[ovs-dev] [PATCH v2 2/7] nsh: userland support for network service headers

Pritesh Kothari pritesh.kothari at cisco.com
Tue Feb 25 23:44:15 UTC 2014


NSH service path (nsp) can be set/unset while creating the port
as well as nsp can be matched on incoming packets.

Signed-off-by: Pritesh Kothari <pritesh.kothari at cisco.com>

diff --git a/lib/flow.c b/lib/flow.c
index e7fe4d3..28e96ec 100644
--- a/lib/flow.c
+++ b/lib/flow.c
@@ -563,6 +563,8 @@ flow_tun_flag_to_string(uint32_t flags)
         return "csum";
     case FLOW_TNL_F_KEY:
         return "key";
+    case FLOW_TNL_F_NSP:
+        return "nsp";
     default:
         return NULL;
     }
diff --git a/lib/flow.h b/lib/flow.h
index 3109a84..6086904 100644
--- a/lib/flow.h
+++ b/lib/flow.h
@@ -56,11 +56,13 @@ BUILD_ASSERT_DECL(FLOW_NW_FRAG_LATER == NX_IP_FRAG_LATER);
 #define FLOW_TNL_F_DONT_FRAGMENT (1 << 0)
 #define FLOW_TNL_F_CSUM (1 << 1)
 #define FLOW_TNL_F_KEY (1 << 2)
+#define FLOW_TNL_F_NSP (1 << 3)
 
 const char *flow_tun_flag_to_string(uint32_t flags);
 
 struct flow_tnl {
     ovs_be64 tun_id;
+    ovs_be32 nsp;
     ovs_be32 ip_src;
     ovs_be32 ip_dst;
     uint16_t flags;
diff --git a/lib/match.c b/lib/match.c
index b2a25fd..091ada5 100644
--- a/lib/match.c
+++ b/lib/match.c
@@ -70,6 +70,9 @@ match_wc_init(struct match *match, const struct flow *flow)
         if (flow->tunnel.flags & FLOW_TNL_F_KEY) {
             memset(&wc->masks.tunnel.tun_id, 0xff, sizeof wc->masks.tunnel.tun_id);
         }
+        if (flow->tunnel.flags & FLOW_TNL_F_NSP) {
+            memset(&wc->masks.tunnel.nsp, 0xff, sizeof wc->masks.tunnel.nsp);
+        }
         memset(&wc->masks.tunnel.ip_src, 0xff, sizeof wc->masks.tunnel.ip_src);
         memset(&wc->masks.tunnel.ip_dst, 0xff, sizeof wc->masks.tunnel.ip_dst);
         memset(&wc->masks.tunnel.flags, 0xff, sizeof wc->masks.tunnel.flags);
@@ -79,6 +82,10 @@ match_wc_init(struct match *match, const struct flow *flow)
         memset(&wc->masks.tunnel.tun_id, 0xff, sizeof wc->masks.tunnel.tun_id);
     }
 
+    if (flow->tunnel.nsp) {
+        memset(&wc->masks.tunnel.nsp, 0xff, sizeof wc->masks.tunnel.nsp);
+    }
+
     memset(&wc->masks.metadata, 0xff, sizeof wc->masks.metadata);
     memset(&wc->masks.in_port, 0xff, sizeof wc->masks.in_port);
     memset(&wc->masks.vlan_tci, 0xff, sizeof wc->masks.vlan_tci);
@@ -751,6 +758,19 @@ match_set_nd_target_masked(struct match *match,
     match->wc.masks.nd_target = *mask;
 }
 
+void
+match_set_nsp_masked(struct match *match, ovs_be32 nsp, ovs_be32 mask)
+{
+    match->wc.masks.tunnel.nsp = mask;
+    match->flow.tunnel.nsp = nsp & mask;
+}
+
+void
+match_set_nsp(struct match *match, ovs_be32 nsp)
+{
+    match_set_nsp_masked(match, nsp, htonl(UINT32_MAX));
+}
+
 /* Returns true if 'a' and 'b' wildcard the same fields and have the same
  * values for fixed fields, otherwise false. */
 bool
@@ -867,6 +887,7 @@ format_flow_tunnel(struct ds *s, const struct match *match)
     const struct flow_tnl *tnl = &match->flow.tunnel;
 
     format_be64_masked(s, "tun_id", tnl->tun_id, wc->masks.tunnel.tun_id);
+    format_uint32_masked(s, "nsp", tnl->nsp, wc->masks.tunnel.nsp);
     format_ip_netmask(s, "tun_src", tnl->ip_src, wc->masks.tunnel.ip_src);
     format_ip_netmask(s, "tun_dst", tnl->ip_dst, wc->masks.tunnel.ip_dst);
 
