]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blobdiff - fs/namespace.c
CIFS: Fix maximum SMB2 header size
[mirror_ubuntu-zesty-kernel.git] / fs / namespace.c
index b5b1259e064f8d9661110ba1f1f73d1a0ff19d51..44bcebc9ba48b580b067a880583615047ff10f34 100644 (file)
@@ -462,6 +462,7 @@ void __mnt_drop_write(struct vfsmount *mnt)
        mnt_dec_writers(real_mount(mnt));
        preempt_enable();
 }
+EXPORT_SYMBOL_GPL(__mnt_drop_write);
 
 /**
  * mnt_drop_write - give up write access to a mount
@@ -636,28 +637,6 @@ struct mount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry)
        return NULL;
 }
 
-/*
- * find the last mount at @dentry on vfsmount @mnt.
- * mount_lock must be held.
- */
-struct mount *__lookup_mnt_last(struct vfsmount *mnt, struct dentry *dentry)
-{
-       struct mount *p, *res = NULL;
-       p = __lookup_mnt(mnt, dentry);
-       if (!p)
-               goto out;
-       if (!(p->mnt.mnt_flags & MNT_UMOUNT))
-               res = p;
-       hlist_for_each_entry_continue(p, mnt_hash) {
-               if (&p->mnt_parent->mnt != mnt || p->mnt_mountpoint != dentry)
-                       break;
-               if (!(p->mnt.mnt_flags & MNT_UMOUNT))
-                       res = p;
-       }
-out:
-       return res;
-}
-
 /*
  * lookup_mnt - Return the first child mount mounted at path
  *
@@ -742,26 +721,50 @@ static struct mountpoint *lookup_mountpoint(struct dentry *dentry)
        return NULL;
 }
 
-static struct mountpoint *new_mountpoint(struct dentry *dentry)
+static struct mountpoint *get_mountpoint(struct dentry *dentry)
 {
-       struct hlist_head *chain = mp_hash(dentry);
-       struct mountpoint *mp;
+       struct mountpoint *mp, *new = NULL;
        int ret;
 
-       mp = kmalloc(sizeof(struct mountpoint), GFP_KERNEL);
-       if (!mp)
+       if (d_mountpoint(dentry)) {
+mountpoint:
+               read_seqlock_excl(&mount_lock);
+               mp = lookup_mountpoint(dentry);
+               read_sequnlock_excl(&mount_lock);
+               if (mp)
+                       goto done;
+       }
+
+       if (!new)
+               new = kmalloc(sizeof(struct mountpoint), GFP_KERNEL);
+       if (!new)
                return ERR_PTR(-ENOMEM);
 
+
+       /* Exactly one processes may set d_mounted */
        ret = d_set_mounted(dentry);
-       if (ret) {
-               kfree(mp);
-               return ERR_PTR(ret);
-       }
 
-       mp->m_dentry = dentry;
-       mp->m_count = 1;
-       hlist_add_head(&mp->m_hash, chain);
-       INIT_HLIST_HEAD(&mp->m_list);
+       /* Someone else set d_mounted? */
+       if (ret == -EBUSY)
+               goto mountpoint;
+
+       /* The dentry is not available as a mountpoint? */
+       mp = ERR_PTR(ret);
+       if (ret)
+               goto done;
+
+       /* Add the new mountpoint to the hash table */
+       read_seqlock_excl(&mount_lock);
+       new->m_dentry = dentry;
+       new->m_count = 1;
+       hlist_add_head(&new->m_hash, mp_hash(dentry));
+       INIT_HLIST_HEAD(&new->m_list);
+       read_sequnlock_excl(&mount_lock);
+
+       mp = new;
+       new = NULL;
+done:
+       kfree(new);
        return mp;
 }
 
@@ -854,6 +857,13 @@ void mnt_set_mountpoint(struct mount *mnt,
        hlist_add_head(&child_mnt->mnt_mp_list, &mp->m_list);
 }
 
+static void __attach_mnt(struct mount *mnt, struct mount *parent)
+{
+       hlist_add_head_rcu(&mnt->mnt_hash,
+                          m_hash(&parent->mnt, mnt->mnt_mountpoint));
+       list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
+}
+
 /*
  * vfsmount lock must be held for write
  */
@@ -862,28 +872,45 @@ static void attach_mnt(struct mount *mnt,
                        struct mountpoint *mp)
 {
        mnt_set_mountpoint(parent, mp, mnt);
-       hlist_add_head_rcu(&mnt->mnt_hash, m_hash(&parent->mnt, mp->m_dentry));
-       list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
+       __attach_mnt(mnt, parent);
 }
 
