Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Matteo Croce <mcroce@redhat.com>
To:     netdev@vger.kernel.org, dev@openvswitch.org
Cc:     linux-kernel@vger.kernel.org, Pravin B Shelar <pshelar@ovn.org>,
        "David S. Miller" <davem@davemloft.net>,
        Bindiya Kurle <bindiyakurle@gmail.com>
Subject: [PATCH net-next] openvswitch: add TTL decrement action
Date:   Tue, 12 Nov 2019 11:25:18 +0100
Message-Id: <20191112102518.4406-1-mcroce@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=WINDOWS-1252
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

New action to decrement TTL instead of setting it to a fixed value.
This action will decrement the TTL and, in case of expired TTL, send the
packet to userspace via output_userspace() to take care of it.

Supports both IPv4 and IPv6 via the ttl and hop_limit fields, respectively.

Tested with a corresponding change in the userspace:

    # ovs-dpctl dump-flows
    in_port(2),eth(),eth_type(0x0800), packets:0, bytes:0, used:never, acti=
ons:dec_ttl,1
    in_port(1),eth(),eth_type(0x0800), packets:0, bytes:0, used:never, acti=
ons:dec_ttl,2
    in_port(1),eth(),eth_type(0x0806), packets:0, bytes:0, used:never, acti=
ons:2
    in_port(2),eth(),eth_type(0x0806), packets:0, bytes:0, used:never, acti=
ons:1

    # ping -c1 192.168.0.2 -t 42
    IP (tos 0x0, ttl 41, id 61647, offset 0, flags [DF], proto ICMP (1), le=
ngth 84)
        192.168.0.1 > 192.168.0.2: ICMP echo request, id 386, seq 1, length=
 64
    # ping -c1 192.168.0.2 -t 120
    IP (tos 0x0, ttl 119, id 62070, offset 0, flags [DF], proto ICMP (1), l=
ength 84)
        192.168.0.1 > 192.168.0.2: ICMP echo request, id 388, seq 1, length=
 64
    # ping -c1 192.168.0.2 -t 1
    #

Co-authored-by: Bindiya Kurle <bindiyakurle@gmail.com>
Signed-off-by: Bindiya Kurle <bindiyakurle@gmail.com>
Signed-off-by: Matteo Croce <mcroce@redhat.com>
---
 include/uapi/linux/openvswitch.h |  2 ++
 net/openvswitch/actions.c        | 46 ++++++++++++++++++++++++++++++++
 net/openvswitch/flow_netlink.c   |  6 +++++
 3 files changed, 54 insertions(+)

