1 files changed, 187 insertions, 152 deletions

diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index bd88f25889f7..a8577a7f26ab 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -230,8 +230,8 @@ static noinline int create_subvol(struct btrfs_root *root,
         struct btrfs_root_item root_item;
         struct btrfs_inode_item *inode_item;
         struct extent_buffer *leaf;
-        struct btrfs_root *new_root = root;
-        struct inode *dir;
+        struct btrfs_root *new_root;
+        struct inode *dir = dentry->d_parent->d_inode;
         int ret;
         int err;
         u64 objectid;
@@ -241,7 +241,7 @@ static noinline int create_subvol(struct btrfs_root *root,
 
         ret = btrfs_check_metadata_free_space(root);
         if (ret)
-                goto fail_commit;
+                return ret;
 
         trans = btrfs_start_transaction(root, 1);
         BUG_ON(!trans);
@@ -304,11 +304,17 @@ static noinline int create_subvol(struct btrfs_root *root,
         if (ret)
                 goto fail;
 
+        key.offset = (u64)-1;
+        new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
+        BUG_ON(IS_ERR(new_root));
+
+        btrfs_record_root_in_trans(trans, new_root);
+
+        ret = btrfs_create_subvol_root(trans, new_root, new_dirid,
+                                       BTRFS_I(dir)->block_group);
         /*
          * insert the directory item
          */
-        key.offset = (u64)-1;
-        dir = dentry->d_parent->d_inode;
         ret = btrfs_set_inode_index(dir, &index);
         BUG_ON(ret);
 
@@ -322,44 +328,18 @@ static noinline int create_subvol(struct btrfs_root *root,
         ret = btrfs_update_inode(trans, root, dir);
         BUG_ON(ret);
 
-        /* add the backref first */
         ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
-                                 objectid, BTRFS_ROOT_BACKREF_KEY,
-                                 root->root_key.objectid,
+                                 objectid, root->root_key.objectid,
                                  dir->i_ino, index, name, namelen);
 
         BUG_ON(ret);
 
-        /* now add the forward ref */
-        ret = btrfs_add_root_ref(trans, root->fs_info->tree_root,
-                                 root->root_key.objectid, BTRFS_ROOT_REF_KEY,
-                                 objectid,
-                                 dir->i_ino, index, name, namelen);
-
-        BUG_ON(ret);
-
-        ret = btrfs_commit_transaction(trans, root);
-        if (ret)
-                goto fail_commit;
-
-        new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
-        BUG_ON(!new_root);
-
-        trans = btrfs_start_transaction(new_root, 1);
-        BUG_ON(!trans);
-
-        ret = btrfs_create_subvol_root(trans, new_root, dentry, new_dirid,
-                                       BTRFS_I(dir)->block_group);
-        if (ret)
-                goto fail;
-
+        d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
 fail:
         nr = trans->blocks_used;
-        err = btrfs_commit_transaction(trans, new_root);
+        err = btrfs_commit_transaction(trans, root);
         if (err && !ret)
                 ret = err;
-fail_commit:
-        btrfs_btree_balance_dirty(root, nr);
         return ret;
 }
 
@@ -420,14 +400,15 @@ static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
  * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
  * inside this filesystem so it's quite a bit simpler.
  */
-static noinline int btrfs_mksubvol(struct path *parent, char *name,
-                                   int mode, int namelen,
+static noinline int btrfs_mksubvol(struct path *parent,
+                                   char *name, int namelen,
                                    struct btrfs_root *snap_src)
 {
+        struct inode *dir  = parent->dentry->d_inode;
         struct dentry *dentry;
         int error;
 
-        mutex_lock_nested(&parent->dentry->d_inode->i_mutex, I_MUTEX_PARENT);
+        mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
 
         dentry = lookup_one_len(name, parent->dentry, namelen);
         error = PTR_ERR(dentry);
@@ -438,99 +419,39 @@ static noinline int btrfs_mksubvol(struct path *parent, char *name,
         if (dentry->d_inode)
                 goto out_dput;
 
-        if (!IS_POSIXACL(parent->dentry->d_inode))
-                mode &= ~current_umask();
-
         error = mnt_want_write(parent->mnt);
         if (error)
                 goto out_dput;
 