-static void attach_shadowed(struct mount *mnt,
-                       struct mount *parent,
-                       struct mount *shadows)
+void mnt_change_mountpoint(struct mount *parent, struct mountpoint *mp, struct mount *mnt)
 {
-       if (shadows) {
-               hlist_add_behind_rcu(&mnt->mnt_hash, &shadows->mnt_hash);
-               list_add(&mnt->mnt_child, &shadows->mnt_child);
-       } else {
-               hlist_add_head_rcu(&mnt->mnt_hash,
-                               m_hash(&parent->mnt, mnt->mnt_mountpoint));
-               list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
-       }
+       struct mountpoint *old_mp = mnt->mnt_mp;
+       struct dentry *old_mountpoint = mnt->mnt_mountpoint;
+       struct mount *old_parent = mnt->mnt_parent;
+
+       list_del_init(&mnt->mnt_child);
+       hlist_del_init(&mnt->mnt_mp_list);
+       hlist_del_init_rcu(&mnt->mnt_hash);
+
+       attach_mnt(mnt, parent, mp);
+
+       put_mountpoint(old_mp);
+
+       /*
+        * Safely avoid even the suggestion this code might sleep or
+        * lock the mount hash by taking advantage of the knowledge that
+        * mnt_change_mountpoint will not release the final reference
+        * to a mountpoint.
+        *
+        * During mounting, the mount passed in as the parent mount will
+        * continue to use the old mountpoint and during unmounting, the
+        * old mountpoint will continue to exist until namespace_unlock,
+        * which happens well after mnt_change_mountpoint.
+        */
+       spin_lock(&old_mountpoint->d_lock);
+       old_mountpoint->d_lockref.count--;
+       spin_unlock(&old_mountpoint->d_lock);
+
+       mnt_add_count(old_parent, -1);
 }
 
 /*
  * vfsmount lock must be held for write
  */
-static void commit_tree(struct mount *mnt, struct mount *shadows)
+static void commit_tree(struct mount *mnt)
 {
        struct mount *parent = mnt->mnt_parent;
        struct mount *m;
@@ -901,7 +928,7 @@ static void commit_tree(struct mount *mnt, struct mount *shadows)
        n->mounts += n->pending_mounts;
        n->pending_mounts = 0;
 
-       attach_shadowed(mnt, parent, shadows);
+       __attach_mnt(mnt, parent);
        touch_mnt_namespace(n);
 }
 
@@ -965,6 +992,21 @@ vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void
 }
 EXPORT_SYMBOL_GPL(vfs_kern_mount);
 
+struct vfsmount *
+vfs_submount(const struct dentry *mountpoint, struct file_system_type *type,
+            const char *name, void *data)
+{
+       /* Until it is worked out how to pass the user namespace
+        * through from the parent mount to the submount don't support
+        * unprivileged mounts with submounts.
+        */
+       if (mountpoint->d_sb->s_user_ns != &init_user_ns)
+               return ERR_PTR(-EPERM);
+
+       return vfs_kern_mount(type, MS_SUBMOUNT, name, data);
+}
+EXPORT_SYMBOL_GPL(vfs_submount);
+
 static struct mount *clone_mnt(struct mount *old, struct dentry *root,
                                        int flag)
 {
@@ -1548,7 +1590,7 @@ static int do_umount(struct mount *mnt, int flags)
                 * Special case for "unmounting" root ...
                 * we just try to remount it readonly.
                 */
-               if (!capable(CAP_SYS_ADMIN))
+               if (!ns_capable(sb->s_user_ns, CAP_SYS_ADMIN))
                        return -EPERM;
                down_write(&sb->s_umount);
                if (!(sb->s_flags & MS_RDONLY))
@@ -1595,11 +1637,11 @@ void __detach_mounts(struct dentry *dentry)
        struct mount *mnt;
 
        namespace_lock();
+       lock_mount_hash();
        mp = lookup_mountpoint(dentry);
        if (IS_ERR_OR_NULL(mp))
                goto out_unlock;
 
-       lock_mount_hash();
        event++;
        while (!hlist_empty(&mp->m_list)) {
                mnt = hlist_entry(mp->m_list.first, struct mount, mnt_mp_list);
@@ -1609,9 +1651,9 @@ void __detach_mounts(struct dentry *dentry)
                }
                else umount_tree(mnt, UMOUNT_CONNECTED);
        }
-       unlock_mount_hash();
        put_mountpoint(mp);
 out_unlock:
+       unlock_mount_hash();
        namespace_unlock();
 }
 
@@ -1740,7 +1782,6 @@ struct mount *copy_tree(struct mount *mnt, struct dentry *dentry,
                        continue;
 
                for (s = r; s; s = next_mnt(s, r)) {
-                       struct mount *t = NULL;
                        if (!(flag & CL_COPY_UNBINDABLE) &&
                            IS_MNT_UNBINDABLE(s)) {
                                s = skip_mnt_tree(s);
@@ -1762,14 +1803,7 @@ struct mount *copy_tree(struct mount *mnt, struct dentry *dentry,
                                goto out;
                        lock_mount_hash();
                        list_add_tail(&q->mnt_list, &res->mnt_list);
-                       mnt_set_mountpoint(parent, p->mnt_mp, q);
-                       if (!list_empty(&parent->mnt_mounts)) {
-                               t = list_last_entry(&parent->mnt_mounts,
-                                       struct mount, mnt_child);
-                               if (t->mnt_mp != p->mnt_mp)
-                                       t = NULL;
-                       }
-                       attach_shadowed(q, parent, t);
+                       attach_mnt(q, parent, p->mnt_mp);
                        unlock_mount_hash();
                }
        }
@@ -1848,6 +1882,7 @@ int iterate_mounts(int (*f)(struct vfsmount *, void *), void *arg,
        }
        return 0;
 }
+EXPORT_SYMBOL_GPL(iterate_mounts);
 
