> On Oct 21, 2015, at 3:59 AM, Thomas Graf <tgraf@xxxxxxx> wrote: > > On 10/20/15 at 03:20pm, Jarno Rajahalme wrote: >> Extend OVS conntrack interface to cover NAT. New nested >> OVS_CT_ATTR_NAT may be used to include NAT with a CT action. A bare >> OVS_CT_ATTR_NAT only mangles existing connections. If >> OVS_NAT_ATTR_SRC or OVS_NAT_ATTR_DST is included within the nested >> attributes, new (non-committed/non-confirmed) connections are mangled >> according to the rest of the nested attributes. >> >> This work extends on a branch by Thomas Graf at >> https://github.com/tgraf/ovs/tree/nat. >> >> Signed-off-by: Jarno Rajahalme <jrajahalme@xxxxxxxxxx> > > Awesome work. This is a great start. > > There are some, probably unintended, unrelated style changes. More > comments below. > Thanks for the review! >> +enum ovs_nat_attr { >> + OVS_NAT_ATTR_UNSPEC, >> + OVS_NAT_ATTR_SRC, >> + OVS_NAT_ATTR_DST, >> + OVS_NAT_ATTR_IP_MIN, >> + OVS_NAT_ATTR_IP_MAX, >> + OVS_NAT_ATTR_PROTO_MIN, >> + OVS_NAT_ATTR_PROTO_MAX, >> + OVS_NAT_ATTR_PERSISTENT, >> + OVS_NAT_ATTR_PROTO_HASH, >> + OVS_NAT_ATTR_PROTO_RANDOM, > > Simplify this with an OVS_NAT_ATTR_FLAGS? The use of separate flag attributes was actually intentional, as it makes the interface easier to understand, and code also easier to read. > >> @@ -137,11 +159,17 @@ static void __ovs_ct_update_key(struct sw_flow_key *key, u8 state, >> ovs_ct_get_label(ct, &key->ct.label); >> } >> >> -/* Update 'key' based on skb->nfct. If 'post_ct' is true, then OVS has >> - * previously sent the packet to conntrack via the ct action. >> +/* Update 'key' based on skb->nfct. If 'post_ct' is true, then OVS has >> + * previously sent the packet to conntrack via the ct action. If >> + * 'keep_nat_flags' is true, the existing NAT flags retained, else they are >> + * initialized from the connection status. >> */ >> static void ovs_ct_update_key(const struct sk_buff *skb, >> - struct sw_flow_key *key, bool post_ct) >> + struct sw_flow_key *key, bool post_ct >> +#ifdef CONFIG_NF_NAT_NEEDED >> + , bool keep_nat_flags >> +#endif >> + ) > > I suggest keeping the argument even for !CONFIG_NF_NAT_NEEDED. This > unclutters the call sites of this function. An ifdef inside the > keep_nat_flags branch should be enough. The compiler will optimize > the code away and it's much prettier to read. > OK. >> { >> const struct nf_conntrack_zone *zone = &nf_ct_zone_dflt; >> enum ip_conntrack_info ctinfo; >> @@ -151,8 +179,20 @@ static void ovs_ct_update_key(const struct sk_buff *skb, >> ct = nf_ct_get(skb, &ctinfo); >> if (ct) { >> state = ovs_ct_get_state(ctinfo); >> + /* OVS persists the related flag for the duration of the >> + * connection. */ >> if (ct->master) >> state |= OVS_CS_F_RELATED; >> +#ifdef CONFIG_NF_NAT_NEEDED >> + if (keep_nat_flags) >> + state |= key->ct.state & OVS_CS_F_NAT_MASK; >> + else { >> + if (ct->status & IPS_SRC_NAT) >> + state |= OVS_CS_F_SRC_NAT; >> + if (ct->status & IPS_DST_NAT) >> + state |= OVS_CS_F_DST_NAT; >> + } >> +#endif >> zone = nf_ct_zone(ct); >> } else if (post_ct) { >> state = OVS_CS_F_TRACKED | OVS_CS_F_INVALID; >> @@ -291,7 +337,16 @@ static int ovs_ct_helper(struct sk_buff *skb, u16 proto) >> return NF_DROP; >> } >> >> - return helper->help(skb, protoff, ct, ctinfo); >> + if (helper->help(skb, protoff, ct, ctinfo) != NF_ACCEPT) >> + return