ubifs: Fix unattached inode when powercut happens in creating

For selinux or encryption scenarios, UBIFS could become inconsistent
while creating new files in powercut case. Encryption/selinux related
xattrs will be created before creating file dentry, which makes creation
process is not atomic, details are shown as:

Encryption case:
ubifs_create
 ubifs_new_inode
  fscrypt_set_context
   ubifs_xattr_set
    create_xattr
     ubifs_jnl_update  // Disk: xentry xinode inode(LAST_OF_NODE_GROUP)
 >> power cut <<
 ubifs_jnl_update  // Disk: dentry inode parent_inode(LAST_OF_NODE_GROUP)

Selinux case:
ubifs_create
 ubifs_new_inode
 ubifs_init_security
  security_inode_init_security
   ubifs_xattr_set
    create_xattr
     ubifs_jnl_update  // Disk: xentry xinode inode(LAST_OF_NODE_GROUP)
 >> power cut <<
 ubifs_jnl_update  // Disk: dentry inode parent_inode(LAST_OF_NODE_GROUP)

Above process will make chk_fs failed in next mounting:
 UBIFS error (ubi0:0 pid 7995): dbg_check_filesystem [ubifs]: inode 66
 nlink is 1, but calculated nlink is 0

Fix it by allocating orphan inode for each non-xattr file creation, then
removing orphan list in journal writing process, which ensures that both
xattr and dentry be effective in atomic when powercut happens.

Fixes: d7f0b70d30ff ("UBIFS: Add security.* XATTR support for the UBIFS")
Fixes: d475a507457b ("ubifs: Add skeleton for fscrypto")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=218309
Suggested-by: Zhang Yi <yi.zhang@huawei.com>
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Signed-off-by: Richard Weinberger <richard@nod.at>

3 files changed, 51 insertions, 26 deletions

diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index fe16443243ab..c77ea57fe696 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -71,8 +71,13 @@ static int inherit_flags(const struct inode *dir, umode_t mode)
  * @is_xattr: whether the inode is xattr inode
  *
  * This function finds an unused inode number, allocates new inode and
- * initializes it. Returns new inode in case of success and an error code in
- * case of failure.
+ * initializes it. Non-xattr new inode may be written with xattrs(selinux/
+ * encryption) before writing dentry, which could cause inconsistent problem
+ * when powercut happens between two operations. To deal with it, non-xattr
+ * new inode is initialized with zero-nlink and added into orphan list, caller
+ * should make sure that inode is relinked later, and make sure that orphan
+ * removing and journal writing into an committing atomic operation. Returns
+ * new inode in case of success and an error code in case of failure.
  */
 struct inode *ubifs_new_inode(struct ubifs_info *c, struct inode *dir,
 			      umode_t mode, bool is_xattr)
@@ -163,9 +168,25 @@ struct inode *ubifs_new_inode(struct ubifs_info *c, struct inode *dir,
 	ui->creat_sqnum = ++c->max_sqnum;
 	spin_unlock(&c->cnt_lock);
 
+	if (!is_xattr) {
+		set_nlink(inode, 0);
+		err = ubifs_add_orphan(c, inode->i_ino);
+		if (err) {
+			ubifs_err(c, "ubifs_add_orphan failed: %i", err);
+			goto out_iput;
+		}
+		down_read(&c->commit_sem);
+		ui->del_cmtno = c->cmt_no;
+		up_read(&c->commit_sem);
+	}
+
 	if (encrypted) {
 		err = fscrypt_set_context(inode, NULL);
 		if (err) {
+			if (!is_xattr) {
+				set_nlink(inode, 1);
+				ubifs_delete_orphan(c, inode->i_ino);
+			}
 			ubifs_err(c, "fscrypt_set_context failed: %i", err);
 			goto out_iput;
 		}
@@ -320,12 +341,13 @@ static int ubifs_create(struct mnt_idmap *idmap, struct inode *dir,
 	if (err)
 		goto out_inode;
 