diff --git a/lib/match.h b/lib/match.h
index 7a8ae68..1c8dd15 100644
--- a/lib/match.h
+++ b/lib/match.h
@@ -124,6 +124,8 @@ void match_set_ipv6_label_masked(struct match *, ovs_be32, ovs_be32);
 void match_set_nd_target(struct match *, const struct in6_addr *);
 void match_set_nd_target_masked(struct match *, const struct in6_addr *,
                                 const struct in6_addr *);
+void match_set_nsp(struct match *, ovs_be32 nsp);
+void match_set_nsp_masked(struct match *, ovs_be32 nsp, ovs_be32 mask);
 
 bool match_equal(const struct match *, const struct match *);
 uint32_t match_hash(const struct match *, uint32_t basis);
diff --git a/lib/meta-flow.c b/lib/meta-flow.c
index 3afcd4c..5a620a1 100644
--- a/lib/meta-flow.c
+++ b/lib/meta-flow.c
@@ -763,7 +763,25 @@ const struct mf_field mf_fields[MFF_N_IDS] = {
         OFPUTIL_P_NXM_OXM_ANY,
         OFPUTIL_P_NXM_OXM_ANY,
         -1,
-    }
+    },
+
+    /* ## ---- ## */
+    /* ## L"S" ## */
+    /* ## ---- ## */
+
+    {
+        MFF_NSP, "nsp", NULL,
+        sizeof(ovs_be32), 24,
+        MFM_FULLY,
+        MFS_HEXADECIMAL,
+        MFP_NONE,
+        false,
+        0, NULL,
+        0, NULL,
+        OFPUTIL_P_NONE,
+        OFPUTIL_P_NONE,
+        -1,
+    },
 };
 
 /* Maps an NXM or OXM header value to an mf_field. */
@@ -888,6 +906,8 @@ mf_is_all_wild(const struct mf_field *mf, const struct flow_wildcards *wc)
     case MFF_TUN_TTL:
     case MFF_TUN_FLAGS:
         return !wc->masks.tunnel.tun_id;
+    case MFF_NSP:
+        return !wc->masks.tunnel.nsp;
     case MFF_METADATA:
         return !wc->masks.metadata;
     case MFF_IN_PORT:
@@ -1162,6 +1182,7 @@ mf_is_value_valid(const struct mf_field *mf, const union mf_value *value)
     case MFF_ND_TARGET:
     case MFF_ND_SLL:
     case MFF_ND_TLL:
+    case MFF_NSP:
         return true;
 
     case MFF_IN_PORT_OXM: {
@@ -1396,6 +1417,10 @@ mf_get_value(const struct mf_field *mf, const struct flow *flow,
         value->ipv6 = flow->nd_target;
         break;
 
+    case MFF_NSP:
+        value->be32 = flow->tunnel.nsp;
+        break;
+
     case MFF_N_IDS:
     default:
         OVS_NOT_REACHED();
@@ -1593,6 +1618,10 @@ mf_set_value(const struct mf_field *mf,
         match_set_nd_target(match, &value->ipv6);
         break;
 
+    case MFF_NSP:
+        match_set_nsp(match, value->be32);
+        break;
+
     case MFF_N_IDS:
     default:
         OVS_NOT_REACHED();
@@ -1810,6 +1839,10 @@ mf_set_flow_value(const struct mf_field *mf,
         flow->nd_target = value->ipv6;
         break;
 
+    case MFF_NSP:
+        flow->tunnel.nsp = value->be32;
+        break;
+
     case MFF_N_IDS:
     default:
         OVS_NOT_REACHED();
@@ -2019,6 +2052,10 @@ mf_set_wild(const struct mf_field *mf, struct match *match)
         memset(&match->flow.nd_target, 0, sizeof match->flow.nd_target);
         break;
 
+    case MFF_NSP:
+        match_set_nsp_masked(match, htonl(0), htonl(0));
+        break;
+
     case MFF_N_IDS:
     default:
         OVS_NOT_REACHED();
@@ -2187,6 +2224,10 @@ mf_set(const struct mf_field *mf,
         match_set_tcp_flags_masked(match, value->be16, mask->be16);
         break;
 
+    case MFF_NSP:
+        match_set_nsp_masked(match, value->be32, mask->be32);
+        break;
+
     case MFF_N_IDS:
     default:
         OVS_NOT_REACHED();
diff --git a/lib/meta-flow.h b/lib/meta-flow.h
index cf92556..a2977e4 100644
--- a/lib/meta-flow.h
+++ b/lib/meta-flow.h
@@ -139,6 +139,9 @@ enum OVS_PACKED_ENUM mf_field_id {
     MFF_ND_SLL,                 /* mac */
     MFF_ND_TLL,                 /* mac */
 
+    /* Network Service Headers (NSH) Fields */
+    MFF_NSP,                    /* be32 */
+
     MFF_N_IDS
 };
 
diff --git a/lib/netdev-vport.c b/lib/netdev-vport.c
index 165c1c6..ce33150 100644
--- a/lib/netdev-vport.c
+++ b/lib/netdev-vport.c
@@ -45,6 +45,7 @@ VLOG_DEFINE_THIS_MODULE(netdev_vport);
 
 #define VXLAN_DST_PORT 4789
 #define LISP_DST_PORT 4341
+#define NSH_DST_PORT 6633
 
