ext4: Use correct locking for ext4_end_io_nolock()

We must hold i_completed_io_lock when manipulating anything on the
i_completed_io_list linked list.  This includes io->lock, which we
were checking in ext4_end_io_nolock().

So move this check to ext4_end_io_work().  This also has the bonus of
avoiding extra work if it is already done without needing to take the
mutex.

Signed-off-by: Tao Ma <boyu.mt@taobao.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
diff --git a/fs/ext4/fsync.c b/fs/ext4/fsync.c
index c942924..851ac5b 100644
--- a/fs/ext4/fsync.c
+++ b/fs/ext4/fsync.c
@@ -83,9 +83,6 @@
 	int ret = 0;
 	int ret2 = 0;
 
-	if (list_empty(&ei->i_completed_io_list))
-		return ret;
-
 	dump_completed_IO(inode);
 	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
 	while (!list_empty(&ei->i_completed_io_list)){
diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c
index 92f38ee..aed4096 100644
--- a/fs/ext4/page-io.c
+++ b/fs/ext4/page-io.c
@@ -87,6 +87,9 @@
 
 /*
  * check a range of space and convert unwritten extents to written.
+ *
+ * Called with inode->i_mutex; we depend on this when we manipulate
+ * io->flag, since we could otherwise race with ext4_flush_completed_IO()
  */
 int ext4_end_io_nolock(ext4_io_end_t *io)
 {
@@ -100,9 +103,6 @@
 		   "list->prev 0x%p\n",
 		   io, inode->i_ino, io->list.next, io->list.prev);
 
-	if (list_empty(&io->list))
-		return ret;
-
 	if (!(io->flag & EXT4_IO_END_UNWRITTEN))
 		return ret;
 
@@ -142,6 +142,13 @@
 	unsigned long		flags;
 	int			ret;
 
+	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
+	if (list_empty(&io->list)) {
+		spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
+		goto free;
+	}
+	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
+
 	if (!mutex_trylock(&inode->i_mutex)) {
 		/*
 		 * Requeue the work instead of waiting so that the work
@@ -170,6 +177,7 @@
 		list_del_init(&io->list);
 	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
 	mutex_unlock(&inode->i_mutex);
+free:
 	ext4_free_io_end(io);
 }