NF_DROP; > > Return the returned value here instead of hardcoding NF_DROP? > OK >> +#ifdef CONFIG_NF_NAT_NEEDED >> + /* Adjust seqs after helper. */ > > A comment on why this is needed would be helpful. > Will add a comment like: /* Needed when a helper adjusts payload size (e.g., FTP PORT command). */ >> + if (test_bit(IPS_SEQ_ADJUST_BIT, &ct->status) >> + && !nf_ct_seq_adjust(skb, ct, ctinfo, protoff)) >> + return NF_DROP; >> +#endif >> + return NF_ACCEPT; > >> @@ -377,7 +432,211 @@ static bool skb_nfct_cached(const struct net *net, const struct sk_buff *skb, >> return true; >> } >> >> -static int __ovs_ct_lookup(struct net *net, const struct sw_flow_key *key, >> +#ifdef CONFIG_NF_NAT_NEEDED >> +/* Modeled after nf_nat_ipv[46]_fn(). >> + * range is only used for new, uninitialized NAT state. >> + * Returns either NF_ACCEPT or NF_DROP. */ >> +static int ovs_ct_nat_execute(struct sk_buff *skb, struct nf_conn *ct, >> + enum ip_conntrack_info ctinfo, >> + const struct nf_nat_range *range, >> + enum nf_nat_manip_type maniptype) >> +{ >> + int hooknum, nh_off, err = NF_ACCEPT; >> + >> + nh_off = skb_network_offset(skb); >> + skb_pull(skb, nh_off); >> + >> + /* See HOOK2MANIP(). */ >> + if (maniptype == NF_NAT_MANIP_SRC) >> + hooknum = NF_INET_LOCAL_IN; /* Source NAT */ >> + else >> + hooknum = NF_INET_LOCAL_OUT; /* Destination NAT */ >> + >> + switch (ctinfo) { >> + case IP_CT_RELATED: >> + case IP_CT_RELATED_REPLY: >> + if (skb->protocol == htons(ETH_P_IP) >> + && ip_hdr(skb)->protocol == IPPROTO_ICMP) { >> + if (!nf_nat_icmp_reply_translation(skb, ct, ctinfo, >> + hooknum)) >> + err = NF_DROP; >> + goto push; >> + } else if (skb->protocol == htons(ETH_P_IPV6)) { >> + __be16 frag_off; >> + u8 nexthdr = ipv6_hdr(skb)->nexthdr; >> + int hdrlen = ipv6_skip_exthdr(skb, >> + sizeof(struct ipv6hdr), >> + &nexthdr, &frag_off); >> + >> + if (hdrlen >= 0 && nexthdr == IPPROTO_ICMPV6) { >> + if (!nf_nat_icmpv6_reply_translation(skb, ct, >> + ctinfo, >> + hooknum, >> + hdrlen)) >> + err = NF_DROP; >> + goto push; >> + } >> + } >> + /* Non-ICMP, fall thru to initialize if needed. */ >> + case IP_CT_NEW: >> + /* Seen it before? This can happen for loopback, retrans, >> + * or local packets. >> + */ >> + if (!nf_nat_initialized(ct, maniptype)) { > > Explicit unlikely()? > Normally initialization is needed, but for IP_CT_RELATED the expectation handling may have already initialized NAT. As such, I do not see a strong case for if (likely(!nf_nat_initialized(ct, maniptype)). I’ll see if I can improve the comment, though, as the current one is copied from nf_nat_ipv[46]_fn(). >> + /* Initialize according to the NAT action. */ >> + err = (range && range->flags & NF_NAT_RANGE_MAP_IPS) >> + /* Action is set up to establish a new >> + * mapping */ >> + ? nf_nat_setup_info(ct, range, maniptype) >> + : nf_nat_alloc_null_binding(ct, hooknum); >> + } >> + break; >> + >> + case IP_CT_ESTABLISHED: >> + case IP_CT_ESTABLISHED_REPLY: >> + break; >> + >> + default: >> + err = NF_DROP; >> + goto push; >> + } >> + >> + if (err == NF_ACCEPT) >> + err = nf_nat_packet(ct, ctinfo, hooknum, skb); > > If you goto push on init failure (IP_CT_NEW branch), then this > conditional is no longer needed and a more straight forward exception > handling is seen. > OK. >> +push: >> + skb_push(skb, nh_off); >> + >> + return err; >> +} > >> +/* Returns