Re: [net,v2] net/packet: fix packet receive on L3 devices without visible hard header

["Jason A. Donenfeld" <Jason@zx2c4.com>]

On 11/21/20, Eyal Birger <eyal.birger@gmail.com> wrote:
> In the patchset merged by commit b9fcf0a0d826
> ("Merge branch 'support-AF_PACKET-for-layer-3-devices'") L3 devices which
> did not have header_ops were given one for the purpose of protocol parsing
> on af_packet transmit path.
>
> That change made af_packet receive path regard these devices as having a
> visible L3 header and therefore aligned incoming skb->data to point to the
> skb's mac_header. Some devices, such as ipip, xfrmi, and others, do not
> reset their mac_header prior to ingress and therefore their incoming
> packets became malformed.
>
> Ideally these devices would reset their mac headers, or af_packet would be
> able to rely on dev->hard_header_len being 0 for such cases, but it seems
> this is not the case.
>
> Fix by changing af_packet RX ll visibility criteria to include the
> existence of a '.create()' header operation, which is used when creating
> a device hard header - via dev_hard_header() - by upper layers, and does
> not exist in these L3 devices.
>
> As this predicate may be useful in other situations, add it as a common
> dev_has_header() helper in netdevice.h.
>
> Fixes: b9fcf0a0d826 ("Merge branch
> 'support-AF_PACKET-for-layer-3-devices'")
> Signed-off-by: Eyal Birger <eyal.birger@gmail.com>
>
> ---
>
> v2:
>   - add common dev_has_header() helper as suggested by Willem de Bruijn
> ---
>  include/linux/netdevice.h |  5 +++++
>  net/packet/af_packet.c    | 18 +++++++++---------
>  2 files changed, 14 insertions(+), 9 deletions(-)
>
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index 964b494b0e8d..fa275a054f46 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -3137,6 +3137,11 @@ static inline bool dev_validate_header(const struct
> net_device *dev,
>  	return false;
>  }
>
> +static inline bool dev_has_header(const struct net_device *dev)
> +{
> +	return dev->header_ops && dev->header_ops->create;
> +}
> +
>  typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr,
>  			   int len, int size);
>  int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
> diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
> index cefbd50c1090..7a18ffff8551 100644
> --- a/net/packet/af_packet.c
> +++ b/net/packet/af_packet.c
> @@ -93,8 +93,8 @@
>
>  /*
>     Assumptions:
> -   - If the device has no dev->header_ops, there is no LL header visible
> -     above the device. In this case, its hard_header_len should be 0.
> +   - If the device has no dev->header_ops->create, there is no LL header
> +     visible above the device. In this case, its hard_header_len should be
> 0.
>       The device may prepend its own header internally. In this case, its
>       needed_headroom should be set to the space needed for it to add its
>       internal header.
> @@ -108,26 +108,26 @@
>  On receive:
>  -----------
>
> -Incoming, dev->header_ops != NULL
> +Incoming, dev_has_header(dev) == true
>     mac_header -> ll header
>     data       -> data
>
> -Outgoing, dev->header_ops != NULL
> +Outgoing, dev_has_header(dev) == true
>     mac_header -> ll header
>     data       -> ll header
>
> -Incoming, dev->header_ops == NULL
> +Incoming, dev_has_header(dev) == false
>     mac_header -> data
>       However drivers often make it point to the ll header.
>       This is incorrect because the ll header should be invisible to us.
>     data       -> data
>
> -Outgoing, dev->header_ops == NULL
> +Outgoing, dev_has_header(dev) == false
>     mac_header -> data. ll header is invisible to us.
>     data       -> data
>
>  Resume
> -  If dev->header_ops == NULL we are unable to restore the ll header,
> +  If dev_has_header(dev) == false we are unable to restore the ll header,
>      because it is invisible to us.
>
>
> @@ -2069,7 +2069,7 @@ static int packet_rcv(struct sk_buff *skb, struct
> net_device *dev,
>
>  	skb->dev = dev;
>
> -	if (dev->header_ops) {
> +	if (dev_has_header(dev)) {
>  		/* The device has an explicit notion of ll header,
>  		 * exported to higher levels.
>  		 *
> @@ -2198,7 +2198,7 @@ static int tpacket_rcv(struct sk_buff *skb, struct
> net_device *dev,
>  	if (!net_eq(dev_net(dev), sock_net(sk)))
>  		goto drop;
>
> -	if (dev->header_ops) {
> +	if (dev_has_header(dev)) {
>  		if (sk->sk_type != SOCK_DGRAM)
>  			skb_push(skb, skb->data - skb_mac_header(skb));
>  		else if (skb->pkt_type == PACKET_OUTGOING) {

Thanks for fixing this. Patch seems correct to me.

Acked-by: Jason A. Donenfeld <Jason@zx2c4.com>