 static void cleanup_group_ids(struct mount *mnt, struct mount *end)
 {
@@ -1968,10 +2003,18 @@ static int attach_recursive_mnt(struct mount *source_mnt,
 {
        HLIST_HEAD(tree_list);
        struct mnt_namespace *ns = dest_mnt->mnt_ns;
+       struct mountpoint *smp;
        struct mount *child, *p;
        struct hlist_node *n;
        int err;
 
+       /* Preallocate a mountpoint in case the new mounts need
+        * to be tucked under other mounts.
+        */
+       smp = get_mountpoint(source_mnt->mnt.mnt_root);
+       if (IS_ERR(smp))
+               return PTR_ERR(smp);
+
        /* Is there space to add these mounts to the mount namespace? */
        if (!parent_path) {
                err = count_mounts(ns, source_mnt);
@@ -1998,16 +2041,19 @@ static int attach_recursive_mnt(struct mount *source_mnt,
                touch_mnt_namespace(source_mnt->mnt_ns);
        } else {
                mnt_set_mountpoint(dest_mnt, dest_mp, source_mnt);
-               commit_tree(source_mnt, NULL);
+               commit_tree(source_mnt);
        }
 
        hlist_for_each_entry_safe(child, n, &tree_list, mnt_hash) {
                struct mount *q;
                hlist_del_init(&child->mnt_hash);
-               q = __lookup_mnt_last(&child->mnt_parent->mnt,
-                                     child->mnt_mountpoint);
-               commit_tree(child, q);
+               q = __lookup_mnt(&child->mnt_parent->mnt,
+                                child->mnt_mountpoint);
+               if (q)
+                       mnt_change_mountpoint(child, smp, q);
+               commit_tree(child);
        }
+       put_mountpoint(smp);
        unlock_mount_hash();
 
        return 0;
@@ -2022,6 +2068,11 @@ static int attach_recursive_mnt(struct mount *source_mnt,
        cleanup_group_ids(source_mnt, NULL);
  out:
        ns->pending_mounts = 0;
+
+       read_seqlock_excl(&mount_lock);
+       put_mountpoint(smp);
+       read_sequnlock_excl(&mount_lock);
+
        return err;
 }
 
@@ -2038,9 +2089,7 @@ retry:
        namespace_lock();
        mnt = lookup_mnt(path);
        if (likely(!mnt)) {
-               struct mountpoint *mp = lookup_mountpoint(dentry);
-               if (!mp)
-                       mp = new_mountpoint(dentry);
+               struct mountpoint *mp = get_mountpoint(dentry);
                if (IS_ERR(mp)) {
                        namespace_unlock();
                        inode_unlock(dentry->d_inode);
@@ -2059,7 +2108,11 @@ retry:
 static void unlock_mount(struct mountpoint *where)
 {
        struct dentry *dentry = where->m_dentry;
+
+       read_seqlock_excl(&mount_lock);
        put_mountpoint(where);
+       read_sequnlock_excl(&mount_lock);
+
        namespace_unlock();
        inode_unlock(dentry->d_inode);
 }
@@ -2276,7 +2329,7 @@ static int do_remount(struct path *path, int flags, int mnt_flags,
        down_write(&sb->s_umount);
        if (flags & MS_BIND)
                err = change_mount_flags(path->mnt, flags);
-       else if (!capable(CAP_SYS_ADMIN))
+       else if (!ns_capable(sb->s_user_ns, CAP_SYS_ADMIN))
                err = -EPERM;
        else
                err = do_remount_sb(sb, flags, data, 0);
@@ -2768,7 +2821,7 @@ long do_mount(const char *dev_name, const char __user *dir_name,
 
        flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE | MS_BORN |
                   MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT |
-                  MS_STRICTATIME | MS_NOREMOTELOCK);
+                  MS_STRICTATIME | MS_NOREMOTELOCK | MS_SUBMOUNT);
 
        if (flags & MS_REMOUNT)
                retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
@@ -3135,9 +3188,9 @@ SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
        touch_mnt_namespace(current->nsproxy->mnt_ns);
        /* A moved mount should not expire automatically */
        list_del_init(&new_mnt->mnt_expire);
+       put_mountpoint(root_mp);
        unlock_mount_hash();
        chroot_fs_refs(&root, &new);
-       put_mountpoint(root_mp);
        error = 0;
 out4:
        unlock_mount(old_mp);