+	set_nlink(inode, 1);
 	mutex_lock(&dir_ui->ui_mutex);
 	dir->i_size += sz_change;
 	dir_ui->ui_size = dir->i_size;
 	inode_set_mtime_to_ts(dir,
 			      inode_set_ctime_to_ts(dir, inode_get_ctime(inode)));
-	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 0);
+	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 1);
 	if (err)
 		goto out_cancel;
 	mutex_unlock(&dir_ui->ui_mutex);
@@ -340,8 +362,8 @@ out_cancel:
 	dir->i_size -= sz_change;
 	dir_ui->ui_size = dir->i_size;
 	mutex_unlock(&dir_ui->ui_mutex);
+	set_nlink(inode, 0);
 out_inode:
-	make_bad_inode(inode);
 	iput(inode);
 out_fname:
 	fscrypt_free_filename(&nm);
@@ -386,7 +408,6 @@ static struct inode *create_whiteout(struct inode *dir, struct dentry *dentry)
 	return inode;
 
 out_inode:
-	make_bad_inode(inode);
 	iput(inode);
 out_free:
 	ubifs_err(c, "cannot create whiteout file, error %d", err);
@@ -470,6 +491,7 @@ static int ubifs_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
 	if (err)
 		goto out_inode;
 
+	set_nlink(inode, 1);
 	mutex_lock(&ui->ui_mutex);
 	insert_inode_hash(inode);
 	d_tmpfile(file, inode);
@@ -479,7 +501,7 @@ static int ubifs_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
 	mutex_unlock(&ui->ui_mutex);
 
 	lock_2_inodes(dir, inode);
-	err = ubifs_jnl_update(c, dir, &nm, inode, 1, 0, 0);
+	err = ubifs_jnl_update(c, dir, &nm, inode, 1, 0, 1);
 	if (err)
 		goto out_cancel;
 	unlock_2_inodes(dir, inode);
@@ -492,7 +514,6 @@ static int ubifs_tmpfile(struct mnt_idmap *idmap, struct inode *dir,
 out_cancel:
 	unlock_2_inodes(dir, inode);
 out_inode:
-	make_bad_inode(inode);
 	if (!instantiated)
 		iput(inode);
 out_budg:
@@ -1011,6 +1032,7 @@ static int ubifs_mkdir(struct mnt_idmap *idmap, struct inode *dir,
 	if (err)
 		goto out_inode;
 
+	set_nlink(inode, 1);
 	mutex_lock(&dir_ui->ui_mutex);
 	insert_inode_hash(inode);
 	inc_nlink(inode);
@@ -1019,7 +1041,7 @@ static int ubifs_mkdir(struct mnt_idmap *idmap, struct inode *dir,
 	dir_ui->ui_size = dir->i_size;
 	inode_set_mtime_to_ts(dir,
 			      inode_set_ctime_to_ts(dir, inode_get_ctime(inode)));
-	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 0);
+	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 1);
 	if (err) {
 		ubifs_err(c, "cannot create directory, error %d", err);
 		goto out_cancel;
@@ -1036,8 +1058,8 @@ out_cancel:
 	dir_ui->ui_size = dir->i_size;
 	drop_nlink(dir);
 	mutex_unlock(&dir_ui->ui_mutex);
+	set_nlink(inode, 0);
 out_inode:
-	make_bad_inode(inode);
 	iput(inode);
 out_fname:
 	fscrypt_free_filename(&nm);
@@ -1107,13 +1129,14 @@ static int ubifs_mknod(struct mnt_idmap *idmap, struct inode *dir,
 	ui = ubifs_inode(inode);
 	ui->data = dev;
 	ui->data_len = devlen;
+	set_nlink(inode, 1);
 
