Re: [PATCH nf-next] netfilter: xt_CHECKSUM: avoid bad offload warnings on GSO packets

[Michal Kubecek]

On Thu, Aug 24, 2017 at 03:08:42PM +0200, Davide Caratti wrote:
> On Thu, 2017-08-24 at 13:07 +0200, Michal Kubecek wrote:
> > On Thu, Aug 24, 2017 at 12:51:18PM +0200, Florian Westphal wrote:
> > > Michal Kubecek <mkubecek@xxxxxxx> wrote:
> > > > When --checksum_fill action is applied to a GSO packet, checksum_tg() calls
> > > > skb_checksum_help() which is only meant to be applied to non-GSO packets so
> > > > that it issues a warning.
> > > > 
> > > > This can be easily triggered by using e.g.
> > > > 
> > > >   iptables -t mangle -A OUTPUT -j CHECKSUM --checksum-fill
> > > > 
> > > > and sending TCP stream via a device with GSO enabled.
> > > > 
> > > > While this can be considered a misconfiguration, I believe the bad offload
> > > > warning is supposed to catch bugs in drivers and networking stack, not
> > > > misconfigured firewalls. So let's ignore such packets and only issue a one
> > > > time warning with pr_warn_once() rather than a WARN with stack trace and
> > > > tainted kernel.
> > > 
> > > Why issue a warning at all?
> > > What kind of action should be taken upon seeing such warning?
> > 
> > Check and fix the configuration. The reason why I left at least some
> > kind of warning is that the module does something that is unexpected as
> > the checksum is not calculated (this module is often used in
> > virtualization environments where "hardware checksum offload" in fact
> > means the checksum is not computed at all).
> > 
> 
> hello Michal,
> 
> GSO should be capable of computing the checksum on individual segments
> later, so I also think the warning can be removed.
> 
> Small nit: may I suggest you to call skb_csum_hwoffload_help() instead of
> skb_checksum_help(), so that we avoid corrupting SCTP packets in case they
> hit xt_CHECKSUM target?

That sounds like an independent issue so it should be probably handled
by a separate patch.

Michal Kubecek