locks: close potential race between setlease and open
As Al Viro points out, there is an unlikely, but possible race between
opening a file and setting a lease on it. generic_add_lease is done with
the i_lock held, but the inode->i_flock check in break_lease is
lockless. It's possible for another task doing an open to do the entire
pathwalk and call break_lease between the point where generic_add_lease
checks for a conflicting open and adds the lease to the list. If this
occurs, we can end up with a lease set on the file with a conflicting
open.
To guard against that, check again for a conflicting open after adding
the lease to the i_flock list. If the above race occurs, then we can
simply unwind the lease setting and return -EAGAIN.
Because we take dentry references and acquire write access on the file
before calling break_lease, we know that if the i_flock list is empty
when the open caller goes to check it then the necessary refcounts have
already been incremented. Thus the additional check for a conflicting
open will see that there is one and the setlease call will fail.
Cc: Bruce Fields <bfields@fieldses.org>
Cc: David Howells <dhowells@redhat.com>
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@fieldses.org>
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 09f553c..df84744 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -1964,6 +1964,12 @@
static inline int break_lease(struct inode *inode, unsigned int mode)
{
+ /*
+ * Since this check is lockless, we must ensure that any refcounts
+ * taken are done before checking inode->i_flock. Otherwise, we could
+ * end up racing with tasks trying to set a new lease on this file.
+ */
+ smp_mb();
if (inode->i_flock)
return __break_lease(inode, mode, FL_LEASE);
return 0;