From mboxrd@z Thu Jan  1 00:00:00 1970
From: Christoph Hellwig <hch@lst.de>
Subject: truncate in ecryptfs
Date: Wed, 14 Oct 2009 17:12:21 +0200
Message-ID: <20091014151221.GA18371@lst.de>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: linux-fsdevel@vger.kernel.org
To: tyhicks@linux.vnet.ibm.com, kirkland@canonical.com
Return-path: <linux-fsdevel-owner@vger.kernel.org>
Received: from verein.lst.de ([213.95.11.210]:43533 "EHLO verein.lst.de"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1755555AbZJNPNB (ORCPT <rfc822;linux-fsdevel@vger.kernel.org>);
	Wed, 14 Oct 2009 11:13:01 -0400
Content-Disposition: inline
Sender: linux-fsdevel-owner@vger.kernel.org
List-ID: <linux-fsdevel.vger.kernel.org>

It looks like ecryptfs calls vmtruncate directly on the lower filesystem,
which is quite wrong.  Vmtruncate is only a helper for the filesystem,
and while most filesystems end up calling vmtruncate from their
->setattr implementation if ATTR_SIZE is set there are many that require
additional work.  I think ecryptfs needs to got through notify_change()
/ ->setattr if it wants truncate to work reliably and without silent
corruption or leaking blocks on a variety of filesystems.