-        error = btrfs_may_create(parent->dentry->d_inode, dentry);
+        error = btrfs_may_create(dir, dentry);
         if (error)
                 goto out_drop_write;
 
-        /*
-         * Actually perform the low-level subvolume creation after all
-         * this VFS fuzz.
-         *
-         * Eventually we want to pass in an inode under which we create this
-         * subvolume, but for now all are under the filesystem root.
-         *
-         * Also we should pass on the mode eventually to allow creating new
-         * subvolume with specific mode bits.
-         */
+        down_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
+
+        if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
+                goto out_up_read;
+
         if (snap_src) {
-                struct dentry *dir = dentry->d_parent;
-                struct dentry *test = dir->d_parent;
-                struct btrfs_path *path = btrfs_alloc_path();
-                int ret;
-                u64 test_oid;
-                u64 parent_oid = BTRFS_I(dir->d_inode)->root->root_key.objectid;
-
-                test_oid = snap_src->root_key.objectid;
-
-                ret = btrfs_find_root_ref(snap_src->fs_info->tree_root,
-                                          path, parent_oid, test_oid);
-                if (ret == 0)
-                        goto create;
-                btrfs_release_path(snap_src->fs_info->tree_root, path);
-
-                /* we need to make sure we aren't creating a directory loop
-                 * by taking a snapshot of something that has our current
-                 * subvol in its directory tree.  So, this loops through
-                 * the dentries and checks the forward refs for each subvolume
-                 * to see if is references the subvolume where we are
-                 * placing this new snapshot.
-                 */
-                while (1) {
-                        if (!test ||
-                            dir == snap_src->fs_info->sb->s_root ||
-                            test == snap_src->fs_info->sb->s_root ||
-                            test->d_inode->i_sb != snap_src->fs_info->sb) {
-                                break;
-                        }
-                        if (S_ISLNK(test->d_inode->i_mode)) {
-                                printk(KERN_INFO "Btrfs symlink in snapshot "
-                                       "path, failed\n");
-                                error = -EMLINK;
-                                btrfs_free_path(path);
-                                goto out_drop_write;
-                        }
-                        test_oid =
-                                BTRFS_I(test->d_inode)->root->root_key.objectid;
-                        ret = btrfs_find_root_ref(snap_src->fs_info->tree_root,
-                                  path, test_oid, parent_oid);
-                        if (ret == 0) {
-                                printk(KERN_INFO "Btrfs snapshot creation "
-                                       "failed, looping\n");
-                                error = -EMLINK;
-                                btrfs_free_path(path);
-                                goto out_drop_write;
-                        }
-                        btrfs_release_path(snap_src->fs_info->tree_root, path);
-                        test = test->d_parent;
-                }
-create:
-                btrfs_free_path(path);
-                error = create_snapshot(snap_src, dentry, name, namelen);
+                error = create_snapshot(snap_src, dentry,
+                                        name, namelen);
         } else {
-                error = create_subvol(BTRFS_I(parent->dentry->d_inode)->root,
-                                      dentry, name, namelen);
+                error = create_subvol(BTRFS_I(dir)->root, dentry,
+                                      name, namelen);
         }
-        if (error)
-                goto out_drop_write;
-
-        fsnotify_mkdir(parent->dentry->d_inode, dentry);
+        if (!error)
+                fsnotify_mkdir(dir, dentry);
+out_up_read:
+        up_read(&BTRFS_I(dir)->root->fs_info->subvol_sem);
 out_drop_write:
         mnt_drop_write(parent->mnt);
 out_dput:
         dput(dentry);
 out_unlock:
-        mutex_unlock(&parent->dentry->d_inode->i_mutex);
+        mutex_unlock(&dir->i_mutex);
         return error;
 }
 
-
 static int btrfs_defrag_file(struct file *file)
 {
         struct inode *inode = fdentry(file)->d_inode;
@@ -596,9 +517,8 @@ again:
                 clear_page_dirty_for_io(page);
 
                 btrfs_set_extent_delalloc(inode, page_start, page_end);
-
-                unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
                 set_page_dirty(page);
+                unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
                 unlock_page(page);
                 page_cache_release(page);
                 balance_dirty_pages_ratelimited_nr(inode->i_mapping, 1);
@@ -609,7 +529,8 @@ out_unlock:
         return 0;
 }
 
