From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752628AbcDNXs4 (ORCPT ); Thu, 14 Apr 2016 19:48:56 -0400 Received: from mail-pa0-f50.google.com ([209.85.220.50]:33369 "EHLO mail-pa0-f50.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752290AbcDNXsy (ORCPT ); Thu, 14 Apr 2016 19:48:54 -0400 From: Andrey Ulanov To: linux-fsdevel@vger.kernel.org Cc: Alexander Viro , linux-kernel@vger.kernel.org, Andrey Ulanov Subject: [PATCH] namespace: update event counter when umounting a deleted dentry Date: Thu, 14 Apr 2016 16:48:32 -0700 Message-Id: <1460677712-26909-1-git-send-email-andreyu@google.com> X-Mailer: git-send-email 2.8.0.rc3.226.g39d4020 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org This change fixes a problem originally reported at: https://github.com/coreos/bugs/issues/1153 You can find code that reproduces it at: https://github.com/coreos/bugs/issues/1153#issuecomment-210201681 - m_start() in fs/namespace.c expects that ns->event is incremented each time a mount added or removed from ns->list. - umount_tree() removes items from the list but does not increment event counter, expecting that it's done before the function is called. - There are some codepaths that call umount_tree() without updating "event" counter. e.g. from __detach_mounts(). - When this happens m_start may reuse a cached mount structure that no longer belongs to ns->list (i.e. use after free which usually leads to infinite loop). This change fixes the above problem by incrementing global event counter before invoking umount_tree(). Signed-off-by: Andrey Ulanov --- fs/namespace.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/namespace.c b/fs/namespace.c index 4fb1691..b06bb9a 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -1562,6 +1562,7 @@ void __detach_mounts(struct dentry *dentry) goto out_unlock; lock_mount_hash(); + event++; while (!hlist_empty(&mp->m_list)) { mnt = hlist_entry(mp->m_list.first, struct mount, mnt_mp_list); if (mnt->mnt.mnt_flags & MNT_UMOUNT) { @@ -1576,7 +1577,7 @@ out_unlock: namespace_unlock(); } -/* +/* * Is the caller allowed to modify his namespace? */ static inline bool may_mount(void) @@ -1765,6 +1766,7 @@ void drop_collected_mounts(struct vfsmount *mnt) { namespace_lock(); lock_mount_hash(); + event++; umount_tree(real_mount(mnt), UMOUNT_SYNC); unlock_mount_hash(); namespace_unlock(); -- 2.8.0.rc3.226.g39d4020