linux-kernel - Re: [RFC 1/1] net: introduce OpenVPN Data Channel Offload (ovpn-dco)

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <YuKLMIJcb9OkBHry@lunn.ch>
Date:   Thu, 28 Jul 2022 15:12:16 +0200
From:   Andrew Lunn <andrew@...n.ch>
To:     Antonio Quartulli <antonio@...nvpn.net>
Cc:     netdev@...r.kernel.org, David Miller <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>, linux-kernel@...r.kernel.org
Subject: Re: [RFC 1/1] net: introduce OpenVPN Data Channel Offload (ovpn-dco)

On Thu, Jul 28, 2022 at 09:44:18AM +0200, Antonio Quartulli wrote:
> Hi,
> 
> On 19/07/2022 17:37, Andrew Lunn wrote:
> > > +static int ovpn_net_change_mtu(struct net_device *dev, int new_mtu)
> > > +{
> > > +	if (new_mtu < IPV4_MIN_MTU ||
> > > +	    new_mtu + dev->hard_header_len > IP_MAX_MTU)
> > > +		return -EINVAL;
> > 
> > If you set dev->min_mtu and dev->max_mtu, the core will validate this
> > for you, see dev_validate_mtu().
> 
> Yeah, thanks for the pointer.
> 
> > 
> > > +static int ovpn_get_link_ksettings(struct net_device *dev,
> > > +				   struct ethtool_link_ksettings *cmd)
> > > +{
> > > +	ethtool_convert_legacy_u32_to_link_mode(cmd->link_modes.supported, 0);
> > > +	ethtool_convert_legacy_u32_to_link_mode(cmd->link_modes.advertising, 0);
> > 
> > These two should not be needed. Look at tun, veth etc, they don't set
> > them.
> 
> I found this in tun.c:
> 
> 3512         ethtool_link_ksettings_zero_link_mode(cmd, supported);
> 3513         ethtool_link_ksettings_zero_link_mode(cmd, advertising);
> 
> Which seems a more appropriate version of my code, no?

I would trace is backwards. Where is cmd coming from? In order to
avoid unintentional information leaks, the core should be clearing any
memory which gets passed to a driver which might optionally be filled
in and then returned to user space. So take a look in net/ethtool, and
see if there is a memset() or a kzalloc() etc. If it is already been
zero'ed, you don't need this.

	  Andrew