-static int btrfs_ioctl_resize(struct btrfs_root *root, void __user *arg)
+static noinline int btrfs_ioctl_resize(struct btrfs_root *root,
+                                        void __user *arg)
 {
         u64 new_size;
         u64 old_size;
@@ -718,10 +639,7 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
 {
         struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
         struct btrfs_ioctl_vol_args *vol_args;
-        struct btrfs_dir_item *di;
-        struct btrfs_path *path;
         struct file *src_file;
-        u64 root_dirid;
         int namelen;
         int ret = 0;
 
@@ -739,32 +657,9 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
                 goto out;
         }
 
-        path = btrfs_alloc_path();
-        if (!path) {
-                ret = -ENOMEM;
-                goto out;
-        }
-
-        root_dirid = root->fs_info->sb->s_root->d_inode->i_ino,
-        di = btrfs_lookup_dir_item(NULL, root->fs_info->tree_root,
-                            path, root_dirid,
-                            vol_args->name, namelen, 0);
-        btrfs_free_path(path);
-
-        if (di && !IS_ERR(di)) {
-                ret = -EEXIST;
-                goto out;
-        }
-
-        if (IS_ERR(di)) {
-                ret = PTR_ERR(di);
-                goto out;
-        }
-
         if (subvol) {
-                ret = btrfs_mksubvol(&file->f_path, vol_args->name,
-                                     file->f_path.dentry->d_inode->i_mode,
-                                     namelen, NULL);
+                ret = btrfs_mksubvol(&file->f_path, vol_args->name, namelen,
+                                     NULL);
         } else {
                 struct inode *src_inode;
                 src_file = fget(vol_args->fd);
@@ -781,17 +676,156 @@ static noinline int btrfs_ioctl_snap_create(struct file *file,
                         fput(src_file);
                         goto out;
                 }
-                ret = btrfs_mksubvol(&file->f_path, vol_args->name,
-                             file->f_path.dentry->d_inode->i_mode,
-                             namelen, BTRFS_I(src_inode)->root);
+                ret = btrfs_mksubvol(&file->f_path, vol_args->name, namelen,
+                                     BTRFS_I(src_inode)->root);
                 fput(src_file);
         }
-
 out:
         kfree(vol_args);
         return ret;
 }
 
