id;
        ida_remove(&mnt_group_ida, id);
@@ -138,9 +125,42 @@ void mnt_release_group_id(struct vfsmount *mnt)
        mnt->mnt_group_id = 0;
 }
 
-struct vfsmount *alloc_vfsmnt(const char *name)
+/*
+ * vfsmount lock must be held for read
+ */
+static inline void mnt_add_count(struct mount *mnt, int n)
+{
+#ifdef CONFIG_SMP
+       this_cpu_add(mnt->mnt_pcp->mnt_count, n);
+#else
+       preempt_disable();
+       mnt->mnt_count += n;
+       preempt_enable();
+#endif
+}
+
+/*
+ * vfsmount lock must be held for write
+ */
+unsigned int mnt_get_count(struct mount *mnt)
+{
+#ifdef CONFIG_SMP
+       unsigned int count = 0;
+       int cpu;
+
+       for_each_possible_cpu(cpu) {
+               count += per_cpu_ptr(mnt->mnt_pcp, cpu)->mnt_count;
+       }
+
+       return count;
+#else
+       return mnt->mnt_count;
+#endif
+}
+
+static struct mount *alloc_vfsmnt(const char *name)
 {
-       struct vfsmount *mnt = kmem_cache_zalloc(mnt_cache, GFP_KERNEL);
+       struct mount *mnt = kmem_cache_zalloc(mnt_cache, GFP_KERNEL);
        if (mnt) {
                int err;
 
@@ -154,7 +174,17 @@ struct vfsmount *alloc_vfsmnt(const char *name)
                                goto out_free_id;
                }
 
-               atomic_set(&mnt->mnt_count, 1);
+#ifdef CONFIG_SMP
+               mnt->mnt_pcp = alloc_percpu(struct mnt_pcp);
+               if (!mnt->mnt_pcp)
+                       goto out_free_devname;
+
+               this_cpu_add(mnt->mnt_pcp->mnt_count, 1);
+#else
+               mnt->mnt_count = 1;
+               mnt->mnt_writers = 0;
+#endif
+
                INIT_LIST_HEAD(&mnt->mnt_hash);
                INIT_LIST_HEAD(&mnt->mnt_child);
                INIT_LIST_HEAD(&mnt->mnt_mounts);
@@ -166,13 +196,6 @@ struct vfsmount *alloc_vfsmnt(const char *name)
 #ifdef CONFIG_FSNOTIFY
                INIT_HLIST_HEAD(&mnt->mnt_fsnotify_marks);
 #endif
-#ifdef CONFIG_SMP
-               mnt->mnt_writers = alloc_percpu(int);
-               if (!mnt->mnt_writers)
-                       goto out_free_devname;
-#else
-               mnt->mnt_writers = 0;
-#endif
        }
        return mnt;
 
@@ -216,32 +239,32 @@ int __mnt_is_readonly(struct vfsmount *mnt)
 }
 EXPORT_SYMBOL_GPL(__mnt_is_readonly);
 
-static inline void inc_mnt_writers(struct vfsmount *mnt)
+static inline void mnt_inc_writers(struct mount *mnt)
 {
 #ifdef CONFIG_SMP
-       (*per_cpu_ptr(mnt->mnt_writers, smp_processor_id()))++;
+       this_cpu_inc(mnt->mnt_pcp->mnt_writers);
 #else
        mnt->mnt_writers++;
 #endif
 }
 
-static inline void dec_mnt_writers(struct vfsmount *mnt)
+static inline void mnt_dec_writers(struct mount *mnt)
 {
 #ifdef CONFIG_SMP
-       (*per_cpu_ptr(mnt->mnt_writers, smp_processor_id()))--;
+       this_cpu_dec(mnt->mnt_pcp->mnt_writers);
 #else
        mnt->mnt_writers--;
 #endif
 }
 
-static unsigned int count_mnt_writers(struct vfsmount *mnt)
+static unsigned int mnt_get_writers(struct mount *mnt)
 {
 #ifdef CONFIG_SMP
        unsigned int count = 0;
        int cpu;
 
        for_each_possible_cpu(cpu) {
-               count += *per_cpu_ptr(mnt->mnt_writers, cpu);
+               count += per_cpu_ptr(mnt->mnt_pcp, cpu)->mnt_writers;
        }
 
        return count;
@@ -250,6 +273,15 @@ static unsigned int count_mnt_writers(struct vfsmount *mnt)
 #endif
 }
 
+static int mnt_is_readonly(struct vfsmount *mnt)
+{
+       if (mnt->mnt_sb->s_readonly_remount)
+               return 1;
+       /* Order wrt setting s_flags/s_readonly_remount in do_remount() */
+       smp_rmb();
+       return __mnt_is_readonly(mnt);
+}
+
 /*
  * Most r/o checks on a fs are for operations that take
  * discrete amounts of time, like a write() or unlink().
@@ -260,7 +292,7 @@ static unsigned int count_mnt_writers(struct vfsmount *mnt)
  */
 /**
  * mnt_want_write - get write access to a mount
- * @mnt: the mount on which to take a write
+ * @m: the mount on which to take a write
  *
  * This tells the low-level filesystem that a write is
  * about to be performed to it, and makes sure that
@@ -268,19 +300,20 @@ static unsigned int count_mnt_writers(struct vfsmount *mnt)
  * the write operation is finished, mnt_drop_write()
  * must be called.  This is effectively a refcount.
  */
-int mnt_want_write(struct vfsmount *mnt)
+int mnt_want_write(struct vfsmount *m)
 {
+       struct mount *mnt = real_mount(m);
        int ret = 0;
 
        preempt_disable();
-       inc_mnt_writers(mnt);
+       mnt_inc_writers(mnt);
        /*
-        * The store to inc_mnt_writers must be visible before we pass
+        * The store to mnt_inc_writers must be visible before we pass
         * MNT_WRITE_HOLD loop below, so that the slowpath can see our
         * incremented count after it has set MNT_WRITE_HOLD.
         */
        smp_mb();
-       while (mnt->mnt_flags & MNT_WRITE_HOLD)
+       while (mnt->mnt.mnt_flags & MNT_WRITE_HOLD)
                cpu_relax();
        /*
         * After the slowpath clears MNT_WRITE_HOLD, mnt_is_readonly will
@@ -288,12 +321,10 @@ int mnt_want_write(struct vfsmount *mnt)
         * MNT_WRITE_HOLD is cleared.
         */
        smp_rmb();
-       if (__mnt_is_readonly(mnt)) {
-               dec_mnt_writers(mnt);
+       if (mnt_is_readonly(m)) {
+               mnt_dec_writers(mnt);
                ret = -EROFS;
-               goto out;
        }
-out:
        preempt_enable();
        return ret;
 }
@@ -317,7 +348,7 @@ int mnt_clone_write(struct vfsmount *mnt)
        if (__mnt_is_readonly(mnt))
                return -EROFS;
        preempt_disable();
-       inc_mnt_writers(mnt);
+       mnt_inc_writers(real_mount(mnt));
        preempt_enable();
        return 0;
 }
@@ -351,17 +382,23 @@ EXPORT_SYMBOL_GPL(mnt_want_write_file);
 void mnt_drop_write(struct vfsmount *mnt)
 {
        preempt_disable();
-       dec_mnt_writers(mnt);
+       mnt_dec_writers(real_mount(mnt));
        preempt_enable();
 }
 EXPORT_SYMBOL_GPL(mnt_drop_write);
 
-static int mnt_make_readonly(struct vfsmount *mnt)
+void mnt_drop_write_file(struct file *file)
+{
+       mnt_drop_write(file->f_path.mnt);
+}
+EXPORT_SYMBOL(mnt_drop_write_file);
+
+static int mnt_make_readonly(struct mount *mnt)
 {
        int ret = 0;
 
        br_write_lock(vfsmount_lock);
-       mnt->mnt_flags |= MNT_WRITE_HOLD;
+       mnt->mnt.mnt_flags |= MNT_WRITE_HOLD;
        /*
         * After storing MNT_WRITE_HOLD, we'll read the counters. This store
         * should be visible before we do.
@@ -384,41 +421,69 @@ static int mnt_make_readonly(struct vfsmount *mnt)
         * MNT_WRITE_HOLD, so it can't be decremented by another CPU while
         * we're counting up here.
         */
-       if (count_mnt_writers(mnt) > 0)
+       if (mnt_get_writers(mnt) > 0)
                ret = -EBUSY;
        else
-               mnt->mnt_flags |= MNT_READONLY;
+               mnt->mnt.mnt_flags |= MNT_READONLY;
        /*
         * MNT_READONLY must become visible before ~MNT_WRITE_HOLD, so writers
         * that become unheld will see MNT_READONLY.
         */
        smp_wmb();
-       mnt->mnt_flags &= ~MNT_WRITE_HOLD;
+       mnt->mnt.mnt_flags &= ~MNT_WRITE_HOLD;
        br_write_unlock(vfsmount_lock);
        return ret;
 }
 
-static void __mnt_unmake_readonly(struct vfsmount *mnt)
+static void __mnt_unmake_readonly(struct mount *mnt)
 {
        br_write_lock(vfsmount_lock);
-       mnt->mnt_flags &= ~MNT_READONLY;
+       mnt->mnt.mnt_flags &= ~MNT_READONLY;
        br_write_unlock(vfsmount_lock);
 }
 
-void simple_set_mnt(struct vfsmount *mnt, struct super_block *sb)
+int sb_prepare_remount_readonly(struct super_block *sb)
 {
-       mnt->mnt_sb = sb;
-       mnt->mnt_root = dget(sb->s_root);
-}
+       struct mount *mnt;
+       int err = 0;
 
-EXPORT_SYMBOL(simple_set_mnt);
+       /* Racy optimization.  Recheck the counter under MNT_WRITE_HOLD */
+       if (atomic_long_read(&sb->s_remove_count))
+               return -EBUSY;
 
