Received-SPF: pass (google.com: domain of linux-wireless+bounces-7992-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Message-ID: <9fca4c0d496eb731f571cd8eacd409b9a9e61dae.camel@sipsolutions.net>
Subject: Re: [PATCH v2] wifi: mac80211: Avoid address calculations via out
 of bounds array indexing
From: Johannes Berg <johannes@sipsolutions.net>
To: Simon Horman <horms@kernel.org>, Kenton Groombridge <concord@gentoo.org>
Cc: davem@davemloft.net, edumazet@google.com, kuba@kernel.org,
 pabeni@redhat.com,  linux-wireless@vger.kernel.org, netdev@vger.kernel.org,
  linux-kernel@vger.kernel.org, linux-hardening@vger.kernel.org, Kees Cook
 <keescook@chromium.org>
Date: Thu, 23 May 2024 11:35:37 +0200
In-Reply-To: <20240517204532.GC475595@kernel.org>
References: <20240517145420.8891-1-concord@gentoo.org>
	 <20240517204532.GC475595@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.50.4 (3.50.4-1.fc39) 
Precedence: bulk
MIME-Version: 1.0

On Fri, 2024-05-17 at 21:45 +0100, Simon Horman wrote:
>=20
> FWWIW, it seems unfortunate to me that the __counted_by field (n_channels=
)
> is set some distance away from the allocation of the flex-array (channels=
)
> whose bounds it checks. It seems it would be pretty easy for a bug in the
> code being updated here to result in an overrun.
>=20

In a way, this is a more general problem, this allocates the max we know
we might need, but then filter it down. It'd have to iterate twice to
actually allocate the "correct" size, but then you could still have bugs
by having different filter conditions in the two loops ...

Don't see any good solutions to this kind of code?

johannes