Patchwork ext4: fix ext4_end_io_dio() racing against fsync()

login
register
mail settings
Submitter Theodore Ts'o
Date Nov. 30, 2011, 8:53 p.m.
Message ID <1322686428-17728-1-git-send-email-tytso@mit.edu>
Download mbox | patch
Permalink /patch/128570/
State Accepted
Headers show

Comments

Theodore Ts'o - Nov. 30, 2011, 8:53 p.m.
We need to make sure iocb->private is cleared *before* we put the
io_end structure on i_completed_io_list.  Otherwise fsync() could
potentially run on another CPU and free the iocb structure out from
under us.

Reported-by: Kent Overstreet <koverstreet@google.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
---
 fs/ext4/inode.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

Patch

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 848f436..d2c8663 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -2781,10 +2781,11 @@  static void ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
  		  iocb->private, io_end->inode->i_ino, iocb, offset,
 		  size);
 
+	iocb->private = NULL;
+
 	/* if not aio dio with unwritten extents, just free io and return */
 	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
 		ext4_free_io_end(io_end);
-		iocb->private = NULL;
 out:
 		if (is_async)
 			aio_complete(iocb, ret, 0);
@@ -2807,7 +2808,6 @@  out:
 	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
 
 	/* queue the work to convert unwritten extents to written */
-	iocb->private = NULL;
 	queue_work(wq, &io_end->work);
 
 	/* XXX: probably should move into the real I/O completion handler */