-void free_vfsmnt(struct vfsmount *mnt)
+       br_write_lock(vfsmount_lock);
+       list_for_each_entry(mnt, &sb->s_mounts, mnt_instance) {
+               if (!(mnt->mnt.mnt_flags & MNT_READONLY)) {
+                       mnt->mnt.mnt_flags |= MNT_WRITE_HOLD;
+                       smp_mb();
+                       if (mnt_get_writers(mnt) > 0) {
+                               err = -EBUSY;
+                               break;
+                       }
+               }
+       }
+       if (!err && atomic_long_read(&sb->s_remove_count))
+               err = -EBUSY;
+
+       if (!err) {
+               sb->s_readonly_remount = 1;
+               smp_wmb();
+       }
+       list_for_each_entry(mnt, &sb->s_mounts, mnt_instance) {
+               if (mnt->mnt.mnt_flags & MNT_WRITE_HOLD)
+                       mnt->mnt.mnt_flags &= ~MNT_WRITE_HOLD;
+       }
+       br_write_unlock(vfsmount_lock);
+
+       return err;
+}
+
+static void free_vfsmnt(struct mount *mnt)
 {
        kfree(mnt->mnt_devname);
        mnt_free_id(mnt);
 #ifdef CONFIG_SMP
-       free_percpu(mnt->mnt_writers);
+       free_percpu(mnt->mnt_pcp);
 #endif
        kmem_cache_free(mnt_cache, mnt);
 }
@@ -428,20 +493,20 @@ void free_vfsmnt(struct vfsmount *mnt)
  * @dir. If @dir is set return the first mount else return the last mount.
  * vfsmount_lock must be held for read or write.
  */
-struct vfsmount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry,
+struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry,
                              int dir)
 {
        struct list_head *head = mount_hashtable + hash(mnt, dentry);
        struct list_head *tmp = head;
-       struct vfsmount *p, *found = NULL;
+       struct mount *p, *found = NULL;
 
        for (;;) {
                tmp = dir ? tmp->next : tmp->prev;
                p = NULL;
                if (tmp == head)
                        break;
-               p = list_entry(tmp, struct vfsmount, mnt_hash);
-               if (p->mnt_parent == mnt && p->mnt_mountpoint == dentry) {
+               p = list_entry(tmp, struct mount, mnt_hash);
+               if (&p->mnt_parent->mnt == mnt && p->mnt_mountpoint == dentry) {
                        found = p;
                        break;
                }
@@ -455,16 +520,21 @@ struct vfsmount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry,
  */
 struct vfsmount *lookup_mnt(struct path *path)
 {
-       struct vfsmount *child_mnt;
+       struct mount *child_mnt;
 
        br_read_lock(vfsmount_lock);
-       if ((child_mnt = __lookup_mnt(path->mnt, path->dentry, 1)))
-               mntget(child_mnt);
-       br_read_unlock(vfsmount_lock);
-       return child_mnt;
+       child_mnt = __lookup_mnt(path->mnt, path->dentry, 1);
+       if (child_mnt) {
+               mnt_add_count(child_mnt, 1);
+               br_read_unlock(vfsmount_lock);
+               return &child_mnt->mnt;
+       } else {
+               br_read_unlock(vfsmount_lock);
+               return NULL;
+       }
 }
 
-static inline int check_mnt(struct vfsmount *mnt)
+static inline int check_mnt(struct mount *mnt)
 {
        return mnt->mnt_ns == current->nsproxy->mnt_ns;
 }
@@ -492,65 +562,107 @@ static void __touch_mnt_namespace(struct mnt_namespace *ns)
 }
 
 /*
+ * Clear dentry's mounted state if it has no remaining mounts.
+ * vfsmount_lock must be held for write.
+ */
+static void dentry_reset_mounted(struct dentry *dentry)
+{
+       unsigned u;
+
+       for (u = 0; u < HASH_SIZE; u++) {
+               struct mount *p;
+
+               list_for_each_entry(p, &mount_hashtable[u], mnt_hash) {
+                       if (p->mnt_mountpoint == dentry)
+                               return;
+               }
+       }
+       spin_lock(&dentry->d_lock);
+       dentry->d_flags &= ~DCACHE_MOUNTED;
+       spin_unlock(&dentry->d_lock);
+}
+
+/*
  * vfsmount lock must be held for write
  */
-static void detach_mnt(struct vfsmount *mnt, struct path *old_path)
+static void detach_mnt(struct mount *mnt, struct path *old_path)
 {
        old_path->dentry = mnt->mnt_mountpoint;
-       old_path->mnt = mnt->mnt_parent;
+       old_path->mnt = &mnt->mnt_parent->mnt;
        mnt->mnt_parent = mnt;
-       mnt->mnt_mountpoint = mnt->mnt_root;
+       mnt->mnt_mountpoint = mnt->mnt.mnt_root;
        list_del_init(&mnt->mnt_child);
        list_del_init(&mnt->mnt_hash);
-       old_path->dentry->d_mounted--;
+       dentry_reset_mounted(old_path->dentry);
 }
 
 /*
  * vfsmount lock must be held for write
  */
-void mnt_set_mountpoint(struct vfsmount *mnt, struct dentry *dentry,
-                       struct vfsmount *child_mnt)
+void mnt_set_mountpoint(struct mount *mnt, struct dentry *dentry,
+                       struct mount *child_mnt)
 {
-       child_mnt->mnt_parent = mntget(mnt);
+       mnt_add_count(mnt, 1);  /* essentially, that's mntget */
        child_mnt->mnt_mountpoint = dget(dentry);
-       dentry->d_mounted++;
+       child_mnt->mnt_parent = mnt;
+       spin_lock(&dentry->d_lock);
+       dentry->d_flags |= DCACHE_MOUNTED;
+       spin_unlock(&dentry->d_lock);
 }
 
 /*
  * vfsmount lock must be held for write
  */
-static void attach_mnt(struct vfsmount *mnt, struct path *path)
+static void attach_mnt(struct mount *mnt, struct path *path)
 {
-       mnt_set_mountpoint(path->mnt, path->dentry, mnt);
+       mnt_set_mountpoint(real_mount(path->mnt), path->dentry, mnt);
        list_add_tail(&mnt->mnt_hash, mount_hashtable +
                        hash(path->mnt, path->dentry));
-       list_add_tail(&mnt->mnt_child, &path->mnt->mnt_mounts);
+       list_add_tail(&mnt->mnt_child, &real_mount(path->mnt)->mnt_mounts);
+}
+
+static inline void __mnt_make_longterm(struct mount *mnt)
+{
+#ifdef CONFIG_SMP
+       atomic_inc(&mnt->mnt_longterm);
+#endif
+}
+
+/* needs vfsmount lock for write */
+static inline void __mnt_make_shortterm(struct mount *mnt)
+{
+#ifdef CONFIG_SMP
+       atomic_dec(&mnt->mnt_longterm);
+#endif
 }
 
 /*
  * vfsmount lock must be held for write
  */
-static void commit_tree(struct vfsmount *mnt)
+static void commit_tree(struct mount *mnt)
 {
-       struct vfsmount *parent = mnt->mnt_parent;
-       struct vfsmount *m;
+       struct mount *parent = mnt->mnt_parent;
+       struct mount *m;
        LIST_HEAD(head);
        struct mnt_namespace *n = parent->mnt_ns;
 
        BUG_ON(parent == mnt);
 
        list_add_tail(&head, &mnt->mnt_list);
-       list_for_each_entry(m, &head, mnt_list)
+       list_for_each_entry(m, &head, mnt_list) {
                m->mnt_ns = n;
+               __mnt_make_longterm(m);
+       }
+
        list_splice(&head, n->list.prev);
 
        list_add_tail(&mnt->mnt_hash, mount_hashtable +
-                               hash(parent, mnt->mnt_mountpoint));
+                               hash(&parent->mnt, mnt->mnt_mountpoint));
        list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
        touch_mnt_namespace(n);
 }
 
-static struct vfsmount *next_mnt(struct vfsmount *p, struct vfsmount *root)
+static struct mount *next_mnt(struct mount *p, struct mount *root)
 {
        struct list_head *next = p->mnt_mounts.next;
        if (next == &p->mnt_mounts) {
@@ -563,24 +675,57 @@ static struct vfsmount *next_mnt(struct vfsmount *p, struct vfsmount *root)
                        p = p->mnt_parent;
                }
        }
-       return list_entry(next, struct vfsmount, mnt_child);
+       return list_entry(next, struct mount, mnt_child);
 }
 
-static struct vfsmount *skip_mnt_tree(struct vfsmount *p)
+static struct mount *skip_mnt_tree(struct mount *p)
 {
        struct list_head *prev = p->mnt_mounts.prev;
        while (prev != &p->mnt_mounts) {
-               p = list_entry(prev, struct vfsmount, mnt_child);
+               p = list_entry(prev, struct mount, mnt_child);
                prev = p->mnt_mounts.prev;
        }
        return p;
 }
 
-static struct vfsmount *clone_mnt(struct vfsmount *old, struct dentry *root,
+struct vfsmount *
+vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void *data)
+{
+       struct mount *mnt;
+       struct dentry *root;
+
+       if (!type)
+               return ERR_PTR(-ENODEV);
+
+       mnt = alloc_vfsmnt(name);
+       if (!mnt)
+               return ERR_PTR(-ENOMEM);
+
+       if (flags & MS_KERNMOUNT)
+               mnt->mnt.mnt_flags = MNT_INTERNAL;
+
+       root = mount_fs(type, flags, name, data);
+       if (IS_ERR(root)) {
+               free_vfsmnt(mnt);
+               return ERR_CAST(root);
+       }
+
+       mnt->mnt.mnt_root = root;
+       mnt->mnt.mnt_sb = root->d_sb;
+       mnt->mnt_mountpoint = mnt->mnt.mnt_root;
+       mnt->mnt_parent = mnt;
+       br_write_lock(vfsmount_lock);
+       list_add_tail(&mnt->mnt_instance, &root->d_sb->s_mounts);
+       br_write_unlock(vfsmount_lock);
+       return &mnt->mnt;
+}
+EXPORT_SYMBOL_GPL(vfs_kern_mount);
+
+static struct mount *clone_mnt(struct mount *old, struct dentry *root,
                                        int flag)
 {
-       struct super_block *sb = old->mnt_sb;
-       struct vfsmount *mnt = alloc_vfsmnt(old->mnt_devname);
+       struct super_block *sb = old->mnt.mnt_sb;
+       struct mount *mnt = alloc_vfsmnt(old->mnt_devname);
 
        if (mnt) {
                if (flag & (CL_SLAVE | CL_PRIVATE))
@@ -594,12 +739,15 @@ static struct vfsmount *clone_mnt(struct vfsmount *old, struct dentry *root,
                                goto out_free;
                }
 
