Received-SPF: pass (google.com: domain of linux-kernel+bounces-122424-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Date: Wed, 27 Mar 2024 23:26:51 -0600
From: Kees Cook <kees@kernel.org>
To: "Luck, Tony" <tony.luck@intel.com>, Kees Cook <keescook@chromium.org>
CC: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
 linux-hardening@vger.kernel.org
Subject: Re: __randomize_layout;
User-Agent: K-9 Mail for Android
In-Reply-To: <SJ1PR11MB608346DDA3F41ABB893C7860FC342@SJ1PR11MB6083.namprd11.prod.outlook.com>
References: <SJ1PR11MB608346DDA3F41ABB893C7860FC342@SJ1PR11MB6083.namprd11.prod.outlook.com>
Message-ID: <1B6E0664-D1B5-4FCC-A874-8A9587924687@kernel.org>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable


On March 27, 2024 9:21:59 AM MDT, "Luck, Tony" <tony=2Eluck@intel=2Ecom> w=
rote:
>This e-mail is to check you on whether that __randomize_layout can shuffl=
e the
>fields inside that nested union/structure=2E I tried some experiments, an=
d in a
>few kernel builds I saw the whole block move to different offsets, but th=
e order
>of x86_vendor, x86, x86_model, and x86_reserved was preserved=2E

Yes, this is an intentional behavior: __randomize_layout will only apply t=
o the struct it is attached to, and is not enabled for any substructs (anon=
ymous or otherwise)=2E

>But experiments aren't proof=2E Nor defense against future versions of
>scripts/gcc-plugins/randomize_layout_plugin=2Ec becoming smarter or
>more aggressive about changing layout=2E

The behavior is also supported natively by Clang -- neither implementation=
 is likely to ever change its treatment of substructs as it would kind of c=
ause chaos=2E

So you're all good! :)

--=20
Kees Cook