NF_DROP if the packet should be dropped, NF_ACCEPT otherwise. >> + * This action can be used to both NAT and reverse NAT, however, reverse NAT >> + * can also be done with the conntrack action. */ >> +static int ovs_ct_nat(struct net *net, struct sw_flow_key *key, >> + const struct ovs_conntrack_info *info, >> + struct sk_buff *skb) >> +{ >> + enum nf_nat_manip_type maniptype; >> + enum ip_conntrack_info ctinfo; >> + struct nf_conn *ct; >> + int err; >> + >> + /* No NAT action or already NATed? */ >> + if (!(info->flags & OVS_CT_F_NAT_MASK) >> + || key->ct.state & OVS_CS_F_NAT_MASK) >> + return NF_ACCEPT; >> + >> + ct = nf_ct_get(skb, &ctinfo); >> + /* Check if an existing conntrack entry may be found for this skb. >> + * This happens when we lose the ct entry pointer due to an upcall. >> + * Don't lookup invalid connections. */ >> + if (!ct && key->ct.state & OVS_CS_F_TRACKED >> + && !(key->ct.state & OVS_CS_F_INVALID)) >> + ct = ovs_ct_find_existing(net, &info->zone, info->family, skb, >> + &ctinfo); >> + if (!ct || nf_ct_is_untracked(ct)) >> + /* A NAT action may only be performed on tracked packets. */ >> + return NF_ACCEPT; > > Braces > Needed due to the comment? >> + /* Add NAT extension if not commited yet. */ >> + if (!nf_ct_is_confirmed(ct)) { >> + if (!nf_ct_nat_ext_add(ct)) >> + return NF_ACCEPT; /* Can't NAT. */ >> + } > > && > Sure. >> + /* Determine NAT type. >> + * Check if the NAT type can be deduced from the tracked connection. >> + * Make sure expected traffic is NATted only when commiting. */ >> + if (info->flags & OVS_CT_F_NAT && ctinfo != IP_CT_NEW >> + && ct->status & IPS_NAT_MASK >> + && (!(ct->status & IPS_EXPECTED_BIT) >> + || info->flags & OVS_CT_F_COMMIT)) { >> + /* NAT an established or related connection like before. */ >> + if (CTINFO2DIR(ctinfo) == IP_CT_DIR_REPLY) >> + /* This is the REPLY direction for a connection >> + * for which NAT was applied in the forward >> + * direction. Do the reverse NAT. */ >> + maniptype = ct->status & IPS_SRC_NAT >> + ? NF_NAT_MANIP_DST : NF_NAT_MANIP_SRC; >> + else >> + maniptype = ct->status & IPS_SRC_NAT >> + ? NF_NAT_MANIP_SRC : NF_NAT_MANIP_DST; >> + } >> + else if (info->flags & OVS_CT_F_SRC_NAT) >> + maniptype = NF_NAT_MANIP_SRC; >> + else if (info->flags & OVS_CT_F_DST_NAT) >> + maniptype = NF_NAT_MANIP_DST; >> + else >> + return NF_ACCEPT; /* Connection is not NATed. */ >> + >> + err = ovs_ct_nat_execute(skb, ct, ctinfo, &info->range, maniptype); >> + >> + /* Mark NAT done if successful. */ >> + if (err == NF_ACCEPT) >> + key->ct.state |= (maniptype == NF_NAT_MANIP_SRC) >> + ? OVS_CS_F_SRC_NAT : OVS_CS_F_DST_NAT; >> + return err; >> +} >> +#endif >> + >> +static int __ovs_ct_lookup(struct net *net, struct sw_flow_key *key, >> const struct ovs_conntrack_info *info, >> struct sk_buff *skb) >> { > > >> @@ -538,6 +819,131 @@ static int ovs_ct_add_helper(struct ovs_conntrack_info *info, const char *name, >> return 0; >> } >> >> +#ifdef CONFIG_NF_NAT_NEEDED >> +static int parse_nat(const struct nlattr *attr, >> + struct ovs_conntrack_info *info, bool log) >> +{ >> + struct nlattr *a; >> + int rem; >> + bool have_ip_max = false; >> + bool have_proto_max = false; >> + bool ip_vers = (info->family == NFPROTO_IPV6); >> + >> + nla_for_each_nested(a, attr, rem) { >> + static const u16 ovs_nat_attr_lens[OVS_NAT_ATTR_MAX + 1][2] = { >> + [OVS_NAT_ATTR_SRC] = {0, 0}, >> + [OVS_NAT_ATTR_DST] = {0, 0}, >> + [OVS_NAT_ATTR_IP_MIN] = {sizeof(struct in_addr), >> + sizeof(struct in6_addr)}, >> + [OVS_NAT_ATTR_IP_MAX] = {sizeof(struct in_addr), >> + sizeof(struct in6_addr)}, >> + [OVS_NAT_ATTR_PROTO_MIN] = {sizeof(u16),sizeof(u16)}, >> + [OVS_NAT_ATTR_PROTO_MAX] = {sizeof(u16),sizeof(u16)}, >> + [OVS_NAT_ATTR_PERSISTENT] = {0, 0}, >> + [OVS_NAT_ATTR_PROTO_HASH] = {0, 0}, >> + [OVS_NAT_ATTR_PROTO_RANDOM] = {0, 0}, >> + }; >> + int type = nla_type(a); >> + >> + if (type > OVS_NAT_ATTR_MAX) { >> + OVS_NLERR(log, "Unknown nat attribute (type=%d, max=%d).\n", >> + type, OVS_NAT_ATTR_MAX); > > Formatting Not readily apparent what you mean here, care to elaborate? > >> + return -EINVAL; >> + } >> + >> + case OVS_NAT_ATTR_IP_MIN: >> + nla_memcpy(&info->range.min_addr, a, nla_len(a)); > > The length attribute should be sizeof of min_addr like for max_addr > below. > Right. >> + info->range.flags |= NF_NAT_RANGE_MAP_IPS; >> + break; >> + >> + case OVS_NAT_ATTR_IP_MAX: >> + have_ip_max = true; >> + nla_memcpy(&info->range.max_addr, a, >> + sizeof(info->range.max_addr)); >> + info->range.flags |= NF_NAT_RANGE_MAP_IPS; >> + break; >> + >> + } > >> static const struct ovs_ct_len_tbl ovs_ct_attr_lens[OVS_CT_ATTR_MAX + 1] = { >> [OVS_CT_ATTR_COMMIT] = { .minlen = 0, >> .maxlen = 0 }, >> @@ -548,7 +954,11 @@ static const struct ovs_ct_len_tbl ovs_ct_attr_lens[OVS_CT_ATTR_MAX + 1] = { >> [OVS_CT_ATTR_LABEL] = { .minlen = sizeof(struct md_label), >> .maxlen = sizeof(struct md_label) }, >> [OVS_CT_ATTR_HELPER] = { .minlen = 1, >> - .maxlen = NF_CT_HELPER_NAME_LEN } >> + .maxlen = NF_CT_HELPER_NAME_LEN }, >> +#ifdef CONFIG_NF_NAT_NEEDED >> + [OVS_CT_ATTR_NAT] = { .minlen = 0, >> + .maxlen = 96 } >> +#endif > > Is the 96 a temporary hack here? > It is not an exact value. It is much better than my temporary hack of 512 was. As trailing garbage is checked for, I’m not sure if this should be very accurately calculated? Maybe it would be better to disable the length checks for this altogether? >> @@ -607,6 +1017,14 @@ static int parse_ct(const struct nlattr *attr, struct ovs_conntrack_info *info, >> return -EINVAL; >> } >> break; >> +#ifdef CONFIG_NF_NAT_NEEDED >> + case OVS_CT_ATTR_NAT: { >> + int err = parse_nat(a, info, log); >> + if (err) >> + return err; >> + break; >> + } >> +#endif > > We should probably bark if user space provides a OVS_CT_ATTR_NAT but the > kernel is compiled without support for it. > We do issue -EINVAL and log “Unknown conntrack attr” in that case. >> +#ifdef CONFIG_NF_NAT_NEEDED >> +static bool ovs_ct_nat_to_attr(const struct ovs_conntrack_info *info, >> + struct sk_buff *skb) >> +{ >> + struct nlattr *start; >> + >> + start = nla_nest_start(skb, OVS_CT_ATTR_NAT); >> + if (!start) >> + return false; >> + >> + if (info->flags & OVS_CT_F_SRC_NAT) { >> + if (nla_put_flag(skb, OVS_NAT_ATTR_SRC)) >> + return false; >> + } else if (info->flags & OVS_CT_F_DST_NAT) { >> + if (nla_put_flag(skb, OVS_NAT_ATTR_DST)) >> + return false; >> + } else { >> + goto out; > > Is the empty nested attribute intended here? Yes. On netlink interface empty nested attribute (NAT without any arguments) signifies to NAT packets of previously NATted connections (only). Jarno -- To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