-               mnt->mnt_flags = old->mnt_flags & ~MNT_WRITE_HOLD;
+               mnt->mnt.mnt_flags = old->mnt.mnt_flags & ~MNT_WRITE_HOLD;
                atomic_inc(&sb->s_active);
-               mnt->mnt_sb = sb;
-               mnt->mnt_root = dget(root);
-               mnt->mnt_mountpoint = mnt->mnt_root;
+               mnt->mnt.mnt_sb = sb;
+               mnt->mnt.mnt_root = dget(root);
+               mnt->mnt_mountpoint = mnt->mnt.mnt_root;
                mnt->mnt_parent = mnt;
+               br_write_lock(vfsmount_lock);
+               list_add_tail(&mnt->mnt_instance, &sb->s_mounts);
+               br_write_unlock(vfsmount_lock);
 
                if (flag & CL_SLAVE) {
                        list_add(&mnt->mnt_slave, &old->mnt_slave_list);
@@ -629,9 +777,11 @@ static struct vfsmount *clone_mnt(struct vfsmount *old, struct dentry *root,
        return NULL;
 }
 
-static inline void __mntput(struct vfsmount *mnt)
+static inline void mntfree(struct mount *mnt)
 {
-       struct super_block *sb = mnt->mnt_sb;
+       struct vfsmount *m = &mnt->mnt;
+       struct super_block *sb = m->mnt_sb;
+
        /*
         * This probably indicates that somebody messed
         * up a mnt_want/drop_write() pair.  If this
@@ -639,58 +789,90 @@ static inline void __mntput(struct vfsmount *mnt)
         * to make r/w->r/o transitions.
         */
        /*
-        * atomic_dec_and_lock() used to deal with ->mnt_count decrements
-        * provides barriers, so count_mnt_writers() below is safe.  AV
+        * The locking used to deal with mnt_count decrement provides barriers,
+        * so mnt_get_writers() below is safe.
         */
-       WARN_ON(count_mnt_writers(mnt));
-       fsnotify_vfsmount_delete(mnt);
-       dput(mnt->mnt_root);
+       WARN_ON(mnt_get_writers(mnt));
+       fsnotify_vfsmount_delete(m);
+       dput(m->mnt_root);
        free_vfsmnt(mnt);
        deactivate_super(sb);
 }
 
-void mntput_no_expire(struct vfsmount *mnt)
+static void mntput_no_expire(struct mount *mnt)
 {
-repeat:
-       if (atomic_add_unless(&mnt->mnt_count, -1, 1))
+put_again:
+#ifdef CONFIG_SMP
+       br_read_lock(vfsmount_lock);
+       if (likely(atomic_read(&mnt->mnt_longterm))) {
+               mnt_add_count(mnt, -1);
+               br_read_unlock(vfsmount_lock);
                return;
+       }
+       br_read_unlock(vfsmount_lock);
+
        br_write_lock(vfsmount_lock);
-       if (!atomic_dec_and_test(&mnt->mnt_count)) {
+       mnt_add_count(mnt, -1);
+       if (mnt_get_count(mnt)) {
                br_write_unlock(vfsmount_lock);
                return;
        }
-       if (likely(!mnt->mnt_pinned)) {
-               br_write_unlock(vfsmount_lock);
-               __mntput(mnt);
+#else
+       mnt_add_count(mnt, -1);
+       if (likely(mnt_get_count(mnt)))
                return;
+       br_write_lock(vfsmount_lock);
+#endif
+       if (unlikely(mnt->mnt_pinned)) {
+               mnt_add_count(mnt, mnt->mnt_pinned + 1);
+               mnt->mnt_pinned = 0;
+               br_write_unlock(vfsmount_lock);
+               acct_auto_close_mnt(&mnt->mnt);
+               goto put_again;
        }
-       atomic_add(mnt->mnt_pinned + 1, &mnt->mnt_count);
-       mnt->mnt_pinned = 0;
+       list_del(&mnt->mnt_instance);
        br_write_unlock(vfsmount_lock);
-       acct_auto_close_mnt(mnt);
-       goto repeat;
+       mntfree(mnt);
 }
-EXPORT_SYMBOL(mntput_no_expire);
+
+void mntput(struct vfsmount *mnt)
+{
+       if (mnt) {
+               struct mount *m = real_mount(mnt);
+               /* avoid cacheline pingpong, hope gcc doesn't get "smart" */
+               if (unlikely(m->mnt_expiry_mark))
+                       m->mnt_expiry_mark = 0;
+               mntput_no_expire(m);
+       }
+}
+EXPORT_SYMBOL(mntput);
+
+struct vfsmount *mntget(struct vfsmount *mnt)
+{
+       if (mnt)
+               mnt_add_count(real_mount(mnt), 1);
+       return mnt;
+}
+EXPORT_SYMBOL(mntget);
 
 void mnt_pin(struct vfsmount *mnt)
 {
        br_write_lock(vfsmount_lock);
-       mnt->mnt_pinned++;
+       real_mount(mnt)->mnt_pinned++;
        br_write_unlock(vfsmount_lock);
 }
-
 EXPORT_SYMBOL(mnt_pin);
 
-void mnt_unpin(struct vfsmount *mnt)
+void mnt_unpin(struct vfsmount *m)
 {
+       struct mount *mnt = real_mount(m);
        br_write_lock(vfsmount_lock);
        if (mnt->mnt_pinned) {
-               atomic_inc(&mnt->mnt_count);
+               mnt_add_count(mnt, 1);
                mnt->mnt_pinned--;
        }
        br_write_unlock(vfsmount_lock);
 }
-
 EXPORT_SYMBOL(mnt_unpin);
 
 static inline void mangle(struct seq_file *m, const char *s)
@@ -704,12 +886,12 @@ static inline void mangle(struct seq_file *m, const char *s)
  *
  * See also save_mount_options().
  */
-int generic_show_options(struct seq_file *m, struct vfsmount *mnt)
+int generic_show_options(struct seq_file *m, struct dentry *root)
 {
        const char *options;
 
        rcu_read_lock();
-       options = rcu_dereference(mnt->mnt_sb->s_options);
+       options = rcu_dereference(root->d_sb->s_options);
 
        if (options != NULL && options[0]) {
                seq_putc(m, ',');
@@ -753,10 +935,10 @@ void replace_mount_options(struct super_block *sb, char *options)
 EXPORT_SYMBOL(replace_mount_options);
 
 #ifdef CONFIG_PROC_FS
-/* iterator */
+/* iterator; we want it to have access to namespace_sem, thus here... */
 static void *m_start(struct seq_file *m, loff_t *pos)
 {
-       struct proc_mounts *p = m->private;
+       struct proc_mounts *p = container_of(m, struct proc_mounts, m);
 
        down_read(&namespace_sem);
        return seq_list_start(&p->ns->list, *pos);
@@ -764,7 +946,7 @@ static void *m_start(struct seq_file *m, loff_t *pos)
 
 static void *m_next(struct seq_file *m, void *v, loff_t *pos)
 {
-       struct proc_mounts *p = m->private;
+       struct proc_mounts *p = container_of(m, struct proc_mounts, m);
 
        return seq_list_next(v, &p->ns->list, pos);
 }
@@ -774,200 +956,18 @@ static void m_stop(struct seq_file *m, void *v)
        up_read(&namespace_sem);
 }
 
-int mnt_had_events(struct proc_mounts *p)
-{
-       struct mnt_namespace *ns = p->ns;
-       int res = 0;
-
-       br_read_lock(vfsmount_lock);
-       if (p->event != ns->event) {
-               p->event = ns->event;
-               res = 1;
-       }
-       br_read_unlock(vfsmount_lock);
-
-       return res;
-}
-
-struct proc_fs_info {
-       int flag;
-       const char *str;
-};
-
-static int show_sb_opts(struct seq_file *m, struct super_block *sb)
-{
-       static const struct proc_fs_info fs_info[] = {
-               { MS_SYNCHRONOUS, ",sync" },
-               { MS_DIRSYNC, ",dirsync" },
-               { MS_MANDLOCK, ",mand" },
-               { 0, NULL }
-       };
-       const struct proc_fs_info *fs_infop;
-
-       for (fs_infop = fs_info; fs_infop->flag; fs_infop++) {
-               if (sb->s_flags & fs_infop->flag)
-                       seq_puts(m, fs_infop->str);
-       }
-
-       return security_sb_show_options(m, sb);
-}
-
-static void show_mnt_opts(struct seq_file *m, struct vfsmount *mnt)
+static int m_show(struct seq_file *m, void *v)
 {
-       static const struct proc_fs_info mnt_info[] = {
-               { MNT_NOSUID, ",nosuid" },
-               { MNT_NODEV, ",nodev" },
-               { MNT_NOEXEC, ",noexec" },
-               { MNT_NOATIME, ",noatime" },
-               { MNT_NODIRATIME, ",nodiratime" },
-               { MNT_RELATIME, ",relatime" },
-               { 0, NULL }
-       };
-       const struct proc_fs_info *fs_infop;
-
-       for (fs_infop = mnt_info; fs_infop->flag; fs_infop++) {
-               if (mnt->mnt_flags & fs_infop->flag)
-                       seq_puts(m, fs_infop->str);
-       }
-}
-
-static void show_type(struct seq_file *m, struct super_block *sb)
-{
-       mangle(m, sb->s_type->name);
-       if (sb->s_subtype && sb->s_subtype[0]) {
-               seq_putc(m, '.');
-               mangle(m, sb->s_subtype);
-       }
-}
-
-static int show_vfsmnt(struct seq_file *m, void *v)
-{
-       struct vfsmount *mnt = list_entry(v, struct vfsmount, mnt_list);
-       int err = 0;
-       struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
-
-       mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
-       seq_putc(m, ' ');
-       seq_path(m, &mnt_path, " \t\n\\");
-       seq_putc(m, ' ');
-       show_type(m, mnt->mnt_sb);
-       seq_puts(m, __mnt_is_readonly(mnt) ? " ro" : " rw");
-       err = show_sb_opts(m, mnt->mnt_sb);
-       if (err)
-               goto out;
-       show_mnt_opts(m, mnt);
-       if (mnt->mnt_sb->s_op->show_options)
-               err = mnt->mnt_sb->s_op->show_options(m, mnt);
-       seq_puts(m, " 0 0\n");
-out:
-       return err;
+       struct proc_mounts *p = container_of(m, struct proc_mounts, m);
+       struct mount *r = list_entry(v, struct mount, mnt_list);
+       return p->show(m, &r->mnt);
 }
 
