From: Richard Weinberger <richard@xxxxxx>
commit e58725d51fa8da9133f3f1c54170aa2e43056b91 upstream.
UBIFS's recovery code strictly assumes that a deleted inode will never
come back, therefore it removes all data which belongs to that inode
as soon it faces an inode with link count 0 in the replay list.
Before O_TMPFILE this assumption was perfectly fine. With O_TMPFILE
it can lead to data loss upon a power-cut.
Consider a journal with entries like:
0: inode X (nlink = 0) /* O_TMPFILE was created */
1: data for inode X /* Someone writes to the temp file */
2: inode X (nlink = 0) /* inode was changed, xattr, chmod, … */
3: inode X (nlink = 1) /* inode was re-linked via linkat() */
Upon replay of entry #2 UBIFS will drop all data that belongs to inode X,
this will lead to an empty file after mounting.
As solution for this problem, scan the replay list for a re-link entry
before dropping data.
Fixes: 474b93704f32 ("ubifs: Implement O_TMPFILE")
Cc: stable@xxxxxxxxxxxxxxx # 4.9-4.18
Cc: Russell Senior <russell@xxxxxxxxxxxxxxxxx>
Cc: Rafał Miłecki <zajec5@xxxxxxxxx>
Reported-by: Russell Senior <russell@xxxxxxxxxxxxxxxxx>
Reported-by: Rafał Miłecki <zajec5@xxxxxxxxx>
Tested-by: Rafał Miłecki <rafal@xxxxxxxxxx>
Signed-off-by: Richard Weinberger <richard@xxxxxx>
[rmilecki: update ubifs_assert() calls to compile with 4.18 and older]
Signed-off-by: Rafał Miłecki <rafal@xxxxxxxxxx>
(cherry picked from commit e58725d51fa8da9133f3f1c54170aa2e43056b91)
---
fs/ubifs/replay.c | 37 +++++++++++++++++++++++++++++++++++++
1 file changed, 37 insertions(+)
diff --git a/fs/ubifs/replay.c b/fs/ubifs/replay.c
index ae5c02f22f3e..d998fbf7de30 100644
--- a/fs/ubifs/replay.c
+++ b/fs/ubifs/replay.c
@@ -210,6 +210,38 @@ static int trun_remove_range(struct ubifs_info *c, struct replay_entry *r)
}
/**
+ * inode_still_linked - check whether inode in question will be re-linked.
+ * @c: UBIFS file-system description object
+ * @rino: replay entry to test
+ *
+ * O_TMPFILE files can be re-linked, this means link count goes from 0 to 1.
+ * This case needs special care, otherwise all references to the inode will
+ * be removed upon the first replay entry of an inode with link count 0
+ * is found.
+ */
+static bool inode_still_linked(struct ubifs_info *c, struct replay_entry *rino)
+{
+ struct replay_entry *r;
+
+ ubifs_assert(rino->deletion);
+ ubifs_assert(key_type(c, &rino->key) == UBIFS_INO_KEY);
+
+ /*
+ * Find the most recent entry for the inode behind @rino and check
+ * whether it is a deletion.
+ */
+ list_for_each_entry_reverse(r, &c->replay_list, list) {
+ ubifs_assert(r->sqnum >= rino->sqnum);
+ if (key_inum(c, &r->key) == key_inum(c, &rino->key))
+ return r->deletion == 0;
+
+ }
+
+ ubifs_assert(0);
+ return false;
+}
+
+/**
* apply_replay_entry - apply a replay entry to the TNC.
* @c: UBIFS file-system description object
* @r: replay entry to apply
@@ -239,6 +271,11 @@ static int apply_replay_entry(struct ubifs_info *c, struct replay_entry *r)
{
ino_t inum = key_inum(c, &r->key);
+ if (inode_still_linked(c, r)) {
+ err = 0;
+ break;
+ }
+
err = ubifs_tnc_remove_ino(c, inum);
break;
}
--
2.13.7
