From mboxrd@z Thu Jan  1 00:00:00 1970
From: ebiederm@xmission.com (Eric W. Biederman)
Subject: Re: [PATCH] Add PR_{GET,SET}_NO_NEW_PRIVS to prevent execve from granting privs
Date: Tue, 17 Jan 2012 15:57:20 -0800
Message-ID: <m1pqeh9927.fsf@fess.ebiederm.org>
References: <e2f7f5ab9f24294a6b4ef8b6a7eaa760e414d777.1326420331.git.luto@amacapital.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Will Drewry <wad@chromium.org>, torvalds@linux-foundation.org,
	linux-kernel@vger.kernel.org, keescook@chromium.org,
	john.johansen@canonical.com, serge.hallyn@canonical.com,
	coreyb@linux.vnet.ibm.com, pmoore@redhat.com, eparis@redhat.com,
	djm@mindrot.org, segoon@openwall.com, rostedt@goodmis.org,
	jmorris@namei.org, scarybeasts@gmail.com, avi@redhat.com,
	penberg@cs.helsinki.fi, viro@zeniv.linux.org.uk, luto@MIT.EDU,
	mingo@elte.hu, akpm@linux-foundation.org, khilman@ti.com,
	borislav.petkov@amd.com, amwang@redhat.com, oleg@redhat.com,
	ak@linux.intel.com, eric.dumazet@gmail.com, gregkh@suse.de,
	dhowells@redhat.com, daniel.lezcano@free.fr,
	linux-fsdevel@vger.kernel.org,
	linux-security-module@vger.kernel.org, olofj@chromium.org,
	mhalcrow@google.com, dlaor@redhat.com, corbet@lwn.net,
	alan@lxorguk.ukuu.org.uk
To: Andy Lutomirski <luto@amacapital.net>
Return-path: <linux-kernel-owner@vger.kernel.org>
In-Reply-To: <e2f7f5ab9f24294a6b4ef8b6a7eaa760e414d777.1326420331.git.luto@amacapital.net>
	(Andy Lutomirski's message of "Thu, 12 Jan 2012 18:12:27 -0800")
Sender: linux-kernel-owner@vger.kernel.org
List-Id: linux-fsdevel.vger.kernel.org

Andy Lutomirski <luto@amacapital.net> writes:

> With this set, a lot of dangerous operations (chroot, unshare, etc)
> become a lot less dangerous because there is no possibility of
> subverting privileged binaries.
>
> This patch completely breaks apparmor.  Someone who understands (and
> uses) apparmor should fix it or at least give me a hint.

Foolish implementation question.

Is there a reason why we are putting this in a new field in the task
struct instead of adding a new flag in securebits in struct cred?

It seems to me like putting this information on the task_struct instead
of struct cred will tend to make this frozen state transition less
discoverable for future developers.

Also I would expect if you were applying this policy that you would
to ensure that task->read_cred == task->cred.

Do we want to disable setuid() and it's friends as well?  Certainly
freezing everything at exec is good I'm just wondering if handling
the handful of other cases where we explicitly change the permissions
on a process might be interesting.

My gut says you are trying to implement SECURE_NO_CHANGEPRIVS.

Eric