 const struct seq_operations mounts_op = {
        .start  = m_start,
        .next   = m_next,
        .stop   = m_stop,
-       .show   = show_vfsmnt
-};
-
-static int show_mountinfo(struct seq_file *m, void *v)
-{
-       struct proc_mounts *p = m->private;
-       struct vfsmount *mnt = list_entry(v, struct vfsmount, mnt_list);
-       struct super_block *sb = mnt->mnt_sb;
-       struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
-       struct path root = p->root;
-       int err = 0;
-
-       seq_printf(m, "%i %i %u:%u ", mnt->mnt_id, mnt->mnt_parent->mnt_id,
-                  MAJOR(sb->s_dev), MINOR(sb->s_dev));
-       seq_dentry(m, mnt->mnt_root, " \t\n\\");
-       seq_putc(m, ' ');
-       seq_path_root(m, &mnt_path, &root, " \t\n\\");
-       if (root.mnt != p->root.mnt || root.dentry != p->root.dentry) {
-               /*
-                * Mountpoint is outside root, discard that one.  Ugly,
-                * but less so than trying to do that in iterator in a
-                * race-free way (due to renames).
-                */
-               return SEQ_SKIP;
-       }
-       seq_puts(m, mnt->mnt_flags & MNT_READONLY ? " ro" : " rw");
-       show_mnt_opts(m, mnt);
-
-       /* Tagged fields ("foo:X" or "bar") */
-       if (IS_MNT_SHARED(mnt))
-               seq_printf(m, " shared:%i", mnt->mnt_group_id);
-       if (IS_MNT_SLAVE(mnt)) {
-               int master = mnt->mnt_master->mnt_group_id;
-               int dom = get_dominating_id(mnt, &p->root);
-               seq_printf(m, " master:%i", master);
-               if (dom && dom != master)
-                       seq_printf(m, " propagate_from:%i", dom);
-       }
-       if (IS_MNT_UNBINDABLE(mnt))
-               seq_puts(m, " unbindable");
-
-       /* Filesystem specific data */
-       seq_puts(m, " - ");
-       show_type(m, sb);
-       seq_putc(m, ' ');
-       mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
-       seq_puts(m, sb->s_flags & MS_RDONLY ? " ro" : " rw");
-       err = show_sb_opts(m, sb);
-       if (err)
-               goto out;
-       if (sb->s_op->show_options)
-               err = sb->s_op->show_options(m, mnt);
-       seq_putc(m, '\n');
-out:
-       return err;
-}
-
-const struct seq_operations mountinfo_op = {
-       .start  = m_start,
-       .next   = m_next,
-       .stop   = m_stop,
-       .show   = show_mountinfo,
-};
-
-static int show_vfsstat(struct seq_file *m, void *v)
-{
-       struct vfsmount *mnt = list_entry(v, struct vfsmount, mnt_list);
-       struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
-       int err = 0;
-
-       /* device */
-       if (mnt->mnt_devname) {
-               seq_puts(m, "device ");
-               mangle(m, mnt->mnt_devname);
-       } else
-               seq_puts(m, "no device");
-
-       /* mount point */
-       seq_puts(m, " mounted on ");
-       seq_path(m, &mnt_path, " \t\n\\");
-       seq_putc(m, ' ');
-
-       /* file system type */
-       seq_puts(m, "with fstype ");
-       show_type(m, mnt->mnt_sb);
-
-       /* optional statistics */
-       if (mnt->mnt_sb->s_op->show_stats) {
-               seq_putc(m, ' ');
-               err = mnt->mnt_sb->s_op->show_stats(m, mnt);
-       }
-
-       seq_putc(m, '\n');
-       return err;
-}
-
-const struct seq_operations mountstats_op = {
-       .start  = m_start,
-       .next   = m_next,
-       .stop   = m_stop,
-       .show   = show_vfsstat,
+       .show   = m_show,
 };
 #endif  /* CONFIG_PROC_FS */
 
@@ -979,18 +979,21 @@ const struct seq_operations mountstats_op = {
  * open files, pwds, chroots or sub mounts that are
  * busy.
  */
-int may_umount_tree(struct vfsmount *mnt)
+int may_umount_tree(struct vfsmount *m)
 {
+       struct mount *mnt = real_mount(m);
        int actual_refs = 0;
        int minimum_refs = 0;
-       struct vfsmount *p;
+       struct mount *p;
+       BUG_ON(!m);
 
-       br_read_lock(vfsmount_lock);
+       /* write lock needed for mnt_get_count */
+       br_write_lock(vfsmount_lock);
        for (p = mnt; p; p = next_mnt(p, mnt)) {
-               actual_refs += atomic_read(&p->mnt_count);
+               actual_refs += mnt_get_count(p);
                minimum_refs += 2;
        }
-       br_read_unlock(vfsmount_lock);
+       br_write_unlock(vfsmount_lock);
 
        if (actual_refs > minimum_refs)
                return 0;
@@ -1017,10 +1020,10 @@ int may_umount(struct vfsmount *mnt)
 {
        int ret = 1;
        down_read(&namespace_sem);
-       br_read_lock(vfsmount_lock);
-       if (propagate_mount_busy(mnt, 2))
+       br_write_lock(vfsmount_lock);
+       if (propagate_mount_busy(real_mount(mnt), 2))
                ret = 0;
-       br_read_unlock(vfsmount_lock);
+       br_write_unlock(vfsmount_lock);
        up_read(&namespace_sem);
        return ret;
 }
@@ -1029,25 +1032,25 @@ EXPORT_SYMBOL(may_umount);
 
 void release_mounts(struct list_head *head)
 {
-       struct vfsmount *mnt;
+       struct mount *mnt;
        while (!list_empty(head)) {
-               mnt = list_first_entry(head, struct vfsmount, mnt_hash);
+               mnt = list_first_entry(head, struct mount, mnt_hash);
                list_del_init(&mnt->mnt_hash);
-               if (mnt->mnt_parent != mnt) {
+               if (mnt_has_parent(mnt)) {
                        struct dentry *dentry;
-                       struct vfsmount *m;
+                       struct mount *m;
 
                        br_write_lock(vfsmount_lock);
                        dentry = mnt->mnt_mountpoint;
                        m = mnt->mnt_parent;
-                       mnt->mnt_mountpoint = mnt->mnt_root;
+                       mnt->mnt_mountpoint = mnt->mnt.mnt_root;
                        mnt->mnt_parent = mnt;
                        m->mnt_ghosts--;
                        br_write_unlock(vfsmount_lock);
                        dput(dentry);
-                       mntput(m);
+                       mntput(&m->mnt);
                }
-               mntput(mnt);
+               mntput(&mnt->mnt);
        }
 }
 
@@ -1055,39 +1058,43 @@ void release_mounts(struct list_head *head)
  * vfsmount lock must be held for write
  * namespace_sem must be held for write
  */
-void umount_tree(struct vfsmount *mnt, int propagate, struct list_head *kill)
+void umount_tree(struct mount *mnt, int propagate, struct list_head *kill)
 {
-       struct vfsmount *p;
+       LIST_HEAD(tmp_list);
+       struct mount *p;
 
        for (p = mnt; p; p = next_mnt(p, mnt))
-               list_move(&p->mnt_hash, kill);
+               list_move(&p->mnt_hash, &tmp_list);
 
        if (propagate)
-               propagate_umount(kill);
+               propagate_umount(&tmp_list);
 
-       list_for_each_entry(p, kill, mnt_hash) {
+       list_for_each_entry(p, &tmp_list, mnt_hash) {
                list_del_init(&p->mnt_expire);
                list_del_init(&p->mnt_list);
                __touch_mnt_namespace(p->mnt_ns);
+               if (p->mnt_ns)
+                       __mnt_make_shortterm(p);
                p->mnt_ns = NULL;
                list_del_init(&p->mnt_child);
-               if (p->mnt_parent != p) {
+               if (mnt_has_parent(p)) {
                        p->mnt_parent->mnt_ghosts++;
-                       p->mnt_mountpoint->d_mounted--;
+                       dentry_reset_mounted(p->mnt_mountpoint);
                }
                change_mnt_propagation(p, MS_PRIVATE);
        }
+       list_splice(&tmp_list, kill);
 }
 
-static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts);
+static void shrink_submounts(struct mount *mnt, struct list_head *umounts);
 
-static int do_umount(struct vfsmount *mnt, int flags)
+static int do_umount(struct mount *mnt, int flags)
 {
-       struct super_block *sb = mnt->mnt_sb;
+       struct super_block *sb = mnt->mnt.mnt_sb;
        int retval;
        LIST_HEAD(umount_list);
 
-       retval = security_sb_umount(mnt, flags);
+       retval = security_sb_umount(&mnt->mnt, flags);
        if (retval)
                return retval;
 
@@ -1098,12 +1105,20 @@ static int do_umount(struct vfsmount *mnt, int flags)
         *  (2) the usage count == 1 [parent vfsmount] + 1 [sys_umount]
         */
        if (flags & MNT_EXPIRE) {
-               if (mnt == current->fs->root.mnt ||
+               if (&mnt->mnt == current->fs->root.mnt ||
                    flags & (MNT_FORCE | MNT_DETACH))
                        return -EINVAL;
 
-               if (atomic_read(&mnt->mnt_count) != 2)
+               /*
+                * probably don't strictly need the lock here if we examined
+                * all race cases, but it's a slowpath.
+                */
+               br_write_lock(vfsmount_lock);
+               if (mnt_get_count(mnt) != 2) {
+                       br_write_unlock(vfsmount_lock);
                        return -EBUSY;
+               }
+               br_write_unlock(vfsmount_lock);
 
                if (!xchg(&mnt->mnt_expiry_mark, 1))
                        return -EAGAIN;
@@ -1132,7 +1147,7 @@ static int do_umount(struct vfsmount *mnt, int flags)
         * /reboot - static binary that would close all descriptors and
         * call reboot(9). Then init(8) could umount root and exec /reboot.
         */
-       if (mnt == current->fs->root.mnt && !(flags & MNT_DETACH)) {
+       if (&mnt->mnt == current->fs->root.mnt && !(flags & MNT_DETACH)) {
                /*
                 * Special case for "unmounting" root ...
                 * we just try to remount it readonly.
@@ -1174,6 +1189,7 @@ static int do_umount(struct vfsmount *mnt, int flags)
 SYSCALL_DEFINE2(umount, char __user *, name, int, flags)
 {
        struct path path;
+       struct mount *mnt;
        int retval;
        int lookup_flags = 0;
 
