ext4: treat buffers with write errors as containing valid data

I got some errors when I repair an ext4 volume which stacked by an
iscsi target:
    Entry 'test60' in / (2) has deleted/unused inode 73750.  Clear?
It can be reproduced when the network not good enough.

When I debug this I found ext4 will read entry buffer from disk and
the buffer is marked with write_io_error.

If the buffer is marked with write_io_error, it means it already
wroten to journal, and not checked out to disk. IOW, the journal
is newer than the data in disk.
If this journal record 'delete test60', it means the 'test60' still
on the disk metadata.

In this case, if we read the buffer from disk successfully and create
file continue, the new journal record will overwrite the journal
which record 'delete test60', then the entry corruptioned.

So, use the buffer rather than read from disk if the buffer is marked
with write_io_error.

Signed-off-by: Zhang Xiaoxu <zhangxiaoxu5@huawei.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

2 files changed, 15 insertions, 2 deletions

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 17cc2dc13174..2348be3d66b7 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -3344,6 +3344,19 @@ static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
 
 extern const struct iomap_ops ext4_iomap_ops;
 
+static inline int ext4_buffer_uptodate(struct buffer_head *bh)
+{
+	/*
+	 * If the buffer has the write error flag, we have failed
+	 * to write out data in the block.  In this  case, we don't
+	 * have to read the block because we may read the old data
+	 * successfully.
+	 */
+	if (!buffer_uptodate(bh) && buffer_write_io_error(bh))
+		set_buffer_uptodate(bh);
+	return buffer_uptodate(bh);
+}
+
 #endif	/* __KERNEL__ */
 
 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 4b92c7603907..9db896fc6af8 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1024,7 +1024,7 @@ struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
 	bh = ext4_getblk(handle, inode, block, map_flags);
 	if (IS_ERR(bh))
 		return bh;
-	if (!bh || buffer_uptodate(bh))
+	if (!bh || ext4_buffer_uptodate(bh))
 		return bh;
 	ll_rw_block(REQ_OP_READ, REQ_META | REQ_PRIO, 1, &bh);
 	wait_on_buffer(bh);
@@ -1051,7 +1051,7 @@ int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
 
 	for (i = 0; i < bh_count; i++)
 		/* Note that NULL bhs[i] is valid because of holes. */
-		if (bhs[i] && !buffer_uptodate(bhs[i]))
+		if (bhs[i] && !ext4_buffer_uptodate(bhs[i]))
 			ll_rw_block(REQ_OP_READ, REQ_META | REQ_PRIO, 1,
 				    &bhs[i]);