 	mutex_lock(&dir_ui->ui_mutex);
 	dir->i_size += sz_change;
 	dir_ui->ui_size = dir->i_size;
 	inode_set_mtime_to_ts(dir,
 			      inode_set_ctime_to_ts(dir, inode_get_ctime(inode)));
-	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 0);
+	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 1);
 	if (err)
 		goto out_cancel;
 	mutex_unlock(&dir_ui->ui_mutex);
@@ -1128,8 +1151,8 @@ out_cancel:
 	dir->i_size -= sz_change;
 	dir_ui->ui_size = dir->i_size;
 	mutex_unlock(&dir_ui->ui_mutex);
+	set_nlink(inode, 0);
 out_inode:
-	make_bad_inode(inode);
 	iput(inode);
 out_fname:
 	fscrypt_free_filename(&nm);
@@ -1208,13 +1231,14 @@ static int ubifs_symlink(struct mnt_idmap *idmap, struct inode *dir,
 	 */
 	ui->data_len = disk_link.len - 1;
 	inode->i_size = ubifs_inode(inode)->ui_size = disk_link.len - 1;
+	set_nlink(inode, 1);
 
 	mutex_lock(&dir_ui->ui_mutex);
 	dir->i_size += sz_change;
 	dir_ui->ui_size = dir->i_size;
 	inode_set_mtime_to_ts(dir,
 			      inode_set_ctime_to_ts(dir, inode_get_ctime(inode)));
-	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 0);
+	err = ubifs_jnl_update(c, dir, &nm, inode, 0, 0, 1);
 	if (err)
 		goto out_cancel;
 	mutex_unlock(&dir_ui->ui_mutex);
@@ -1228,10 +1252,10 @@ out_cancel:
 	dir->i_size -= sz_change;
 	dir_ui->ui_size = dir->i_size;
 	mutex_unlock(&dir_ui->ui_mutex);
+	set_nlink(inode, 0);
 out_inode:
 	/* Free inode->i_link before inode is marked as bad. */
 	fscrypt_free_inode(inode);
-	make_bad_inode(inode);
 	iput(inode);
 out_fname:
 	fscrypt_free_filename(&nm);
@@ -1399,14 +1423,10 @@ static int do_rename(struct inode *old_dir, struct dentry *old_dentry,
 		 */
 		err = ubifs_budget_space(c, &wht_req);
 		if (err) {
-			/*
-			 * Whiteout inode can not be written on flash by
-			 * ubifs_jnl_write_inode(), because it's neither
-			 * dirty nor zero-nlink.
-			 */
 			iput(whiteout);
 			goto out_release;
 		}
+		set_nlink(whiteout, 1);
 
 		/* Add the old_dentry size to the old_dir size. */
 		old_sz -= CALC_DENT_SIZE(fname_len(&old_nm));
@@ -1485,7 +1505,7 @@ static int do_rename(struct inode *old_dir, struct dentry *old_dentry,
 	}
 
 	err = ubifs_jnl_rename(c, old_dir, old_inode, &old_nm, new_dir,
-			       new_inode, &new_nm, whiteout, sync);
+			       new_inode, &new_nm, whiteout, sync, !!whiteout);
 	if (err)
 		goto out_cancel;
 