@@ -1186,21 +1202,22 @@ SYSCALL_DEFINE2(umount, char __user *, name, int, flags)
        retval = user_path_at(AT_FDCWD, name, lookup_flags, &path);
        if (retval)
                goto out;
+       mnt = real_mount(path.mnt);
        retval = -EINVAL;
        if (path.dentry != path.mnt->mnt_root)
                goto dput_and_out;
-       if (!check_mnt(path.mnt))
+       if (!check_mnt(mnt))
                goto dput_and_out;
 
        retval = -EPERM;
        if (!capable(CAP_SYS_ADMIN))
                goto dput_and_out;
 
-       retval = do_umount(path.mnt, flags);
+       retval = do_umount(mnt, flags);
 dput_and_out:
        /* we mustn't call path_put() as that would clear mnt_expiry_mark */
        dput(path.dentry);
-       mntput_no_expire(path.mnt);
+       mntput_no_expire(mnt);
 out:
        return retval;
 }
@@ -1235,10 +1252,10 @@ static int mount_is_safe(struct path *path)
 #endif
 }
 
-struct vfsmount *copy_tree(struct vfsmount *mnt, struct dentry *dentry,
+struct mount *copy_tree(struct mount *mnt, struct dentry *dentry,
                                        int flag)
 {
-       struct vfsmount *res, *p, *q, *r, *s;
+       struct mount *res, *p, *q, *r;
        struct path path;
 
        if (!(flag & CL_COPY_ALL) && IS_MNT_UNBINDABLE(mnt))
@@ -1251,6 +1268,7 @@ struct vfsmount *copy_tree(struct vfsmount *mnt, struct dentry *dentry,
 
        p = mnt;
        list_for_each_entry(r, &mnt->mnt_mounts, mnt_child) {
+               struct mount *s;
                if (!is_subdir(r->mnt_mountpoint, dentry))
                        continue;
 
@@ -1264,9 +1282,9 @@ struct vfsmount *copy_tree(struct vfsmount *mnt, struct dentry *dentry,
                                q = q->mnt_parent;
                        }
                        p = s;
-                       path.mnt = q;
+                       path.mnt = &q->mnt;
                        path.dentry = p->mnt_mountpoint;
-                       q = clone_mnt(p, p->mnt_root, flag);
+                       q = clone_mnt(p, p->mnt.mnt_root, flag);
                        if (!q)
                                goto Enomem;
                        br_write_lock(vfsmount_lock);
@@ -1289,11 +1307,12 @@ Enomem:
 
 struct vfsmount *collect_mounts(struct path *path)
 {
-       struct vfsmount *tree;
+       struct mount *tree;
        down_write(&namespace_sem);
-       tree = copy_tree(path->mnt, path->dentry, CL_COPY_ALL | CL_PRIVATE);
+       tree = copy_tree(real_mount(path->mnt), path->dentry,
+                        CL_COPY_ALL | CL_PRIVATE);
        up_write(&namespace_sem);
-       return tree;
+       return tree ? &tree->mnt : NULL;
 }
 
 void drop_collected_mounts(struct vfsmount *mnt)
@@ -1301,7 +1320,7 @@ void drop_collected_mounts(struct vfsmount *mnt)
        LIST_HEAD(umount_list);
        down_write(&namespace_sem);
        br_write_lock(vfsmount_lock);
-       umount_tree(mnt, 0, &umount_list);
+       umount_tree(real_mount(mnt), 0, &umount_list);
        br_write_unlock(vfsmount_lock);
        up_write(&namespace_sem);
        release_mounts(&umount_list);
@@ -1310,21 +1329,21 @@ void drop_collected_mounts(struct vfsmount *mnt)
 int iterate_mounts(int (*f)(struct vfsmount *, void *), void *arg,
                   struct vfsmount *root)
 {
-       struct vfsmount *mnt;
+       struct mount *mnt;
        int res = f(root, arg);
        if (res)
                return res;
-       list_for_each_entry(mnt, &root->mnt_list, mnt_list) {
-               res = f(mnt, arg);
+       list_for_each_entry(mnt, &real_mount(root)->mnt_list, mnt_list) {
+               res = f(&mnt->mnt, arg);
                if (res)
                        return res;
        }
        return 0;
 }
 
-static void cleanup_group_ids(struct vfsmount *mnt, struct vfsmount *end)
+static void cleanup_group_ids(struct mount *mnt, struct mount *end)
 {
-       struct vfsmount *p;
+       struct mount *p;
 
        for (p = mnt; p != end; p = next_mnt(p, mnt)) {
                if (p->mnt_group_id && !IS_MNT_SHARED(p))
@@ -1332,9 +1351,9 @@ static void cleanup_group_ids(struct vfsmount *mnt, struct vfsmount *end)
        }
 }
 
-static int invent_group_ids(struct vfsmount *mnt, bool recurse)
+static int invent_group_ids(struct mount *mnt, bool recurse)
 {
-       struct vfsmount *p;
+       struct mount *p;
 
        for (p = mnt; p; p = recurse ? next_mnt(p, mnt) : NULL) {
                if (!p->mnt_group_id && !IS_MNT_SHARED(p)) {
@@ -1412,13 +1431,13 @@ static int invent_group_ids(struct vfsmount *mnt, bool recurse)
  * Must be called without spinlocks held, since this function can sleep
  * in allocations.
  */
-static int attach_recursive_mnt(struct vfsmount *source_mnt,
+static int attach_recursive_mnt(struct mount *source_mnt,
                        struct path *path, struct path *parent_path)
 {
        LIST_HEAD(tree_list);
-       struct vfsmount *dest_mnt = path->mnt;
+       struct mount *dest_mnt = real_mount(path->mnt);
        struct dentry *dest_dentry = path->dentry;
-       struct vfsmount *child, *p;
+       struct mount *child, *p;
        int err;
 
        if (IS_MNT_SHARED(dest_mnt)) {
@@ -1439,7 +1458,7 @@ static int attach_recursive_mnt(struct vfsmount *source_mnt,
        if (parent_path) {
                detach_mnt(source_mnt, parent_path);
                attach_mnt(source_mnt, path);
-               touch_mnt_namespace(parent_path->mnt->mnt_ns);
+               touch_mnt_namespace(source_mnt->mnt_ns);
        } else {
                mnt_set_mountpoint(dest_mnt, dest_dentry, source_mnt);
                commit_tree(source_mnt);
@@ -1460,26 +1479,46 @@ static int attach_recursive_mnt(struct vfsmount *source_mnt,
        return err;
 }
 
-static int graft_tree(struct vfsmount *mnt, struct path *path)
+static int lock_mount(struct path *path)
 {
-       int err;
-       if (mnt->mnt_sb->s_flags & MS_NOUSER)
+       struct vfsmount *mnt;
+retry:
+       mutex_lock(&path->dentry->d_inode->i_mutex);
+       if (unlikely(cant_mount(path->dentry))) {
+               mutex_unlock(&path->dentry->d_inode->i_mutex);
+               return -ENOENT;
+       }
+       down_write(&namespace_sem);
+       mnt = lookup_mnt(path);
+       if (likely(!mnt))
+               return 0;
+       up_write(&namespace_sem);
+       mutex_unlock(&path->dentry->d_inode->i_mutex);
+       path_put(path);
+       path->mnt = mnt;
+       path->dentry = dget(mnt->mnt_root);
+       goto retry;
+}
+
+static void unlock_mount(struct path *path)
+{
+       up_write(&namespace_sem);
+       mutex_unlock(&path->dentry->d_inode->i_mutex);
+}
+
+static int graft_tree(struct mount *mnt, struct path *path)
+{
+       if (mnt->mnt.mnt_sb->s_flags & MS_NOUSER)
                return -EINVAL;
 
        if (S_ISDIR(path->dentry->d_inode->i_mode) !=
-             S_ISDIR(mnt->mnt_root->d_inode->i_mode))
+             S_ISDIR(mnt->mnt.mnt_root->d_inode->i_mode))
                return -ENOTDIR;
 
-       err = -ENOENT;
-       mutex_lock(&path->dentry->d_inode->i_mutex);
-       if (cant_mount(path->dentry))
-               goto out_unlock;
+       if (d_unlinked(path->dentry))
+               return -ENOENT;
 
-       if (!d_unlinked(path->dentry))
-               err = attach_recursive_mnt(mnt, path, NULL);
-out_unlock:
-       mutex_unlock(&path->dentry->d_inode->i_mutex);
-       return err;
+       return attach_recursive_mnt(mnt, path, NULL);
 }
 
 /*
@@ -1488,7 +1527,7 @@ out_unlock:
 
 static int flags_to_propagation_type(int flags)
 {
-       int type = flags & ~MS_REC;
+       int type = flags & ~(MS_REC | MS_SILENT);
 
        /* Fail if any non-propagation flags are set */
        if (type & ~(MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
@@ -1504,7 +1543,8 @@ static int flags_to_propagation_type(int flags)
  */
 static int do_change_type(struct path *path, int flag)
 {
-       struct vfsmount *m, *mnt = path->mnt;
+       struct mount *m;
+       struct mount *mnt = real_mount(path->mnt);
        int recurse = flag & MS_REC;
        int type;
        int err = 0;
@@ -1542,46 +1582,50 @@ static int do_change_type(struct path *path, int flag)
 static int do_loopback(struct path *path, char *old_name,
                                int recurse)
 {
+       LIST_HEAD(umount_list);
        struct path old_path;
-       struct vfsmount *mnt = NULL;
+       struct mount *mnt = NULL, *old;
        int err = mount_is_safe(path);
        if (err)
                return err;
        if (!old_name || !*old_name)
                return -EINVAL;
-       err = kern_path(old_name, LOOKUP_FOLLOW, &old_path);
+       err = kern_path(old_name, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &old_path);
        if (err)
                return err;
 
