Re: [PATCH RFC 14/14] net: introduce and use KERNEL_MAX_HEADER_PARSE_ADDRLEN

[Florian Westphal <fw@strlen.de>]

Willem de Bruijn <willemb@google.com> wrote:
> > index 6fb6bdf..97eb50c 100644
> > --- a/net/mac802154/iface.c
> > +++ b/net/mac802154/iface.c
> > @@ -433,6 +433,7 @@ mac802154_header_parse(const struct sk_buff *skb, unsigned char *haddr)
> >                 return 0;
> >         }
> >
> > +       dev_validate_header_parse_addrlen(sizeof(*addr));
> >         *addr = hdr.source;
> >         return sizeof(*addr);
> >  }
> > diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
> > index 9db8369..1a4dcbb 100644
> > --- a/net/packet/af_packet.c
> > +++ b/net/packet/af_packet.c
> > @@ -1816,7 +1816,8 @@ static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
> >                 skb = nskb;
> >         }
> >
> > -       sock_skb_cb_check_size(sizeof(*PACKET_SKB_CB(skb)) + MAX_ADDR_LEN - 8);
> > +       sock_skb_cb_check_size(sizeof(*PACKET_SKB_CB(skb)) +
> > +                              KERNEL_MAX_HEADER_PARSE_ADDRLEN);
> 
> KERNEL_MAX_HEADER_PARSE_ADDRLEN - 8, as the first 8
> bytes are still written into the space allocated in packet_skb_cb?

Good catch, thanks!

I'll make this a FIELD_SIZEOF construct using sll_addr to make
it a bit more obvious where that 8 is coming from.