From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mike Snitzer Subject: Re: [PATCH 0/3] dm-crypt: Adds support for wiping key when doing suspend/hibernation Date: Thu, 9 Apr 2015 09:12:08 -0400 Message-ID: <20150409131208.GA9504@redhat.com> References: <1428254419-7334-1-git-send-email-pali.rohar@gmail.com> <20150406130045.GA18583@redhat.com> <201504061529.57299@pali> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Content-Disposition: inline In-Reply-To: <201504061529.57299@pali> Sender: linux-pm-owner@vger.kernel.org To: Pali =?iso-8859-1?Q?Roh=E1r?= Cc: Alasdair Kergon , Neil Brown , "Rafael J. Wysocki" , Len Brown , Pavel Machek , dm-devel@redhat.com, linux-raid@vger.kernel.org, linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org List-Id: linux-raid.ids On Mon, Apr 06 2015 at 9:29am -0400, Pali Roh=E1r wrote: > On Monday 06 April 2015 15:00:46 Mike Snitzer wrote: > > On Sun, Apr 05 2015 at 1:20pm -0400, > >=20 > > Pali Roh=E1r wrote: > > > This patch series increase security of suspend and hibernate > > > actions. It allows user to safely wipe crypto keys before > > > suspend and hibernate actions starts without race > > > conditions on userspace process with heavy I/O. > > >=20 > > > To automatically wipe cryto key for before > > > hibernate action call: $ dmsetup message 0 key > > > wipe_on_hibernation 1 > > >=20 > > > To automatically wipe cryto key for before suspend > > > action call: $ dmsetup message 0 key > > > wipe_on_suspend 1 > > >=20 > > > (Value 0 after wipe_* string reverts original behaviour - to > > > not wipe key) > >=20 > > Can you elaborate on the attack vector your changes are meant > > to protect against? The user already authorized access, why > > is it inherently dangerous to _not_ wipe the associated key > > across these events? >=20 > Hi, >=20 > yes, I will try to explain current problems with cryptsetup=20 > luksSuspend command and hibernation. >=20 > First, sometimes it is needed to put machine into other hands.=20 > You can still watch other person what is doing with machine, but=20 > once if you let machine unlocked (e.g opened luks disk), she/he=20 > can access encrypted data. >=20 > If you turn off machine, it could be safe, because luks disk=20 > devices are locked. But if you enter machine into suspend or=20 > hibernate state luks devices are still open. And my patches try=20 > to achieve similar security as when machine is off (=3D no crypto=20 > keys in RAM or on swap). >=20 > When doing hibernate on unencrypted swap it is to prevent leaking=20 > crypto keys to hibernate image (which is stored in swap). >=20 > When doing suspend action it is again to prevent leaking crypto=20 > keys. E.g when you suspend laptop and put it off (somebody can=20 > remove RAMs and do some cold boot attack). >=20 > The most common situation is: > You have mounted partition from dm-crypt device (e.g. /home/),=20 > some userspace processes access it (e.g opened firefox which=20 > still reads/writes to cache ~/.firefox/) and you want to drop=20 > crypto keys from kernel for some time. >=20 > For that operation there is command cryptsetup luksSuspend, which=20 > suspend dm device and then tell kernel to wipe crypto keys. All=20 > I/O operations are then stopped and userspace processes which=20 > want to do some those I/O operations are stopped too (until you=20 > call cryptsetup luksResume and enter correct key). >=20 > Now if you want to suspend/hiberate your machine (when some of dm=20 > devices are suspeneded and some processes are stopped due to=20 > pending I/O) it is not possible. Kernel freeze_processes function=20 > will fail because userspace processes are still stopped inside=20 > some I/O syscall (read/write, etc,...). >=20 > My patches fixes this problem and do those operations (suspend dm=20 > device, wipe crypto keys, enter suspend/hiberate) in correct=20 > order and without race condition. >=20 > dm device is suspended *after* userspace processes are freezed=20 > and after that are crypto keys wiped. And then computer/laptop=20 > enters into suspend/hibernate state. Wouldn't it be better to fix freeze_processes() to be tolerant of processes that are hung as a side-effect of their backing storage being suspended? A hibernate shouldn't fail simply because a user chose to suspend a DM device. Then this entire problem goes away and the key can be wiped from userspace (like you said above).