[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CADvbK_e+tgefsiB1N-7CHUR35P-sDfaOqRVp281VhrQO2ot_hQ@mail.gmail.com>
Date: Wed, 23 Nov 2022 13:54:41 -0500
From: Xin Long <lucien.xin@...il.com>
To: Marcelo Ricardo Leitner <marcelo.leitner@...il.com>
Cc: dev@...nvswitch.org, ovs-dev@...nvswitch.org,
Davide Caratti <dcaratti@...hat.com>,
Jiri Pirko <jiri@...nulli.us>,
network dev <netdev@...r.kernel.org>,
Paul Blakey <paulb@...dia.com>,
Florian Westphal <fw@...len.de>,
Jamal Hadi Salim <jhs@...atatu.com>,
Ilya Maximets <i.maximets@....org>,
Eric Dumazet <edumazet@...gle.com>,
Cong Wang <xiyou.wangcong@...il.com>, kuba@...nel.org,
Paolo Abeni <pabeni@...hat.com>, davem@...emloft.net,
Pablo Neira Ayuso <pablo@...filter.org>
Subject: Re: [ovs-dev] [PATCHv2 net-next 5/5] net: move the nat function to
nf_nat_ovs for ovs and tc
On Wed, Nov 23, 2022 at 1:48 PM Marcelo Ricardo Leitner
<marcelo.leitner@...il.com> wrote:
>
> On Wed, Nov 23, 2022 at 12:31:38PM -0500, Xin Long wrote:
> > On Wed, Nov 23, 2022 at 10:13 AM Marcelo Ricardo Leitner
> > <marcelo.leitner@...il.com> wrote:
> > >
> > > On Wed, Nov 23, 2022 at 12:09:55PM -0300, Marcelo Ricardo Leitner wrote:
> > > > On Tue, Nov 22, 2022 at 12:32:21PM -0500, Xin Long wrote:
> > > > > +int nf_ct_nat(struct sk_buff *skb, struct nf_conn *ct,
> > > > > + enum ip_conntrack_info ctinfo, int *action,
> > > > > + const struct nf_nat_range2 *range, bool commit)
> > > > > +{
> > > > > + enum nf_nat_manip_type maniptype;
> > > > > + int err, ct_action = *action;
> > > > > +
> > > > > + *action = 0;
> > > > > +
> > > > > + /* Add NAT extension if not confirmed yet. */
> > > > > + if (!nf_ct_is_confirmed(ct) && !nf_ct_nat_ext_add(ct))
> > > > > + return NF_ACCEPT; /* Can't NAT. */
> > > > > +
> > > > > + if (ctinfo != IP_CT_NEW && (ct->status & IPS_NAT_MASK) &&
> > > > > + (ctinfo != IP_CT_RELATED || 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 (ct_action & (1 << NF_NAT_MANIP_SRC)) {
> > > > > + maniptype = NF_NAT_MANIP_SRC;
> > > > > + } else if (ct_action & (1 << NF_NAT_MANIP_DST)) {
> > > > > + maniptype = NF_NAT_MANIP_DST;
> > > > > + } else {
> > > > > + return NF_ACCEPT;
> > > > > + }
> > > > > +
> > > > > + err = nf_ct_nat_execute(skb, ct, ctinfo, action, range, maniptype);
> > > > > + if (err == NF_ACCEPT && ct->status & IPS_DST_NAT) {
> > > > > + if (ct->status & IPS_SRC_NAT) {
> > > > > + if (maniptype == NF_NAT_MANIP_SRC)
> > > > > + maniptype = NF_NAT_MANIP_DST;
> > > > > + else
> > > > > + maniptype = NF_NAT_MANIP_SRC;
> > > > > +
> > > > > + err = nf_ct_nat_execute(skb, ct, ctinfo, action, range,
> > > > > + maniptype);
> > > > > + } else if (CTINFO2DIR(ctinfo) == IP_CT_DIR_ORIGINAL) {
> > > > > + err = nf_ct_nat_execute(skb, ct, ctinfo, action, NULL,
> > > > > + NF_NAT_MANIP_SRC);
> > > > > + }
> > > > > + }
> > > > > + return err;
> > > > > +}
> > > > > +EXPORT_SYMBOL_GPL(nf_ct_nat);
> > > > > diff --git a/net/openvswitch/conntrack.c b/net/openvswitch/conntrack.c
> > > > > index cc643a556ea1..d03c75165663 100644
> > > > > --- a/net/openvswitch/conntrack.c
> > > > > +++ b/net/openvswitch/conntrack.c
> > > > > @@ -726,144 +726,27 @@ static void ovs_nat_update_key(struct sw_flow_key *key,
> > > > > }
> > > > > }
> > > > >
> > > > > -/* Modelled 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_range2 *range,
> > > > > - enum nf_nat_manip_type maniptype, struct sw_flow_key *key)
> > > > > -{
> > > > > - int hooknum, err = NF_ACCEPT;
> > > > > -
> > > > > - /* 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 (IS_ENABLED(CONFIG_NF_NAT) &&
> > > > > - 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 out;
> > > > > - } else if (IS_ENABLED(CONFIG_IPV6) &&
> > > > > - 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 out;
> > > > > - }
> > > > > - }
> > > > > - /* Non-ICMP, fall thru to initialize if needed. */
> > > > > - fallthrough;
> > > > > - case IP_CT_NEW:
> > > > > - /* Seen it before? This can happen for loopback, retrans,
> > > > > - * or local packets.