-       down_write(&namespace_sem);
-       err = -EINVAL;
-       if (IS_MNT_UNBINDABLE(old_path.mnt))
+       err = lock_mount(path);
+       if (err)
                goto out;
 
-       if (!check_mnt(path->mnt) || !check_mnt(old_path.mnt))
-               goto out;
+       old = real_mount(old_path.mnt);
+
+       err = -EINVAL;
+       if (IS_MNT_UNBINDABLE(old))
+               goto out2;
+
+       if (!check_mnt(real_mount(path->mnt)) || !check_mnt(old))
+               goto out2;
 
        err = -ENOMEM;
        if (recurse)
-               mnt = copy_tree(old_path.mnt, old_path.dentry, 0);
+               mnt = copy_tree(old, old_path.dentry, 0);
        else
-               mnt = clone_mnt(old_path.mnt, old_path.dentry, 0);
+               mnt = clone_mnt(old, old_path.dentry, 0);
 
        if (!mnt)
-               goto out;
+               goto out2;
 
        err = graft_tree(mnt, path);
        if (err) {
-               LIST_HEAD(umount_list);
-
                br_write_lock(vfsmount_lock);
                umount_tree(mnt, 0, &umount_list);
                br_write_unlock(vfsmount_lock);
-               release_mounts(&umount_list);
        }
-
+out2:
+       unlock_mount(path);
+       release_mounts(&umount_list);
 out:
-       up_write(&namespace_sem);
        path_put(&old_path);
        return err;
 }
@@ -1597,9 +1641,9 @@ static int change_mount_flags(struct vfsmount *mnt, int ms_flags)
                return 0;
 
        if (readonly_request)
-               error = mnt_make_readonly(mnt);
+               error = mnt_make_readonly(real_mount(mnt));
        else
-               __mnt_unmake_readonly(mnt);
+               __mnt_unmake_readonly(real_mount(mnt));
        return error;
 }
 
@@ -1613,16 +1657,21 @@ static int do_remount(struct path *path, int flags, int mnt_flags,
 {
        int err;
        struct super_block *sb = path->mnt->mnt_sb;
+       struct mount *mnt = real_mount(path->mnt);
 
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
 
-       if (!check_mnt(path->mnt))
+       if (!check_mnt(mnt))
                return -EINVAL;
 
        if (path->dentry != path->mnt->mnt_root)
                return -EINVAL;
 
+       err = security_sb_remount(sb, data);
+       if (err)
+               return err;
+
        down_write(&sb->s_umount);
        if (flags & MS_BIND)
                err = change_mount_flags(path->mnt, flags);
@@ -1630,22 +1679,22 @@ static int do_remount(struct path *path, int flags, int mnt_flags,
                err = do_remount_sb(sb, flags, data, 0);
        if (!err) {
                br_write_lock(vfsmount_lock);
-               mnt_flags |= path->mnt->mnt_flags & MNT_PROPAGATION_MASK;
-               path->mnt->mnt_flags = mnt_flags;
+               mnt_flags |= mnt->mnt.mnt_flags & MNT_PROPAGATION_MASK;
+               mnt->mnt.mnt_flags = mnt_flags;
                br_write_unlock(vfsmount_lock);
        }
        up_write(&sb->s_umount);
        if (!err) {
                br_write_lock(vfsmount_lock);
-               touch_mnt_namespace(path->mnt->mnt_ns);
+               touch_mnt_namespace(mnt->mnt_ns);
                br_write_unlock(vfsmount_lock);
        }
        return err;
 }
 
-static inline int tree_contains_unbindable(struct vfsmount *mnt)
+static inline int tree_contains_unbindable(struct mount *mnt)
 {
-       struct vfsmount *p;
+       struct mount *p;
        for (p = mnt; p; p = next_mnt(p, mnt)) {
                if (IS_MNT_UNBINDABLE(p))
                        return 1;
@@ -1656,7 +1705,8 @@ static inline int tree_contains_unbindable(struct vfsmount *mnt)
 static int do_move_mount(struct path *path, char *old_name)
 {
        struct path old_path, parent_path;
-       struct vfsmount *p;
+       struct mount *p;
+       struct mount *old;
        int err = 0;
        if (!capable(CAP_SYS_ADMIN))
                return -EPERM;
@@ -1666,17 +1716,15 @@ static int do_move_mount(struct path *path, char *old_name)
        if (err)
                return err;
 
-       down_write(&namespace_sem);
-       while (d_mountpoint(path->dentry) &&
-              follow_down(path))
-               ;
-       err = -EINVAL;
-       if (!check_mnt(path->mnt) || !check_mnt(old_path.mnt))
+       err = lock_mount(path);
+       if (err < 0)
                goto out;
 
-       err = -ENOENT;
-       mutex_lock(&path->dentry->d_inode->i_mutex);
-       if (cant_mount(path->dentry))
+       old = real_mount(old_path.mnt);
+       p = real_mount(path->mnt);
+
+       err = -EINVAL;
+       if (!check_mnt(p) || !check_mnt(old))
                goto out1;
 
        if (d_unlinked(path->dentry))
@@ -1686,7 +1734,7 @@ static int do_move_mount(struct path *path, char *old_name)
        if (old_path.dentry != old_path.mnt->mnt_root)
                goto out1;
 
-       if (old_path.mnt == old_path.mnt->mnt_parent)
+       if (!mnt_has_parent(old))
                goto out1;
 
        if (S_ISDIR(path->dentry->d_inode->i_mode) !=
@@ -1695,108 +1743,183 @@ static int do_move_mount(struct path *path, char *old_name)
        /*
         * Don't move a mount residing in a shared parent.
         */
-       if (old_path.mnt->mnt_parent &&
-           IS_MNT_SHARED(old_path.mnt->mnt_parent))
+       if (IS_MNT_SHARED(old->mnt_parent))
                goto out1;
        /*
         * Don't move a mount tree containing unbindable mounts to a destination
         * mount which is shared.
         */
-       if (IS_MNT_SHARED(path->mnt) &&
-           tree_contains_unbindable(old_path.mnt))
+       if (IS_MNT_SHARED(p) && tree_contains_unbindable(old))
                goto out1;
        err = -ELOOP;
-       for (p = path->mnt; p->mnt_parent != p; p = p->mnt_parent)
-               if (p == old_path.mnt)
+       for (; mnt_has_parent(p); p = p->mnt_parent)
+               if (p == old)
                        goto out1;
 
-       err = attach_recursive_mnt(old_path.mnt, path, &parent_path);
+       err = attach_recursive_mnt(old, path, &parent_path);
        if (err)
                goto out1;
 
        /* if the mount is moved, it should no longer be expire
         * automatically */
-       list_del_init(&old_path.mnt->mnt_expire);
+       list_del_init(&old->mnt_expire);
 out1:
-       mutex_unlock(&path->dentry->d_inode->i_mutex);
+       unlock_mount(path);
 out:
-       up_write(&namespace_sem);
        if (!err)
                path_put(&parent_path);
        path_put(&old_path);
        return err;
 }
 
-/*
- * create a new mount for userspace and request it to be added into the
- * namespace's tree
- */
-static int do_new_mount(struct path *path, char *type, int flags,
-                       int mnt_flags, char *name, void *data)
+static struct vfsmount *fs_set_subtype(struct vfsmount *mnt, const char *fstype)
 {
-       struct vfsmount *mnt;
-
-       if (!type)
-               return -EINVAL;
+       int err;
+       const char *subtype = strchr(fstype, '.');
+       if (subtype) {
+               subtype++;
+               err = -EINVAL;
+               if (!subtype[0])
+                       goto err;
+       } else
+               subtype = "";
 
-       /* we need capabilities... */
-       if (!capable(CAP_SYS_ADMIN))
-               return -EPERM;
+       mnt->mnt_sb->s_subtype = kstrdup(subtype, GFP_KERNEL);
+       err = -ENOMEM;
+       if (!mnt->mnt_sb->s_subtype)
+               goto err;
+       return mnt;
 
-       mnt = do_kern_mount(type, flags, name, data);
-       if (IS_ERR(mnt))
-               return PTR_ERR(mnt);
+ err:
+       mntput(mnt);
+       return ERR_PTR(err);
+}
 
-       return do_add_mount(mnt, path, mnt_flags, NULL);
+static struct vfsmount *
+do_kern_mount(const char *fstype, int flags, const char *name, void *data)
+{
+       struct file_system_type *type = get_fs_type(fstype);
+       struct vfsmount *mnt;
+       if (!type)
+               return ERR_PTR(-ENODEV);
+       mnt = vfs_kern_mount(type, flags, name, data);
+       if (!IS_ERR(mnt) && (type->fs_flags & FS_HAS_SUBTYPE) &&
+           !mnt->mnt_sb->s_subtype)
+               mnt = fs_set_subtype(mnt, fstype);
+       put_filesystem(type);
+       return mnt;
 }
 
 /*
  * add a mount into a namespace's mount tree
- * - provide the option of adding the new mount to an expiration list
  */
-int do_add_mount(struct vfsmount *newmnt, struct path *path,
-                int mnt_flags, struct list_head *fslist)
+static int do_add_mount(struct mount *newmnt, struct path *path, int mnt_flags)
 {
        int err;
 
        mnt_flags &= ~(MNT_SHARED | MNT_WRITE_HOLD | MNT_INTERNAL);
 
-       down_write(&namespace_sem);
-       /* Something was mounted here while we slept */
-       while (d_mountpoint(path->dentry) &&
-              follow_down(path))
-               ;
+       err = lock_mount(path);
+       if (err)
+               return err;
+
        err = -EINVAL;
-       if (!(mnt_flags & MNT_SHRINKABLE) && !check_mnt(path->mnt))
+       if (!(mnt_flags & MNT_SHRINKABLE) && !check_mnt(real_mount(path->mnt)))
                goto unlock;
 
        /* Refuse the same filesystem on the same mount point */
        err = -EBUSY;
-       if (path->mnt->mnt_sb == newmnt->mnt_sb &&
+       if (path->mnt->mnt_sb == newmnt->mnt.mnt_sb &&
            path->mnt->mnt_root == path->dentry)
                goto unlock;
 
        err = -EINVAL;
-       if (S_ISLNK(newmnt->mnt_root->d_inode->i_mode))
+       if (S_ISLNK(newmnt->mnt.mnt_root->d_inode->i_mode))
                goto unlock;
 
