Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Sun, 29 Jul 2018 08:59:06 -0500
From:   Josh Poimboeuf <jpoimboe@redhat.com>
To:     Jeremy Cline <jcline@redhat.com>
Cc:     "David S . Miller" <davem@davemloft.net>, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, stable@vger.kernel.org
Subject: Re: [PATCH 2/2] net: socket: Fix potential spectre v1 gadget in
 sock_is_registered
Message-ID: <20180729135906.lgqo5ue6it3hl2da@treble>
References: <20180727224302.5503-1-jcline@redhat.com>
 <20180727224302.5503-3-jcline@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20180727224302.5503-3-jcline@redhat.com>
User-Agent: NeoMutt/20180716
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Fri, Jul 27, 2018 at 10:43:02PM +0000, Jeremy Cline wrote:
> 'family' can be a user-controlled value, so sanitize it after the bounds
> check to avoid speculative out-of-bounds access.
> 
> Cc: Josh Poimboeuf <jpoimboe@redhat.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Jeremy Cline <jcline@redhat.com>
> ---
>  net/socket.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/net/socket.c b/net/socket.c
> index f15d5cbb3ba4..608e29ae6baf 100644
> --- a/net/socket.c
> +++ b/net/socket.c
> @@ -2672,7 +2672,8 @@ EXPORT_SYMBOL(sock_unregister);
>  
>  bool sock_is_registered(int family)
>  {
> -	return family < NPROTO && rcu_access_pointer(net_families[family]);
> +	return family < NPROTO &&
> +		rcu_access_pointer(net_families[array_index_nospec(family, NPROTO)]);
>  }
>  
>  static int __init sock_init(void)

This is another one where I think it would be better to do the nospec
clamp higher up the call chain.  The untrusted 'family' value comes from
__sock_diag_cmd():

__sock_diag_cmd
  sock_load_diag_module
    sock_is_registered

That function has a bounds check, and also uses the value in some other
array accesses:

	if (req->sdiag_family >= AF_MAX)
		return -EINVAL;

	if (sock_diag_handlers[req->sdiag_family] == NULL)
		sock_load_diag_module(req->sdiag_family, 0);

	mutex_lock(&sock_diag_table_mutex);
	hndl = sock_diag_handlers[req->sdiag_family];
	...

So I think clamping 'req->sdiag_family' right after the bounds check
would be the way to go.

-- 
Josh