[PATCH resend] posix_acl: de-union a_refcount and a_rcu

From: Jeff Layton <jlayton@redhat.com>
To: Al Viro <viro@zeniv.linux.org.uk>,
	Andreas Gruenbacher <agruenba@redhat.com>
Cc: linux-fsdevel@vger.kernel.org
Subject: [PATCH resend] posix_acl: de-union a_refcount and a_rcu
Date: Mon, 11 Jul 2016 09:10:06 -0400	[thread overview]
Message-ID: <1468242606-10826-1-git-send-email-jlayton@redhat.com> (raw)

Currently the two are unioned together, but I don't think that's safe.

It looks like get_cached_acl could race with the last put in
posix_acl_release. get_cached_acl calls atomic_inc_not_zero on
a_refcount, but that field could have already been clobbered by
call_rcu, and may no longer be zero. Fix this by de-unioning the two
fields.

Fixes: b8a7a3a66747 (posix_acl: Inode acl caching fixes)
Signed-off-by: Jeff Layton <jlayton@redhat.com>
---
 include/linux/posix_acl.h | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

This appears to be a regression from v4.6, so I think we do want this in
v4.7.

diff --git a/include/linux/posix_acl.h b/include/linux/posix_acl.h
index 5b5a80cc5926..c818772d9f9d 100644
--- a/include/linux/posix_acl.h
+++ b/include/linux/posix_acl.h
@@ -43,10 +43,8 @@ struct posix_acl_entry {
 };
 
 struct posix_acl {
-	union {
-		atomic_t		a_refcount;
-		struct rcu_head		a_rcu;
-	};
+	atomic_t		a_refcount;
+	struct rcu_head		a_rcu;
 	unsigned int		a_count;
 	struct posix_acl_entry	a_entries[0];
 };
-- 
2.7.4

