ext4: fix racy use-after-free in ext4_end_io_dio()

ext4_end_io_dio() queues io_end->work and then clears iocb->private;
however, io_end->work calls aio_complete() which frees the iocb
object.  If that slab object gets reallocated, then ext4_end_io_dio()
can end up clearing someone else's iocb->private, this use-after-free
can cause a leak of a struct ext4_io_end_t structure.

Detected and tested with slab poisoning.

[ Note: Can also reproduce using 12 fio's against 12 file systems with the
  following configuration file:

  [global]
  direct=1
  ioengine=libaio
  iodepth=1
  bs=4k
  ba=4k
  size=128m

  [create]
  filename=${TESTDIR}
  rw=write

  -- tytso ]

Google-Bug-Id: 5354697
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Reported-by: Kent Overstreet <koverstreet@google.com>
Tested-by: Kent Overstreet <koverstreet@google.com>
Cc: stable@kernel.org

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index fffec40..848f436 100644 (file)
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -2807,8 +2807,8 @@ out:
        spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
 
        /* queue the work to convert unwritten extents to written */
        spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
 
        /* queue the work to convert unwritten extents to written */

-       queue_work(wq, &io_end->work);

        iocb->private = NULL;
        iocb->private = NULL;

+       queue_work(wq, &io_end->work);

 
        /* XXX: probably should move into the real I/O completion handler */
        inode_dio_done(inode);
 
        /* XXX: probably should move into the real I/O completion handler */
        inode_dio_done(inode);