[PATCH] f2fs: fix to convert inline directory correctly

[Chao Yu]

With below serials, we will lose parts of dirents:

1) mount f2fs with inline_dentry option
2) echo 1 > /sys/fs/f2fs/sdX/dir_level
3) mkdir dir
4) touch 180 files named [001-180] in dir
5) touch 181 in dir
6) echo 3 > /proc/sys/vm/drop_caches
7) ll dir

ls: cannot access 2: No such file or directory
ls: cannot access 4: No such file or directory
ls: cannot access 5: No such file or directory
ls: cannot access 6: No such file or directory
ls: cannot access 8: No such file or directory
ls: cannot access 9: No such file or directory
...
total 360
drwxr-xr-x 2 root root 4096 Feb 19 15:12 ./
drwxr-xr-x 3 root root 4096 Feb 19 15:11 ../
-rw-r--r-- 1 root root    0 Feb 19 15:12 1
-rw-r--r-- 1 root root    0 Feb 19 15:12 10
-rw-r--r-- 1 root root    0 Feb 19 15:12 100
-????????? ? ?    ?       ?            ? 101
-????????? ? ?    ?       ?            ? 102
-????????? ? ?    ?       ?            ? 103
...

The reason is: when doing the inline dir conversion, we didn't consider
that directory has hierarchical hash structure which can be configured
through sysfs interface 'dir_level'.

By default, dir_level of directory inode is 0, it means we have one bucket
in hash table located in first level, all dirents will be hashed in this
bucket, so it has no problem for us to do the duplication simply between
inline dentry page and converted normal dentry page.

However, if we configured dir_level with the value N (greater than 0), it
will expand the bucket number of first level hash table by 2^N - 1, it
hashs dirents into different buckets according their hash value, if we
still move all dirents to first bucket, it makes incorrent locating for
inline dirents, the result is, although we can iterate all dirents through
->readdir, we can't stat some of them in ->lookup which based on hash
table searching.

This patch fixes this issue by rehashing dirents into correct position
when converting inline directory.

Signed-off-by: Chao Yu <chao2.yu@xxxxxxxxxxx>
---
 fs/f2fs/data.c          |  2 --
 fs/f2fs/dir.c           | 87 ++++++++++++++++++++++++---------------------
 fs/f2fs/f2fs.h          |  4 ++-
 fs/f2fs/inline.c        | 94 ++++++++++++++++++++++++++++++++++++++++++++++++-
 include/linux/f2fs_fs.h |  2 ++
 5 files changed, 144 insertions(+), 45 deletions(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index d3a6cd0..6baabf5 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -505,8 +505,6 @@ struct page *get_new_data_page(struct inode *inode,
 	} else {
 		f2fs_put_page(page, 1);
 
-		/* if ipage exists, blkaddr should be NEW_ADDR */
-		f2fs_bug_on(F2FS_I_SB(inode), ipage);
 		page = get_lock_data_page(inode, index, true);
 		if (IS_ERR(page))
 			return page;
diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
index aa06700..0fe877d 100644
--- a/fs/f2fs/dir.c
+++ b/fs/f2fs/dir.c
@@ -48,7 +48,6 @@ unsigned char f2fs_filetype_table[F2FS_FT_MAX] = {
 	[F2FS_FT_SYMLINK]	= DT_LNK,
 };
 
-#define S_SHIFT 12
 static unsigned char f2fs_type_by_mode[S_IFMT >> S_SHIFT] = {
 	[S_IFREG >> S_SHIFT]	= F2FS_FT_REG_FILE,
 	[S_IFDIR >> S_SHIFT]	= F2FS_FT_DIR,
@@ -64,6 +63,13 @@ void set_de_type(struct f2fs_dir_entry *de, umode_t mode)
 	de->file_type = f2fs_type_by_mode[(mode & S_IFMT) >> S_SHIFT];
 }
 
+unsigned char get_de_type(struct f2fs_dir_entry *de)
+{
+	if (de->file_type < F2FS_FT_MAX)
+		return f2fs_filetype_table[de->file_type];
+	return DT_UNKNOWN;
+}
+
 static unsigned long dir_block_index(unsigned int level,
 				int dir_level, unsigned int idx)
 {
@@ -519,11 +525,7 @@ void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d,
 	}
 }
 
-/*
- * Caller should grab and release a rwsem by calling f2fs_lock_op() and
- * f2fs_unlock_op().
- */
-int __f2fs_add_link(struct inode *dir, const struct qstr *name,
+int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name,
 				struct inode *inode, nid_t ino, umode_t mode)
 {
 	unsigned int bit_pos;
@@ -536,28 +538,11 @@ int __f2fs_add_link(struct inode *dir, const struct qstr *name,
 	struct f2fs_dentry_block *dentry_blk = NULL;
 	struct f2fs_dentry_ptr d;
 	struct page *page = NULL;
-	struct fscrypt_name fname;
-	struct qstr new_name;
-	int slots, err;
-
-	err = fscrypt_setup_filename(dir, name, 0, &fname);
-	if (err)
-		return err;
-
-	new_name.name = fname_name(&fname);
-	new_name.len = fname_len(&fname);
-
-	if (f2fs_has_inline_dentry(dir)) {
-		err = f2fs_add_inline_entry(dir, &new_name, inode, ino, mode);
-		if (!err || err != -EAGAIN)
-			goto out;
-		else
-			err = 0;
-	}
+	int slots, err = 0;
 