@@ -1538,6 +1558,7 @@ out_cancel:
 	unlock_4_inodes(old_dir, new_dir, new_inode, whiteout);
 	if (whiteout) {
 		ubifs_release_budget(c, &wht_req);
+		set_nlink(whiteout, 0);
 		iput(whiteout);
 	}
 out_release:
diff --git a/fs/ubifs/journal.c b/fs/ubifs/journal.c
index 2b4b05c2d9b2..517f5de76b49 100644
--- a/fs/ubifs/journal.c
+++ b/fs/ubifs/journal.c
@@ -643,7 +643,7 @@ static void set_dent_cookie(struct ubifs_info *c, struct ubifs_dent_node *dent)
  * @inode: inode to update
  * @deletion: indicates a directory entry deletion i.e unlink or rmdir
  * @xent: non-zero if the directory entry is an extended attribute entry
- * @delete_orphan: indicates an orphan entry deletion for @inode
+ * @in_orphan: indicates whether the @inode is in orphan list
  *
  * This function updates an inode by writing a directory entry (or extended
  * attribute entry), the inode itself, and the parent directory inode (or the
@@ -665,7 +665,7 @@ static void set_dent_cookie(struct ubifs_info *c, struct ubifs_dent_node *dent)
  */
 int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 		     const struct fscrypt_name *nm, const struct inode *inode,
-		     int deletion, int xent, int delete_orphan)
+		     int deletion, int xent, int in_orphan)
 {
 	int err, dlen, ilen, len, lnum, ino_offs, dent_offs, orphan_added = 0;
 	int aligned_dlen, aligned_ilen, sync = IS_DIRSYNC(dir);
@@ -751,7 +751,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 	if (err)
 		goto out_release;
 
-	if (last_reference) {
+	if (last_reference && !in_orphan) {
 		err = ubifs_add_orphan(c, inode->i_ino);
 		if (err) {
 			release_head(c, BASEHD);
@@ -807,7 +807,7 @@ int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 	if (err)
 		goto out_ro;
 
-	if (delete_orphan)
+	if (in_orphan && inode->i_nlink)
 		ubifs_delete_orphan(c, inode->i_ino);
 
 	finish_reservation(c);
@@ -1340,6 +1340,7 @@ out_free:
  * @new_nm: new name of the new directory entry
  * @whiteout: whiteout inode
  * @sync: non-zero if the write-buffer has to be synchronized
+ * @delete_orphan: indicates an orphan entry deletion for @whiteout
  *
  * This function implements the re-name operation which may involve writing up
  * to 4 inodes(new inode, whiteout inode, old and new parent directory inodes)
@@ -1352,7 +1353,7 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 		     const struct inode *new_dir,
 		     const struct inode *new_inode,
 		     const struct fscrypt_name *new_nm,
-		     const struct inode *whiteout, int sync)
+		     const struct inode *whiteout, int sync, int delete_orphan)
 {
 	void *p;
 	union ubifs_key key;
@@ -1569,6 +1570,9 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 			goto out_ro;
 	}
 
+	if (delete_orphan)
+		ubifs_delete_orphan(c, whiteout->i_ino);
+
 	finish_reservation(c);
 	if (new_inode) {
 		mark_inode_clean(c, new_ui);
diff --git a/fs/ubifs/ubifs.h b/fs/ubifs/ubifs.h
index d0ed78345617..b2c6554fa857 100644
--- a/fs/ubifs/ubifs.h
+++ b/fs/ubifs/ubifs.h
@@ -1800,7 +1800,7 @@ int ubifs_consolidate_log(struct ubifs_info *c);
 /* journal.c */
 int ubifs_jnl_update(struct ubifs_info *c, const struct inode *dir,
 		     const struct fscrypt_name *nm, const struct inode *inode,
-		     int deletion, int xent, int delete_orphan);
+		     int deletion, int xent, int in_orphan);
 int ubifs_jnl_write_data(struct ubifs_info *c, const struct inode *inode,
 			 const union ubifs_key *key, const void *buf, int len);
 int ubifs_jnl_write_inode(struct ubifs_info *c, const struct inode *inode);
@@ -1817,7 +1817,7 @@ int ubifs_jnl_rename(struct ubifs_info *c, const struct inode *old_dir,
 		     const struct inode *new_dir,
 		     const struct inode *new_inode,
 		     const struct fscrypt_name *new_nm,
-		     const struct inode *whiteout, int sync);
+		     const struct inode *whiteout, int sync, int delete_orphan);
 int ubifs_jnl_truncate(struct ubifs_info *c, const struct inode *inode,
 		       loff_t old_size, loff_t new_size);
 int ubifs_jnl_delete_xattr(struct ubifs_info *c, const struct inode *host,