On Fri, May 6, 2016 at 3:12 PM, Jesse Gross <je...@kernel.org> wrote: > On Thu, Apr 21, 2016 at 6:54 PM, Pravin B Shelar <pshe...@ovn.org> wrote: >> diff --git a/lib/netdev-native-tnl.h b/lib/netdev-native-tnl.h >> index dbe6bd0..a0dfa8c 100644 >> --- a/lib/netdev-native-tnl.h >> +++ b/lib/netdev-native-tnl.h >> @@ -82,6 +82,31 @@ ipv6_hdr(void *eth) >> return (void *)((char *)eth + sizeof (struct eth_header)); >> } >> >> +static inline void * >> +ip_build_header(struct ovs_action_push_tnl *data, >> + uint8_t next_proto, >> + unsigned int *hlen, >> + bool *is_ipv6) > > I would also prefix this function to indicate where it is coming from. > I'm also not entirely sure that this should be an inline (especially > after my next comment). > > There's a somewhat odd split between this function and > tnl_port_build_header(). It seems like it would better to consolidate > all of the IP header building here in a single place. I think it might > also be better to split out the IP and Ethernet layers in the > build/push/pop functions - even though they are always done together > it is easier to understand from a layering perspective.
It it good idea to consolidate header build functionality in one place. But I do not see how can we split out IP and Ether layer functionality from push header function. So I have kept that part same. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