> > > > > - */
> > > > > - if (!nf_nat_initialized(ct, maniptype)) {
> > > > > - /* 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);
> > > > > - if (err != NF_ACCEPT)
> > > > > - goto out;
> > > > > - }
> > > > > - break;
> > > > > -
> > > > > - case IP_CT_ESTABLISHED:
> > > > > - case IP_CT_ESTABLISHED_REPLY:
> > > > > - break;
> > > > > -
> > > > > - default:
> > > > > - err = NF_DROP;
> > > > > - goto out;
> > > > > - }
> > > > > -
> > > > > - err = nf_nat_packet(ct, ctinfo, hooknum, skb);
> > > > > -out:
> > > > > - /* Update the flow key if NAT successful. */
> > > > > - if (err == NF_ACCEPT)
> > > > > - ovs_nat_update_key(key, skb, maniptype);
> > > > > -
> > > > > - return err;
> > > > > -}
> > > > > -
> > > > > /* Returns NF_DROP if the packet should be dropped, NF_ACCEPT otherwise. */
> > > > > static int ovs_ct_nat(struct net *net, struct sw_flow_key *key,
> > > > > const struct ovs_conntrack_info *info,
> > > > > struct sk_buff *skb, struct nf_conn *ct,
> > > > > enum ip_conntrack_info ctinfo)
> > > > > {
> > > > > - enum nf_nat_manip_type maniptype;
> > > > > - int err;
> > > > > + int err, action = 0;
> > > > >
> > > > > if (!(info->nat & OVS_CT_NAT))
> > > > > return NF_ACCEPT;
> > > > > + if (info->nat & OVS_CT_SRC_NAT)
> > > > > + action |= (1 << NF_NAT_MANIP_SRC);
> > > > > + if (info->nat & OVS_CT_DST_NAT)
> > > > > + action |= (1 << NF_NAT_MANIP_DST);
> > > >
> > > > I'm wondering why this dance at this level with supporting multiple
> > > > MANIPs while actually only one can be used at a time.
> > > >
> > > > act_ct will reject an action using both:
> > > > if ((p->ct_action & TCA_CT_ACT_NAT_SRC) &&
> > > > (p->ct_action & TCA_CT_ACT_NAT_DST)) {
> > > > NL_SET_ERR_MSG_MOD(extack, "dnat and snat can't be enabled at the same time");
> > > > return -EOPNOTSUPP;
> > > >
> > > > I couldn't find this kind of check in ovs code right now (didn't look much, I
> > > > confess :)), but even the code here was already doing:
> > > >
> > > > - } else if (info->nat & OVS_CT_SRC_NAT) {
> > > > - maniptype = NF_NAT_MANIP_SRC;
> > > > - } else if (info->nat & OVS_CT_DST_NAT) {
> > > > - maniptype = NF_NAT_MANIP_DST;
> > > >
> > > > And in case of tuple conflict, maniptype will be forcibly updated in
> > > > [*] below.
> > >
> > > Ahh.. just found why.. in case of typle conflict, nf_ct_nat() now may
> > > set both.
> > Right.
> > BTW. The tuple conflict processing actually has provided the
> > code to do full nat (snat and dnat at the same time), which I
> > think is more efficient than doing full nat in two zones. All
> > we have to do is adjust a few lines of code for that.
>
> In this part, yes. But it needs some surgery all around for full
> support. The code in ovs kernel for using only one type starts here:
>
> static int parse_nat(const struct nlattr *attr,
> struct ovs_conntrack_info *info, bool log)
> {
> ...
> switch (type) {
> case OVS_NAT_ATTR_SRC:
> case OVS_NAT_ATTR_DST:
> if (info->nat) {
> OVS_NLERR(log, "Only one type of NAT may be specified");
> return -ERANGE;
> }
> ...
> }
>
> So vswitchd also doesn't support it. And then tc, iproute and drivers
> that offload it as well. Not sure if it has impacts on openflow too.
>
not in one single action, but two actions:
"table=1, in_port=veth1,tcp,tcp_dst=2121,ct_state=+trk+new
actions=ct(nat(dst=7.7.16.3)),ct(commit, nat(src=7.7.16.1),
alg=ftp),veth2"
as long as it allows the 1st one doesn't commit, which is a simple
check in parse_nat().
I tested it, TC already supports it. I'm not sure about drivers, but I
think it should be, as with two actions.
Powered by blists - more mailing lists