+/*
+ * helper to check if the subvolume references other subvolumes
+ */
+static noinline int may_destroy_subvol(struct btrfs_root *root)
+{
+        struct btrfs_path *path;
+        struct btrfs_key key;
+        int ret;
+
+        path = btrfs_alloc_path();
+        if (!path)
+                return -ENOMEM;
+
+        key.objectid = root->root_key.objectid;
+        key.type = BTRFS_ROOT_REF_KEY;
+        key.offset = (u64)-1;
+
+        ret = btrfs_search_slot(NULL, root->fs_info->tree_root,
+                                &key, path, 0, 0);
+        if (ret < 0)
+                goto out;
+        BUG_ON(ret == 0);
+
+        ret = 0;
+        if (path->slots[0] > 0) {
+                path->slots[0]--;
+                btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+                if (key.objectid == root->root_key.objectid &&
+                    key.type == BTRFS_ROOT_REF_KEY)
+                        ret = -ENOTEMPTY;
+        }
+out:
+        btrfs_free_path(path);
+        return ret;
+}
+
+static noinline int btrfs_ioctl_snap_destroy(struct file *file,
+                                             void __user *arg)
+{
+        struct dentry *parent = fdentry(file);
+        struct dentry *dentry;
+        struct inode *dir = parent->d_inode;
+        struct inode *inode;
+        struct btrfs_root *root = BTRFS_I(dir)->root;
+        struct btrfs_root *dest = NULL;
+        struct btrfs_ioctl_vol_args *vol_args;
+        struct btrfs_trans_handle *trans;
+        int namelen;
+        int ret;
+        int err = 0;
+
+        if (!capable(CAP_SYS_ADMIN))
+                return -EPERM;
+
+        vol_args = memdup_user(arg, sizeof(*vol_args));
+        if (IS_ERR(vol_args))
+                return PTR_ERR(vol_args);
+
+        vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
+        namelen = strlen(vol_args->name);
+        if (strchr(vol_args->name, '/') ||
+            strncmp(vol_args->name, "..", namelen) == 0) {
+                err = -EINVAL;
+                goto out;
+        }
+
+        err = mnt_want_write(file->f_path.mnt);
+        if (err)
+                goto out;
+
+        mutex_lock_nested(&dir->i_mutex, I_MUTEX_PARENT);
+        dentry = lookup_one_len(vol_args->name, parent, namelen);
+        if (IS_ERR(dentry)) {
+                err = PTR_ERR(dentry);
+                goto out_unlock_dir;
+        }
+
+        if (!dentry->d_inode) {
+                err = -ENOENT;
+                goto out_dput;
+        }
+
+        inode = dentry->d_inode;
+        if (inode->i_ino != BTRFS_FIRST_FREE_OBJECTID) {
+                err = -EINVAL;
+                goto out_dput;
+        }
+
+        dest = BTRFS_I(inode)->root;
+
+        mutex_lock(&inode->i_mutex);
+        err = d_invalidate(dentry);
+        if (err)
+                goto out_unlock;
+
+        down_write(&root->fs_info->subvol_sem);
+
+        err = may_destroy_subvol(dest);
+        if (err)
+                goto out_up_write;
+
+        trans = btrfs_start_transaction(root, 1);
+        ret = btrfs_unlink_subvol(trans, root, dir,
+                                dest->root_key.objectid,
+                                dentry->d_name.name,
+                                dentry->d_name.len);
+        BUG_ON(ret);
+
+        btrfs_record_root_in_trans(trans, dest);
+
+        memset(&dest->root_item.drop_progress, 0,
+                sizeof(dest->root_item.drop_progress));
+        dest->root_item.drop_level = 0;
+        btrfs_set_root_refs(&dest->root_item, 0);
+
+        ret = btrfs_insert_orphan_item(trans,
+                                root->fs_info->tree_root,
+                                dest->root_key.objectid);
+        BUG_ON(ret);
+
+        ret = btrfs_commit_transaction(trans, root);
+        BUG_ON(ret);
+        inode->i_flags |= S_DEAD;
+out_up_write:
+        up_write(&root->fs_info->subvol_sem);
+out_unlock:
+        mutex_unlock(&inode->i_mutex);
+        if (!err) {
+                btrfs_invalidate_inodes(dest);
+                d_delete(dentry);
+        }
+out_dput:
+        dput(dentry);
+out_unlock_dir:
+        mutex_unlock(&dir->i_mutex);
+        mnt_drop_write(file->f_path.mnt);
+out:
+        kfree(vol_args);
+        return err;
+}
+
 static int btrfs_ioctl_defrag(struct file *file)
 {
         struct inode *inode = fdentry(file)->d_inode;
@@ -865,8 +899,8 @@ static long btrfs_ioctl_rm_dev(struct btrfs_root *root, void __user *arg)
         return ret;
 }
 
-static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
-                u64 off, u64 olen, u64 destoff)
+static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
+                                       u64 off, u64 olen, u64 destoff)
 {
         struct inode *inode = fdentry(file)->d_inode;
         struct btrfs_root *root = BTRFS_I(inode)->root;
@@ -976,7 +1010,7 @@ static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 
         /* punch hole in destination first */
         btrfs_drop_extents(trans, root, inode, off, off + len,
-                           off + len, 0, &hint_byte);
+                           off + len, 0, &hint_byte, 1);
 
         /* clone data */
         key.objectid = src->i_ino;
@@ -1071,8 +1105,7 @@ static long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
                                         datao += off - key.offset;
                                         datal -= off - key.offset;
                                 }
-                                if (key.offset + datao + datal + key.offset >
-                                    off + len)
+                                if (key.offset + datao + datal > off + len)
                                         datal = off + len - key.offset - datao;
                                 /* disko == 0 means it's a hole */
                                 if (!disko)
@@ -1258,6 +1291,8 @@ long btrfs_ioctl(struct file *file, unsigned int
                 return btrfs_ioctl_snap_create(file, argp, 0);
         case BTRFS_IOC_SUBVOL_CREATE:
                 return btrfs_ioctl_snap_create(file, argp, 1);
+        case BTRFS_IOC_SNAP_DESTROY:
+                return btrfs_ioctl_snap_destroy(file, argp);
         case BTRFS_IOC_DEFRAG:
                 return btrfs_ioctl_defrag(file);
         case BTRFS_IOC_RESIZE: