Subject: Re: [PATCH v2] ip6_output: ensure flow saddr actually belongs to
 device
To: "Jason A. Donenfeld" <Jason@zx2c4.com>,
        Netdev <netdev@vger.kernel.org>,
        WireGuard mailing list <wireguard@lists.zx2c4.com>,
        LKML <linux-kernel@vger.kernel.org>,
        YOSHIFUJI Hideaki <hideaki.yoshifuji@miraclelinux.com>,
        Hannes Frederic Sowa <hannes@stressinduktion.org>
References: <405b2e79-854d-4c30-07b0-bd524137d2f6@cumulusnetworks.com>
 <20161113190251.1027-1-Jason@zx2c4.com>
From: David Ahern <dsa@cumulusnetworks.com>
Message-ID: <27cccef1-06d9-74b3-5b8a-912850119a76@cumulusnetworks.com>
Date: Sun, 13 Nov 2016 13:45:31 -0700
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.11; rv:45.0)
 Gecko/20100101 Thunderbird/45.4.0
MIME-Version: 1.0
In-Reply-To: <20161113190251.1027-1-Jason@zx2c4.com>
Content-Type: text/plain; charset=windows-1252
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 3393
Lines: 88

On 11/13/16 12:02 PM, Jason A. Donenfeld wrote:
> This puts the IPv6 routing functions in parity with the IPv4 routing
> functions. Namely, we now check in v6 that if a flowi6 requests an
> saddr, the returned dst actually corresponds to a net device that has
> that saddr. This mirrors the v4 logic with __ip_dev_find in
> __ip_route_output_key_hash. In the event that the returned dst is not
> for a dst with a dev that has the saddr, we return -EINVAL, just like
> v4; this makes it easy to use the same error handlers for both cases.
> 
> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
> Cc: David Ahern <dsa@cumulusnetworks.com>
> ---
> Changes from v1:
>    This moves the check to the top and now sees if it's a valid address
>    on _any_ device, not just the one in dst.
> 
>  include/net/ipv6.h    |  2 ++
>  net/ipv6/ip6_output.c | 28 ++++++++++++++++++++++++++++
>  2 files changed, 30 insertions(+)
> 
> diff --git a/include/net/ipv6.h b/include/net/ipv6.h
> index 8fed1cd..e5dc14f 100644
> --- a/include/net/ipv6.h
> +++ b/include/net/ipv6.h
> @@ -914,6 +914,8 @@ struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
>  					 const struct in6_addr *final_dst);
>  struct dst_entry *ip6_blackhole_route(struct net *net,
>  				      struct dst_entry *orig_dst);
> +struct net_device *__ip6_dev_find(struct net *net, struct in6_addr *addr,
> +				  bool devref);
>  
>  /*
>   *	skb processing functions
> diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
> index 6001e78..371170b 100644
> --- a/net/ipv6/ip6_output.c
> +++ b/net/ipv6/ip6_output.c
> @@ -916,6 +916,30 @@ static struct dst_entry *ip6_sk_dst_check(struct sock *sk,
>  	return dst;
>  }
>  
> +/**
> + * __ip6_dev_find - find the first device with a given source address.
> + * @net: the net namespace
> + * @addr: the source address
> + * @devref: if true, take a reference on the found device
> + *
> + * If a caller uses devref=false, it should be protected by RCU, or RTNL
> + */
> +struct net_device *__ip6_dev_find(struct net *net, struct in6_addr *addr, bool devref)
> +{
> +	struct net_device *result;
> +
> +	rcu_read_lock();
> +	for_each_netdev_rcu(net, result) {
> +		if (ipv6_chk_addr(net, addr, result, 1))
> +			break;
> +	}
> +	if (result && devref)
> +		dev_hold(result);
> +	rcu_read_unlock();
> +	return result;
> +}
> +EXPORT_SYMBOL(__ip6_dev_find);

You don't need a new function to walk all interfaces; just use ipv6_chk_addr with a dev arg of NULL. IPv6 has a hash table with all unicast addresses -- inet6_addr_lst. ipv6_chk_addr is checking that list for the address in question. The actual device is not relevant for verifying the address is a valid local one (though the device can be returned from ifp->idev->dev if ever needed).

So drop the above ...

> +
>  static int ip6_dst_lookup_tail(struct net *net, const struct sock *sk,
>  			       struct dst_entry **dst, struct flowi6 *fl6)
>  {
> @@ -926,6 +950,10 @@ static int ip6_dst_lookup_tail(struct net *net, const struct sock *sk,
>  	int err;
>  	int flags = 0;
>  
> +	if (!ipv6_addr_any(&fl6->saddr) &&
> +	    !__ip6_dev_find(net, &fl6->saddr, false))

... and just use ipv6_chk_addr here.

> +		return -EINVAL;
> +
>  	/* The correct way to handle this would be to do
>  	 * ip6_route_get_saddr, and then ip6_route_output; however,
>  	 * the route-specific preferred source forces the
>