-       newmnt->mnt_flags = mnt_flags;
-       if ((err = graft_tree(newmnt, path)))
-               goto unlock;
+       newmnt->mnt.mnt_flags = mnt_flags;
+       err = graft_tree(newmnt, path);
 
-       if (fslist) /* add to the specified expiration list */
-               list_add_tail(&newmnt->mnt_expire, fslist);
+unlock:
+       unlock_mount(path);
+       return err;
+}
 
-       up_write(&namespace_sem);
-       return 0;
+/*
+ * create a new mount for userspace and request it to be added into the
+ * namespace's tree
+ */
+static int do_new_mount(struct path *path, char *type, int flags,
+                       int mnt_flags, char *name, void *data)
+{
+       struct vfsmount *mnt;
+       int err;
 
-unlock:
-       up_write(&namespace_sem);
-       mntput(newmnt);
+       if (!type)
+               return -EINVAL;
+
+       /* we need capabilities... */
+       if (!capable(CAP_SYS_ADMIN))
+               return -EPERM;
+
+       mnt = do_kern_mount(type, flags, name, data);
+       if (IS_ERR(mnt))
+               return PTR_ERR(mnt);
+
+       err = do_add_mount(real_mount(mnt), path, mnt_flags);
+       if (err)
+               mntput(mnt);
        return err;
 }
 
-EXPORT_SYMBOL_GPL(do_add_mount);
+int finish_automount(struct vfsmount *m, struct path *path)
+{
+       struct mount *mnt = real_mount(m);
+       int err;
+       /* The new mount record should have at least 2 refs to prevent it being
+        * expired before we get a chance to add it
+        */
+       BUG_ON(mnt_get_count(mnt) < 2);
+
+       if (m->mnt_sb == path->mnt->mnt_sb &&
+           m->mnt_root == path->dentry) {
+               err = -ELOOP;
+               goto fail;
+       }
+
+       err = do_add_mount(mnt, path, path->mnt->mnt_flags | MNT_SHRINKABLE);
+       if (!err)
+               return 0;
+fail:
+       /* remove m from any expiration list it may be on */
+       if (!list_empty(&mnt->mnt_expire)) {
+               down_write(&namespace_sem);
+               br_write_lock(vfsmount_lock);
+               list_del_init(&mnt->mnt_expire);
+               br_write_unlock(vfsmount_lock);
+               up_write(&namespace_sem);
+       }
+       mntput(m);
+       mntput(m);
+       return err;
+}
+
+/**
+ * mnt_set_expiry - Put a mount on an expiration list
+ * @mnt: The mount to list.
+ * @expiry_list: The list to add the mount to.
+ */
+void mnt_set_expiry(struct vfsmount *mnt, struct list_head *expiry_list)
+{
+       down_write(&namespace_sem);
+       br_write_lock(vfsmount_lock);
+
+       list_add_tail(&real_mount(mnt)->mnt_expire, expiry_list);
+
+       br_write_unlock(vfsmount_lock);
+       up_write(&namespace_sem);
+}
+EXPORT_SYMBOL(mnt_set_expiry);
 
 /*
  * process a list of expirable mountpoints with the intent of discarding any
@@ -1805,7 +1928,7 @@ EXPORT_SYMBOL_GPL(do_add_mount);
  */
 void mark_mounts_for_expiry(struct list_head *mounts)
 {
-       struct vfsmount *mnt, *next;
+       struct mount *mnt, *next;
        LIST_HEAD(graveyard);
        LIST_HEAD(umounts);
 
@@ -1828,7 +1951,7 @@ void mark_mounts_for_expiry(struct list_head *mounts)
                list_move(&mnt->mnt_expire, &graveyard);
        }
        while (!list_empty(&graveyard)) {
-               mnt = list_first_entry(&graveyard, struct vfsmount, mnt_expire);
+               mnt = list_first_entry(&graveyard, struct mount, mnt_expire);
                touch_mnt_namespace(mnt->mnt_ns);
                umount_tree(mnt, 1, &umounts);
        }
@@ -1846,9 +1969,9 @@ EXPORT_SYMBOL_GPL(mark_mounts_for_expiry);
  * search the list of submounts for a given mountpoint, and move any
  * shrinkable submounts to the 'graveyard' list.
  */
-static int select_submounts(struct vfsmount *parent, struct list_head *graveyard)
+static int select_submounts(struct mount *parent, struct list_head *graveyard)
 {
-       struct vfsmount *this_parent = parent;
+       struct mount *this_parent = parent;
        struct list_head *next;
        int found = 0;
 
@@ -1857,10 +1980,10 @@ repeat:
 resume:
        while (next != &this_parent->mnt_mounts) {
                struct list_head *tmp = next;
-               struct vfsmount *mnt = list_entry(tmp, struct vfsmount, mnt_child);
+               struct mount *mnt = list_entry(tmp, struct mount, mnt_child);
 
                next = tmp->next;
-               if (!(mnt->mnt_flags & MNT_SHRINKABLE))
+               if (!(mnt->mnt.mnt_flags & MNT_SHRINKABLE))
                        continue;
                /*
                 * Descend a level if the d_mounts list is non-empty.
@@ -1892,15 +2015,15 @@ resume:
  *
  * vfsmount_lock must be held for write
  */
-static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts)
+static void shrink_submounts(struct mount *mnt, struct list_head *umounts)
 {
        LIST_HEAD(graveyard);
-       struct vfsmount *m;
+       struct mount *m;
 
        /* extract submounts of 'mountpoint' from the expiration list */
        while (select_submounts(mnt, &graveyard)) {
                while (!list_empty(&graveyard)) {
-                       m = list_first_entry(&graveyard, struct vfsmount,
+                       m = list_first_entry(&graveyard, struct mount,
                                                mnt_expire);
                        touch_mnt_namespace(m->mnt_ns);
                        umount_tree(m, 1, umounts);
@@ -2085,6 +2208,23 @@ static struct mnt_namespace *alloc_mnt_ns(void)
        return new_ns;
 }
 
+void mnt_make_longterm(struct vfsmount *mnt)
+{
+       __mnt_make_longterm(real_mount(mnt));
+}
+
+void mnt_make_shortterm(struct vfsmount *m)
+{
+#ifdef CONFIG_SMP
+       struct mount *mnt = real_mount(m);
+       if (atomic_add_unless(&mnt->mnt_longterm, -1, 1))
+               return;
+       br_write_lock(vfsmount_lock);
+       atomic_dec(&mnt->mnt_longterm);
+       br_write_unlock(vfsmount_lock);
+#endif
+}
+
 /*
  * Allocate a new namespace structure and populate it with contents
  * copied from the namespace of the passed in task structure.
@@ -2094,7 +2234,9 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns,
 {
        struct mnt_namespace *new_ns;
        struct vfsmount *rootmnt = NULL, *pwdmnt = NULL;
-       struct vfsmount *p, *q;
+       struct mount *p, *q;
+       struct mount *old = mnt_ns->root;
+       struct mount *new;
 
        new_ns = alloc_mnt_ns();
        if (IS_ERR(new_ns))
@@ -2102,15 +2244,15 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns,
 
        down_write(&namespace_sem);
        /* First pass: copy the tree topology */
-       new_ns->root = copy_tree(mnt_ns->root, mnt_ns->root->mnt_root,
-                                       CL_COPY_ALL | CL_EXPIRE);
-       if (!new_ns->root) {
+       new = copy_tree(old, old->mnt.mnt_root, CL_COPY_ALL | CL_EXPIRE);
+       if (!new) {
                up_write(&namespace_sem);
                kfree(new_ns);
                return ERR_PTR(-ENOMEM);
        }
+       new_ns->root = new;
        br_write_lock(vfsmount_lock);
-       list_add_tail(&new_ns->list, &new_ns->root->mnt_list);
+       list_add_tail(&new_ns->list, &new->mnt_list);
        br_write_unlock(vfsmount_lock);
 
        /*
@@ -2118,22 +2260,27 @@ static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns,
         * as belonging to new namespace.  We have already acquired a private
         * fs_struct, so tsk->fs->lock is not needed.
         */
-       p = mnt_ns->root;
-       q = new_ns->root;
+       p = old;
+       q = new;
        while (p) {
                q->mnt_ns = new_ns;
+               __mnt_make_longterm(q);
                if (fs) {
-                       if (p == fs->root.mnt) {
-                               rootmnt = p;
-                               fs->root.mnt = mntget(q);
+                       if (&p->mnt == fs->root.mnt) {
+                               fs->root.mnt = mntget(&q->mnt);
+                               __mnt_make_longterm(q);
+                               mnt_make_shortterm(&p->mnt);
+                               rootmnt = &p->mnt;
                        }
-                       if (p == fs->pwd.mnt) {
-                               pwdmnt = p;
-                               fs->pwd.mnt = mntget(q);
+                       if (&p->mnt == fs->pwd.mnt) {
+                               fs->pwd.mnt = mntget(&q->mnt);
+                               __mnt_make_longterm(q);
+                               mnt_make_shortterm(&p->mnt);
+                               pwdmnt = &p->mnt;
                        }
                }
-               p = next_mnt(p, mnt_ns->root);
-               q = next_mnt(q, new_ns->root);
+               p = next_mnt(p, old);
+               q = next_mnt(q, new);
        }
        up_write(&namespace_sem);
 