diff --git a/include/uapi/linux/openvswitch.h b/include/uapi/linux/openvswi=
tch.h
index 1887a451c388..a3bdb1ecd1e7 100644
--- a/include/uapi/linux/openvswitch.h
+++ b/include/uapi/linux/openvswitch.h
@@ -890,6 +890,7 @@ struct check_pkt_len_arg {
  * @OVS_ACTION_ATTR_CHECK_PKT_LEN: Check the packet length and execute a s=
et
  * of actions if greater than the specified packet length, else execute
  * another set of actions.
+ * @OVS_ACTION_ATTR_DEC_TTL: Decrement the IP TTL.
  *
  * Only a single header can be set with a single %OVS_ACTION_ATTR_SET.  No=
t all
  * fields within a header are modifiable, e.g. the IPv4 protocol and fragm=
ent
@@ -925,6 +926,7 @@ enum ovs_action_attr {
 =09OVS_ACTION_ATTR_METER,        /* u32 meter ID. */
 =09OVS_ACTION_ATTR_CLONE,        /* Nested OVS_CLONE_ATTR_*.  */
 =09OVS_ACTION_ATTR_CHECK_PKT_LEN, /* Nested OVS_CHECK_PKT_LEN_ATTR_*. */
+=09OVS_ACTION_ATTR_DEC_TTL,      /* Decrement ttl action */
=20
 =09__OVS_ACTION_ATTR_MAX,=09      /* Nothing past this will be accepted
 =09=09=09=09       * from userspace. */
diff --git a/net/openvswitch/actions.c b/net/openvswitch/actions.c
index 12936c151cc0..077b7f309c93 100644
--- a/net/openvswitch/actions.c
+++ b/net/openvswitch/actions.c
@@ -1174,6 +1174,43 @@ static int execute_check_pkt_len(struct datapath *dp=
, struct sk_buff *skb,
 =09=09=09     nla_len(actions), last, clone_flow_key);
 }
=20
+static int execute_dec_ttl(struct sk_buff *skb, struct sw_flow_key *key)
+{
+=09int err;
+
+=09if (skb->protocol =3D=3D htons(ETH_P_IPV6)) {
+=09=09struct ipv6hdr *nh =3D ipv6_hdr(skb);
+
+=09=09err =3D skb_ensure_writable(skb, skb_network_offset(skb) +
+=09=09=09=09=09  sizeof(*nh));
+=09=09if (unlikely(err))
+=09=09=09return err;
+
+=09=09if (nh->hop_limit <=3D 1)
+=09=09=09return -EHOSTUNREACH;
+
+=09=09key->ip.ttl =3D --nh->hop_limit;
+=09} else {
+=09=09struct iphdr *nh =3D ip_hdr(skb);
+=09=09u8 old_ttl;
+
+=09=09err =3D skb_ensure_writable(skb, skb_network_offset(skb) +
+=09=09=09=09=09  sizeof(*nh));
+=09=09if (unlikely(err))
+=09=09=09return err;
+
+=09=09if (nh->ttl <=3D 1)
+=09=09=09return -EHOSTUNREACH;
+
+=09=09old_ttl =3D nh->ttl--;
+=09=09csum_replace2(&nh->check, htons(old_ttl << 8),
+=09=09=09      htons(nh->ttl << 8));
+=09=09key->ip.ttl =3D nh->ttl;
+=09}
+
+=09return 0;
+}
+
 /* Execute a list of actions against 'skb'. */
 static int do_execute_actions(struct datapath *dp, struct sk_buff *skb,
 =09=09=09      struct sw_flow_key *key,
@@ -1345,6 +1382,15 @@ static int do_execute_actions(struct datapath *dp, s=
truct sk_buff *skb,
=20
 =09=09=09break;
 =09=09}
+
+=09=09case OVS_ACTION_ATTR_DEC_TTL:
+=09=09=09err =3D execute_dec_ttl(skb, key);
+=09=09=09if (err =3D=3D -EHOSTUNREACH) {
+=09=09=09=09output_userspace(dp, skb, key, a, attr,
+=09=09=09=09=09=09 len, OVS_CB(skb)->cutlen);
+=09=09=09=09OVS_CB(skb)->cutlen =3D 0;
+=09=09=09}
+=09=09=09break;
 =09=09}
=20
 =09=09if (unlikely(err)) {
diff --git a/net/openvswitch/flow_netlink.c b/net/openvswitch/flow_netlink.=
c
index 65c2e3458ff5..d17f2d4b420f 100644
--- a/net/openvswitch/flow_netlink.c
+++ b/net/openvswitch/flow_netlink.c
@@ -79,6 +79,7 @@ static bool actions_may_change_flow(const struct nlattr *=
actions)
 =09=09case OVS_ACTION_ATTR_SET_MASKED:
 =09=09case OVS_ACTION_ATTR_METER:
 =09=09case OVS_ACTION_ATTR_CHECK_PKT_LEN:
+=09=09case OVS_ACTION_ATTR_DEC_TTL:
 =09=09default:
 =09=09=09return true;
 =09=09}
@@ -3005,6 +3006,7 @@ static int __ovs_nla_copy_actions(struct net *net, co=
nst struct nlattr *attr,
 =09=09=09[OVS_ACTION_ATTR_METER] =3D sizeof(u32),
 =09=09=09[OVS_ACTION_ATTR_CLONE] =3D (u32)-1,
 =09=09=09[OVS_ACTION_ATTR_CHECK_PKT_LEN] =3D (u32)-1,
+=09=09=09[OVS_ACTION_ATTR_DEC_TTL] =3D 0,
 =09=09};
 =09=09const struct ovs_action_push_vlan *vlan;
 =09=09int type =3D nla_type(a);
@@ -3233,6 +3235,10 @@ static int __ovs_nla_copy_actions(struct net *net, c=
onst struct nlattr *attr,
 =09=09=09break;
 =09=09}
=20
+=09=09case OVS_ACTION_ATTR_DEC_TTL:
+=09=09=09/* Nothing to do.  */
+=09=09=09break;
+
 =09=09default:
 =09=09=09OVS_NLERR(log, "Unknown Action type %d", type);
 =09=09=09return -EINVAL;
--=20
2.23.0