Messages in this thread | | | From | Pravin Shelar <> | Date | Thu, 9 Dec 2021 23:36:09 -0800 | Subject | Re: [PATCH net-next v8] net: openvswitch: IPv6: Add IPv6 extension header support |
| |
()
On Mon, Dec 6, 2021 at 3:00 PM Cpp Code <cpp.code.lv@gmail.com> wrote: > > On Thu, Dec 2, 2021 at 9:28 PM Pravin Shelar <pravin.ovn@gmail.com> wrote: > > > > On Thu, Dec 2, 2021 at 12:20 PM Cpp Code <cpp.code.lv@gmail.com> wrote: > > > > > > On Wed, Dec 1, 2021 at 11:34 PM Pravin Shelar <pravin.ovn@gmail.com> wrote: > > > > > > > > On Wed, Nov 24, 2021 at 11:33 AM Toms Atteka <cpp.code.lv@gmail.com> wrote: > > > > > > > > > > This change adds a new OpenFlow field OFPXMT_OFB_IPV6_EXTHDR and > > > > > packets can be filtered using ipv6_ext flag. > > > > > > > > > > Signed-off-by: Toms Atteka <cpp.code.lv@gmail.com> > > > > > --- > > > > > include/uapi/linux/openvswitch.h | 6 ++ > > > > > net/openvswitch/flow.c | 140 +++++++++++++++++++++++++++++++ > > > > > net/openvswitch/flow.h | 14 ++++ > > > > > net/openvswitch/flow_netlink.c | 26 +++++- > > > > > 4 files changed, 184 insertions(+), 2 deletions(-) > > > > > > > > > > diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswitch.h > > > > > index a87b44cd5590..43790f07e4a2 100644 > > > > > --- a/include/uapi/linux/openvswitch.h > > > > > +++ b/include/uapi/linux/openvswitch.h > > > > > @@ -342,6 +342,7 @@ enum ovs_key_attr { > > > > > OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV4, /* struct ovs_key_ct_tuple_ipv4 */ > > > > > OVS_KEY_ATTR_CT_ORIG_TUPLE_IPV6, /* struct ovs_key_ct_tuple_ipv6 */ > > > > > OVS_KEY_ATTR_NSH, /* Nested set of ovs_nsh_key_* */ > > > > > + OVS_KEY_ATTR_IPV6_EXTHDRS, /* struct ovs_key_ipv6_exthdr */ > > > > > > > > > > #ifdef __KERNEL__ > > > > > OVS_KEY_ATTR_TUNNEL_INFO, /* struct ip_tunnel_info */ > > > > > @@ -421,6 +422,11 @@ struct ovs_key_ipv6 { > > > > > __u8 ipv6_frag; /* One of OVS_FRAG_TYPE_*. */ > > > > > }; > > > > > > > > > > +/* separate structure to support backward compatibility with older user space */ > > > > > +struct ovs_key_ipv6_exthdrs { > > > > > + __u16 hdrs; > > > > > +}; > > > > > + > > > > > struct ovs_key_tcp { > > > > > __be16 tcp_src; > > > > > __be16 tcp_dst; > > > > > diff --git a/net/openvswitch/flow.c b/net/openvswitch/flow.c > > > > > index 9d375e74b607..28acb40437ca 100644 > > > > > --- a/net/openvswitch/flow.c > > > > > +++ b/net/openvswitch/flow.c > > > > > @@ -239,6 +239,144 @@ static bool icmphdr_ok(struct sk_buff *skb) > > > > > sizeof(struct icmphdr)); > > > > > } > > > > > > > > > > +/** > > > > > + * get_ipv6_ext_hdrs() - Parses packet and sets IPv6 extension header flags. > > > > > + * > > > > > + * @skb: buffer where extension header data starts in packet > > > > > + * @nh: ipv6 header > > > > > + * @ext_hdrs: flags are stored here > > > > > + * > > > > > + * OFPIEH12_UNREP is set if more than one of a given IPv6 extension header > > > > > + * is unexpectedly encountered. (Two destination options headers may be > > > > > + * expected and would not cause this bit to be set.) > > > > > + * > > > > > + * OFPIEH12_UNSEQ is set if IPv6 extension headers were not in the order > > > > > + * preferred (but not required) by RFC 2460: > > > > > + * > > > > > + * When more than one extension header is used in the same packet, it is > > > > > + * recommended that those headers appear in the following order: > > > > > + * IPv6 header > > > > > + * Hop-by-Hop Options header > > > > > + * Destination Options header > > > > > + * Routing header > > > > > + * Fragment header > > > > > + * Authentication header > > > > > + * Encapsulating Security Payload header > > > > > + * Destination Options header > > > > > + * upper-layer header > > > > > + */ > > > > > +static void get_ipv6_ext_hdrs(struct sk_buff *skb, struct ipv6hdr *nh, > > > > > + u16 *ext_hdrs) > > > > > +{ > > > > > + u8 next_type = nh->nexthdr; > > > > > + unsigned int start = skb_network_offset(skb) + sizeof(struct ipv6hdr); > > > > > + int dest_options_header_count = 0; > > > > > + > > > > > + *ext_hdrs = 0; > > > > > + > > > > > + while (ipv6_ext_hdr(next_type)) { > > > > > + struct ipv6_opt_hdr _hdr, *hp; > > > > > + > > > > > + switch (next_type) { > > > > > + case IPPROTO_NONE: > > > > > + *ext_hdrs |= OFPIEH12_NONEXT; > > > > > + /* stop parsing */ > > > > > + return; > > > > > + > > > > > + case IPPROTO_ESP: > > > > > + if (*ext_hdrs & OFPIEH12_ESP) > > > > > + *ext_hdrs |= OFPIEH12_UNREP; > > > > > + if ((*ext_hdrs & ~(OFPIEH12_HOP | OFPIEH12_DEST | > > > > > + OFPIEH12_ROUTER | IPPROTO_FRAGMENT | > > > > > + OFPIEH12_AUTH | OFPIEH12_UNREP)) || > > > > > + dest_options_header_count >= 2) { > > > > > + *ext_hdrs |= OFPIEH12_UNSEQ; > > > > > + } > > > > > + *ext_hdrs |= OFPIEH12_ESP; > > > > > + break; > > > > you need to check_header() before looking into each extension header. > > > > > > Could you elaborate why I need to add check_header(), > > > skb_header_pointer() is doing sanitization. > > > > I mean check_header() would allow you to read the header without > > copying the bits, it is used in ovs flow extraction so its usual > > check. > > But check_header() will call *__pskb_pull_tail which in turn will copy > bits if data will be fragmented. > OVS flow extract uses this function to extract flow upto L4, so skb_header_pointer() is not saving any copy operation.
> /* Moves tail of skb head forward, copying data from fragmented part, > * when it is necessary. > * 1. It may fail due to malloc failure. > * 2. It may change skb pointers. > * > * It is pretty complicated. Luckily, it is called only in exceptional cases. > */ > void *__pskb_pull_tail(struct sk_buff *skb, int delta) > > as well I noticed that for example commit > 4a06fa67c4da20148803525151845276cdb995c1 is moving from > pskb_may_pull() to skb_header_pointer() ok, I see advantage of using skb_header_pointer() in this case, but replacing all check_header() with skb_header_pointer() would add lot of copy operation in flow extract. Anyways for this use case skb_header_pointer() is fine.
Acked-by: Pravin B Shelar <pshelar@ovn.org>
| |