@@ -2166,19 +2313,50 @@ struct mnt_namespace *copy_mnt_ns(unsigned long flags, struct mnt_namespace *ns,
  * create_mnt_ns - creates a private namespace and adds a root filesystem
  * @mnt: pointer to the new root filesystem mountpoint
  */
-struct mnt_namespace *create_mnt_ns(struct vfsmount *mnt)
+static struct mnt_namespace *create_mnt_ns(struct vfsmount *m)
 {
-       struct mnt_namespace *new_ns;
-
-       new_ns = alloc_mnt_ns();
+       struct mnt_namespace *new_ns = alloc_mnt_ns();
        if (!IS_ERR(new_ns)) {
+               struct mount *mnt = real_mount(m);
                mnt->mnt_ns = new_ns;
+               __mnt_make_longterm(mnt);
                new_ns->root = mnt;
-               list_add(&new_ns->list, &new_ns->root->mnt_list);
+               list_add(&new_ns->list, &mnt->mnt_list);
+       } else {
+               mntput(m);
        }
        return new_ns;
 }
-EXPORT_SYMBOL(create_mnt_ns);
+
+struct dentry *mount_subtree(struct vfsmount *mnt, const char *name)
+{
+       struct mnt_namespace *ns;
+       struct super_block *s;
+       struct path path;
+       int err;
+
+       ns = create_mnt_ns(mnt);
+       if (IS_ERR(ns))
+               return ERR_CAST(ns);
+
+       err = vfs_path_lookup(mnt->mnt_root, mnt,
+                       name, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &path);
+
+       put_mnt_ns(ns);
+
+       if (err)
+               return ERR_PTR(err);
+
+       /* trade a vfsmount reference for active sb one */
+       s = path.mnt->mnt_sb;
+       atomic_inc(&s->s_active);
+       mntput(path.mnt);
+       /* lock the sucker */
+       down_write(&s->s_umount);
+       /* ... and return the root of (sub)tree on it */
+       return path.dentry;
+}
+EXPORT_SYMBOL(mount_subtree);
 
 SYSCALL_DEFINE5(mount, char __user *, dev_name, char __user *, dir_name,
                char __user *, type, unsigned long, flags, void __user *, data)
@@ -2222,6 +2400,31 @@ out_type:
 }
 
 /*
+ * Return true if path is reachable from root
+ *
+ * namespace_sem or vfsmount_lock is held
+ */
+bool is_path_reachable(struct mount *mnt, struct dentry *dentry,
+                        const struct path *root)
+{
+       while (&mnt->mnt != root->mnt && mnt_has_parent(mnt)) {
+               dentry = mnt->mnt_mountpoint;
+               mnt = mnt->mnt_parent;
+       }
+       return &mnt->mnt == root->mnt && is_subdir(dentry, root->dentry);
+}
+
+int path_is_under(struct path *path1, struct path *path2)
+{
+       int res;
+       br_read_lock(vfsmount_lock);
+       res = is_path_reachable(real_mount(path1->mnt), path1->dentry, path2);
+       br_read_unlock(vfsmount_lock);
+       return res;
+}
+EXPORT_SYMBOL(path_is_under);
+
+/*
  * pivot_root Semantics:
  * Moves the root file system of the current process to the directory put_old,
  * makes new_root as the new root file system of the current process, and sets
@@ -2249,8 +2452,8 @@ out_type:
 SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
                const char __user *, put_old)
 {
-       struct vfsmount *tmp;
        struct path new, old, parent_path, root_parent, root;
+       struct mount *new_mnt, *root_mnt;
        int error;
 
        if (!capable(CAP_SYS_ADMIN))
@@ -2259,89 +2462,75 @@ SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
        error = user_path_dir(new_root, &new);
        if (error)
                goto out0;
-       error = -EINVAL;
-       if (!check_mnt(new.mnt))
-               goto out1;
 
        error = user_path_dir(put_old, &old);
        if (error)
                goto out1;
 
        error = security_sb_pivotroot(&old, &new);
-       if (error) {
-               path_put(&old);
-               goto out1;
-       }
+       if (error)
+               goto out2;
 
        get_fs_root(current->fs, &root);
-       down_write(&namespace_sem);
-       mutex_lock(&old.dentry->d_inode->i_mutex);
+       error = lock_mount(&old);
+       if (error)
+               goto out3;
+
        error = -EINVAL;
-       if (IS_MNT_SHARED(old.mnt) ||
-               IS_MNT_SHARED(new.mnt->mnt_parent) ||
-               IS_MNT_SHARED(root.mnt->mnt_parent))
-               goto out2;
-       if (!check_mnt(root.mnt))
-               goto out2;
+       new_mnt = real_mount(new.mnt);
+       root_mnt = real_mount(root.mnt);
+       if (IS_MNT_SHARED(real_mount(old.mnt)) ||
+               IS_MNT_SHARED(new_mnt->mnt_parent) ||
+               IS_MNT_SHARED(root_mnt->mnt_parent))
+               goto out4;
+       if (!check_mnt(root_mnt) || !check_mnt(new_mnt))
+               goto out4;
        error = -ENOENT;
-       if (cant_mount(old.dentry))
-               goto out2;
        if (d_unlinked(new.dentry))
-               goto out2;
+               goto out4;
        if (d_unlinked(old.dentry))
-               goto out2;
+               goto out4;
        error = -EBUSY;
        if (new.mnt == root.mnt ||
            old.mnt == root.mnt)
-               goto out2; /* loop, on the same file system  */
+               goto out4; /* loop, on the same file system  */
        error = -EINVAL;
        if (root.mnt->mnt_root != root.dentry)
-               goto out2; /* not a mountpoint */
-       if (root.mnt->mnt_parent == root.mnt)
-               goto out2; /* not attached */
+               goto out4; /* not a mountpoint */
+       if (!mnt_has_parent(root_mnt))
+               goto out4; /* not attached */
        if (new.mnt->mnt_root != new.dentry)
-               goto out2; /* not a mountpoint */
-       if (new.mnt->mnt_parent == new.mnt)
-               goto out2; /* not attached */
+               goto out4; /* not a mountpoint */
+       if (!mnt_has_parent(new_mnt))
+               goto out4; /* not attached */
        /* make sure we can reach put_old from new_root */
-       tmp = old.mnt;
+       if (!is_path_reachable(real_mount(old.mnt), old.dentry, &new))
+               goto out4;
        br_write_lock(vfsmount_lock);
-       if (tmp != new.mnt) {
-               for (;;) {
-                       if (tmp->mnt_parent == tmp)
-                               goto out3; /* already mounted on put_old */
-                       if (tmp->mnt_parent == new.mnt)
-                               break;
-                       tmp = tmp->mnt_parent;
-               }
-               if (!is_subdir(tmp->mnt_mountpoint, new.dentry))
-                       goto out3;
-       } else if (!is_subdir(old.dentry, new.dentry))
-               goto out3;
-       detach_mnt(new.mnt, &parent_path);
-       detach_mnt(root.mnt, &root_parent);
+       detach_mnt(new_mnt, &parent_path);
+       detach_mnt(root_mnt, &root_parent);
        /* mount old root on put_old */
-       attach_mnt(root.mnt, &old);
+       attach_mnt(root_mnt, &old);
        /* mount new_root on / */
-       attach_mnt(new.mnt, &root_parent);
+       attach_mnt(new_mnt, &root_parent);
        touch_mnt_namespace(current->nsproxy->mnt_ns);
        br_write_unlock(vfsmount_lock);
        chroot_fs_refs(&root, &new);
        error = 0;
-       path_put(&root_parent);
-       path_put(&parent_path);
-out2:
-       mutex_unlock(&old.dentry->d_inode->i_mutex);
-       up_write(&namespace_sem);
+out4:
+       unlock_mount(&old);
+       if (!error) {
+               path_put(&root_parent);
+               path_put(&parent_path);
+       }
+out3:
        path_put(&root);
+out2:
        path_put(&old);
 out1:
        path_put(&new);
 out0:
        return error;
-out3:
-       br_write_unlock(vfsmount_lock);
-       goto out2;
 }
 
 static void __init init_mount_tree(void)
@@ -2353,6 +2542,7 @@ static void __init init_mount_tree(void)
        mnt = do_kern_mount("rootfs", 0, "rootfs", NULL);
        if (IS_ERR(mnt))
                panic("Can't create rootfs");
+
        ns = create_mnt_ns(mnt);
        if (IS_ERR(ns))
                panic("Can't allocate initial namespace");
@@ -2360,8 +2550,8 @@ static void __init init_mount_tree(void)
        init_task.nsproxy->mnt_ns = ns;
        get_mnt_ns(ns);
 
-       root.mnt = ns->root;
-       root.dentry = ns->root->mnt_root;
+       root.mnt = mnt;
+       root.dentry = mnt->mnt_root;
 
        set_fs_pwd(current->fs, &root);
        set_fs_root(current->fs, &root);
@@ -2374,7 +2564,7 @@ void __init mnt_init(void)
 
        init_rwsem(&namespace_sem);
 
-       mnt_cache = kmem_cache_create("mnt_cache", sizeof(struct vfsmount),
+       mnt_cache = kmem_cache_create("mnt_cache", sizeof(struct mount),
                        0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, NULL);
 
        mount_hashtable = (struct list_head *)__get_free_page(GFP_ATOMIC);
@@ -2382,7 +2572,7 @@ void __init mnt_init(void)
        if (!mount_hashtable)
                panic("Failed to allocate mount hash table\n");
 
-       printk("Mount-cache hash table entries: %lu\n", HASH_SIZE);
+       printk(KERN_INFO "Mount-cache hash table entries: %lu\n", HASH_SIZE);
 
        for (u = 0; u < HASH_SIZE; u++)
                INIT_LIST_HEAD(&mount_hashtable[u]);
@@ -2414,4 +2604,33 @@ void put_mnt_ns(struct mnt_namespace *ns)
        release_mounts(&umount_list);
        kfree(ns);
 }
-EXPORT_SYMBOL(put_mnt_ns);
+
+struct vfsmount *kern_mount_data(struct file_system_type *type, void *data)
+{
+       struct vfsmount *mnt;
+       mnt = vfs_kern_mount(type, MS_KERNMOUNT, type->name, data);
+       if (!IS_ERR(mnt)) {
+               /*
+                * it is a longterm mount, don't release mnt until
+                * we unmount before file sys is unregistered
+               */
+               mnt_make_longterm(mnt);
+       }
+       return mnt;
+}
+EXPORT_SYMBOL_GPL(kern_mount_data);
+
+void kern_unmount(struct vfsmount *mnt)
+{
+       /* release long term mount so mount point can be released */
+       if (!IS_ERR_OR_NULL(mnt)) {
+               mnt_make_shortterm(mnt);
+               mntput(mnt);
+       }
+}
+EXPORT_SYMBOL(kern_unmount);
+
+bool our_mnt(struct vfsmount *mnt)
+{
+       return check_mnt(real_mount(mnt));
+}