 	level = 0;
-	slots = GET_DENTRY_SLOTS(new_name.len);
-	dentry_hash = f2fs_dentry_hash(&new_name);
+	slots = GET_DENTRY_SLOTS(new_name->len);
+	dentry_hash = f2fs_dentry_hash(new_name);
 
 	current_depth = F2FS_I(dir)->i_current_depth;
 	if (F2FS_I(dir)->chash == dentry_hash) {
@@ -566,10 +551,8 @@ int __f2fs_add_link(struct inode *dir, const struct qstr *name,
 	}
 
 start:
-	if (unlikely(current_depth == MAX_DIR_HASH_DEPTH)) {
-		err = -ENOSPC;
-		goto out;
-	}
+	if (unlikely(current_depth == MAX_DIR_HASH_DEPTH))
+		return -ENOSPC;
 
 	/* Increase the depth, if required */
 	if (level == current_depth)
@@ -583,10 +566,8 @@ start:
 
 	for (block = bidx; block <= (bidx + nblock - 1); block++) {
 		dentry_page = get_new_data_page(dir, NULL, block, true);
-		if (IS_ERR(dentry_page)) {
-			err = PTR_ERR(dentry_page);
-			goto out;
-		}
+		if (IS_ERR(dentry_page))
+			return PTR_ERR(dentry_page);
 
 		dentry_blk = kmap(dentry_page);
 		bit_pos = room_for_filename(&dentry_blk->dentry_bitmap,
@@ -606,7 +587,7 @@ add_dentry:
 
 	if (inode) {
 		down_write(&F2FS_I(inode)->i_sem);
-		page = init_inode_metadata(inode, dir, &new_name, NULL);
+		page = init_inode_metadata(inode, dir, new_name, NULL);
 		if (IS_ERR(page)) {
 			err = PTR_ERR(page);
 			goto fail;
@@ -616,7 +597,7 @@ add_dentry:
 	}
 
 	make_dentry_ptr(NULL, &d, (void *)dentry_blk, 1);
-	f2fs_update_dentry(ino, mode, &d, &new_name, dentry_hash, bit_pos);
+	f2fs_update_dentry(ino, mode, &d, new_name, dentry_hash, bit_pos);
 
 	set_page_dirty(dentry_page);
 
@@ -638,7 +619,34 @@ fail:
 	}
 	kunmap(dentry_page);
 	f2fs_put_page(dentry_page, 1);
-out:
+
+	return err;
+}
+
+/*
+ * Caller should grab and release a rwsem by calling f2fs_lock_op() and
+ * f2fs_unlock_op().
+ */
+int __f2fs_add_link(struct inode *dir, const struct qstr *name,
+				struct inode *inode, nid_t ino, umode_t mode)
+{
+	struct fscrypt_name fname;
+	struct qstr new_name;
+	int err;
+
+	err = fscrypt_setup_filename(dir, name, 0, &fname);
+	if (err)
+		return err;
+
+	new_name.name = fname_name(&fname);
+	new_name.len = fname_len(&fname);
+
+	err = -EAGAIN;
+	if (f2fs_has_inline_dentry(dir))
+		err = f2fs_add_inline_entry(dir, &new_name, inode, ino, mode);
+	if (err == -EAGAIN)
+		err = f2fs_add_regular_entry(dir, &new_name, inode, ino, mode);
+
 	fscrypt_free_filename(&fname);
 	f2fs_update_time(F2FS_I_SB(dir), REQ_TIME);
 	return err;
@@ -802,10 +810,7 @@ bool f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d,
 			continue;
 		}
 
-		if (de->file_type < F2FS_FT_MAX)
-			d_type = f2fs_filetype_table[de->file_type];
-		else
-			d_type = DT_UNKNOWN;
+		d_type = get_de_type(de);
 
