netdev - Re: [PATCH] net: bridge: add missing NULL checks

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 9 Apr 2018 01:25:41 +0300
From:   Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
To:     Laszlo Toth <laszlth@...il.com>,
        Stephen Hemminger <stephen@...workplumber.org>,
        "David S. Miller" <davem@...emloft.net>
Cc:     bridge@...ts.linux-foundation.org, netdev@...r.kernel.org
Subject: Re: [PATCH] net: bridge: add missing NULL checks

On 08/04/18 20:49, Laszlo Toth wrote:
> br_port_get_rtnl() can return NULL
> 
> Signed-off-by: Laszlo Toth <laszlth@...il.com>
> ---
>   net/bridge/br_netlink.c | 12 ++++++++++--
>   1 file changed, 10 insertions(+), 2 deletions(-)
> 

Nacked-by: Nikolay Aleksandrov <nikolay@...ulusnetworks.com>
More below.

> diff --git a/net/bridge/br_netlink.c b/net/bridge/br_netlink.c
> index 015f465c..cbec11f 100644
> --- a/net/bridge/br_netlink.c
> +++ b/net/bridge/br_netlink.c
> @@ -939,14 +939,17 @@ static int br_port_slave_changelink(struct net_device *brdev,
>   				    struct nlattr *data[],
>   				    struct netlink_ext_ack *extack)
>   {
> +	struct net_bridge_port *port = br_port_get_rtnl(dev);
>   	struct net_bridge *br = netdev_priv(brdev);
>   	int ret;
>   
>   	if (!data)
>   		return 0;
> +	if (!port)
> +		return -EINVAL;
>   

If we're here, it means the master device of dev is a bridge => dev is a bridge port,
since we're running with RTNL that cannot change, so this check is unnecessary.

Have you actually hit a bug with this code ?

>   	spin_lock_bh(&br->lock);
> -	ret = br_setport(br_port_get_rtnl(dev), data);
> +	ret = br_setport(port, data);
>   	spin_unlock_bh(&br->lock);
>   
>   	return ret;
> @@ -956,7 +959,12 @@ static int br_port_fill_slave_info(struct sk_buff *skb,
>   				   const struct net_device *brdev,
>   				   const struct net_device *dev)
>   {
> -	return br_port_fill_attrs(skb, br_port_get_rtnl(dev));
> +	struct net_bridge_port *port = br_port_get_rtnl(dev);
> +
> +	if (!port)
> +		return -EINVAL;
> +
> +	return br_port_fill_attrs(skb, port);

Same rationale here, fill_slave_info is called via a master device's ops
under RTNL, which means dev is a bridge port and that also cannot change.

If you have hit a bug with this code, can we see the trace ?
The problem might be elsewhere.

Thanks,
  Nik

>   }
>   
>   static size_t br_port_get_slave_size(const struct net_device *brdev,
>