 		de_name.name = d->filename[bit_pos];
 		de_name.len = le16_to_cpu(de->name_len);
diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 4d759f3..8fc17c0 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1706,7 +1706,7 @@ struct dentry *f2fs_get_parent(struct dentry *child);
  */
 extern unsigned char f2fs_filetype_table[F2FS_FT_MAX];
 void set_de_type(struct f2fs_dir_entry *, umode_t);
-
+unsigned char get_de_type(struct f2fs_dir_entry *);
 struct f2fs_dir_entry *find_target_dentry(struct fscrypt_name *,
 			f2fs_hash_t, int *, struct f2fs_dentry_ptr *);
 bool f2fs_fill_dentries(struct dir_context *, struct f2fs_dentry_ptr *,
@@ -1727,6 +1727,8 @@ void f2fs_set_link(struct inode *, struct f2fs_dir_entry *,
 int update_dent_inode(struct inode *, struct inode *, const struct qstr *);
 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *,
 			const struct qstr *, f2fs_hash_t , unsigned int);
+int f2fs_add_regular_entry(struct inode *, const struct qstr *,
+						struct inode *, nid_t, umode_t);
 int __f2fs_add_link(struct inode *, const struct qstr *, struct inode *, nid_t,
 			umode_t);
 void f2fs_delete_entry(struct f2fs_dir_entry *, struct page *, struct inode *,
diff --git a/fs/f2fs/inline.c b/fs/f2fs/inline.c
index d27347e..f651083 100644
--- a/fs/f2fs/inline.c
+++ b/fs/f2fs/inline.c
@@ -371,7 +371,7 @@ int make_empty_inline_dir(struct inode *inode, struct inode *parent,
  * NOTE: ipage is grabbed by caller, but if any error occurs, we should
  * release ipage in this function.
  */
-static int f2fs_convert_inline_dir(struct inode *dir, struct page *ipage,
+static int f2fs_move_inline_dirents(struct inode *dir, struct page *ipage,
 				struct f2fs_inline_dentry *inline_dentry)
 {
 	struct page *page;
@@ -432,6 +432,98 @@ out:
 	return err;
 }
 
+static int f2fs_add_inline_entries(struct inode *dir,
+			struct f2fs_inline_dentry *inline_dentry)
+{
+	struct f2fs_dentry_ptr d;
+	unsigned long bit_pos = 0;
+	int err = 0;
+
+	make_dentry_ptr(NULL, &d, (void *)inline_dentry, 2);
+
+	while (bit_pos < d.max) {
+		struct f2fs_dir_entry *de;
+		struct qstr new_name;
+		nid_t ino;
+		umode_t fake_mode;
+
+		if (!test_bit_le(bit_pos, d.bitmap)) {
+			bit_pos++;
+			continue;
+		}
+
+		de = &d.dentry[bit_pos];
+		new_name.name = d.filename[bit_pos];
+		new_name.len = de->name_len;
+
+		ino = le32_to_cpu(de->ino);
+		fake_mode = get_de_type(de) << S_SHIFT;
+
+		err = f2fs_add_regular_entry(dir, &new_name, NULL,
+							ino, fake_mode);
+		if (err)
+			goto punch_dentry_pages;
+
+		if (unlikely(!de->name_len))
+			d.max = -1;
+
+		bit_pos += GET_DENTRY_SLOTS(le16_to_cpu(de->name_len));
+	}
+	return 0;
+punch_dentry_pages:
+	truncate_inode_pages(&dir->i_data, 0);
+	truncate_blocks(dir, 0, false);
+	remove_dirty_inode(dir);
+	return err;
+}
+
+static int f2fs_move_rehashed_dirents(struct inode *dir, struct page *ipage,
+				struct f2fs_inline_dentry *inline_dentry)
+{
+	struct f2fs_inline_dentry *backup_dentry;
+	int err;
+
+	backup_dentry = kmalloc(sizeof(struct f2fs_inline_dentry),
+							GFP_F2FS_ZERO);
+	if (!backup_dentry)
+		return -ENOMEM;
+
+	memcpy(backup_dentry, inline_dentry, MAX_INLINE_DATA);
+	truncate_inline_inode(ipage, 0);
+
+	unlock_page(ipage);
+
+	err = f2fs_add_inline_entries(dir, backup_dentry);
+	if (err)
+		goto recover;
+
+	lock_page(ipage);
+
+	stat_dec_inline_dir(dir);
+	clear_inode_flag(F2FS_I(dir), FI_INLINE_DENTRY);
+	update_inode(dir, ipage);
+	kfree(backup_dentry);
+	return 0;
+recover:
+	lock_page(ipage);
+	memcpy(inline_dentry, backup_dentry, MAX_INLINE_DATA);
+	i_size_write(dir, MAX_INLINE_DATA);
+	update_inode(dir, ipage);
+	f2fs_put_page(ipage, 1);
+
+	kfree(backup_dentry);
+	return err;
+}
+
+static int f2fs_convert_inline_dir(struct inode *dir, struct page *ipage,
+				struct f2fs_inline_dentry *inline_dentry)
+{
+	if (!F2FS_I(dir)->i_dir_level)
+		return f2fs_move_inline_dirents(dir, ipage, inline_dentry);
+	else
+		return f2fs_move_rehashed_dirents(dir, ipage, inline_dentry);
+}
+
 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *name,
 			struct inode *inode, nid_t ino, umode_t mode)
 {
diff --git a/include/linux/f2fs_fs.h b/include/linux/f2fs_fs.h
index 9eb215a..71a9ba8 100644
--- a/include/linux/f2fs_fs.h
+++ b/include/linux/f2fs_fs.h
@@ -508,4 +508,6 @@ enum {
 	F2FS_FT_MAX
 };
 
+#define S_SHIFT 12
+
 #endif  /* _LINUX_F2FS_FS_H */
-- 
2.7.0