 #define DEFAULT_TTL 64
 
@@ -311,6 +312,33 @@ parse_key(const struct smap *args, const char *name,
     }
 }
 
+static ovs_be32
+parse_nsp(const struct smap *args, const char *name,
+          bool *present, bool *flow)
+{
+    const char *s;
+
+    *present = false;
+    *flow = false;
+
+    s = smap_get(args, name);
+    if (!s) {
+        s = smap_get(args, "nsp");
+        if (!s) {
+            return 0;
+        }
+    }
+
+    *present = true;
+
+    if (!strcmp(s, "flow")) {
+        *flow = true;
+        return 0;
+    } else {
+        return htonl(strtoul(s, NULL, 0));
+    }
+}
+
 static int
 set_tunnel_config(struct netdev *dev_, const struct smap *args)
 {
@@ -414,6 +442,10 @@ set_tunnel_config(struct netdev *dev_, const struct smap *args)
                    !strcmp(node->key, "in_key") ||
                    !strcmp(node->key, "out_key")) {
             /* Handled separately below. */
+        } else if (!strcmp(node->key, "nsp") ||
+                   !strcmp(node->key, "in_nsp") ||
+                   !strcmp(node->key, "out_nsp")) {
+            /* Handled separately below. */
         } else {
             VLOG_WARN("%s: unknown %s argument '%s'", name, type, node->key);
         }
@@ -482,6 +514,16 @@ set_tunnel_config(struct netdev *dev_, const struct smap *args)
                                &tnl_cfg.out_key_present,
                                &tnl_cfg.out_key_flow);
 
+    if (tnl_cfg.dst_port == htons(NSH_DST_PORT)) {
+        tnl_cfg.in_nsp = parse_nsp(args, "in_nsp",
+                                   &tnl_cfg.in_nsp_present,
+                                   &tnl_cfg.in_nsp_flow);
+
+        tnl_cfg.out_nsp = parse_nsp(args, "out_nsp",
+                                    &tnl_cfg.out_nsp_present,
+                                    &tnl_cfg.out_nsp_flow);
+    }
+
     ovs_mutex_lock(&dev->mutex);
     dev->tnl_cfg = tnl_cfg;
     seq_change(connectivity_seq_get());
@@ -495,6 +537,7 @@ get_tunnel_config(const struct netdev *dev, struct smap *args)
 {
     struct netdev_vport *netdev = netdev_vport_cast(dev);
     struct netdev_tunnel_config tnl_cfg;
+    const char *type = netdev_get_type(dev);
 
     ovs_mutex_lock(&netdev->mutex);
     tnl_cfg = netdev->tnl_cfg;
@@ -533,6 +576,27 @@ get_tunnel_config(const struct netdev *dev, struct smap *args)
         }
     }
 
+    if (tnl_cfg.in_nsp_flow && tnl_cfg.out_nsp_flow) {
+        smap_add(args, "nsp", "flow");
+    } else if (tnl_cfg.in_nsp_present && tnl_cfg.out_nsp_present
+               && tnl_cfg.in_nsp == tnl_cfg.out_nsp) {
+        smap_add_format(args, "nsp", "%#"PRIx32, ntohl(tnl_cfg.in_nsp));
+    } else {
+        if (tnl_cfg.in_nsp_flow) {
+            smap_add(args, "in_nsp", "flow");
+        } else if (tnl_cfg.in_nsp_present) {
+            smap_add_format(args, "in_nsp", "%#"PRIx32,
+                            ntohl(tnl_cfg.in_nsp));
+        }
+
+        if (tnl_cfg.out_nsp_flow) {
+            smap_add(args, "out_nsp", "flow");
+        } else if (tnl_cfg.out_nsp_present) {
+            smap_add_format(args, "out_nsp", "%#"PRIx32,
+                            ntohl(tnl_cfg.out_nsp));
+        }
+    }
+
     if (tnl_cfg.ttl_inherit) {
         smap_add(args, "ttl", "inherit");
     } else if (tnl_cfg.ttl != DEFAULT_TTL) {
@@ -547,7 +611,6 @@ get_tunnel_config(const struct netdev *dev, struct smap *args)
 
     if (tnl_cfg.dst_port) {
         uint16_t dst_port = ntohs(tnl_cfg.dst_port);
-        const char *type = netdev_get_type(dev);
 
         if ((!strcmp("vxlan", type) && dst_port != VXLAN_DST_PORT) ||
             (!strcmp("lisp", type) && dst_port != LISP_DST_PORT)) {
diff --git a/lib/netdev.h b/lib/netdev.h
index 410c35b..ba99062 100644
--- a/lib/netdev.h
+++ b/lib/netdev.h
@@ -102,6 +102,14 @@ struct netdev_stats {
 
 /* Configuration specific to tunnels. */
 struct netdev_tunnel_config {
+    bool in_nsp_present;
+    bool in_nsp_flow;
+    ovs_be32 in_nsp;            /* incoming NSH service path */
+
+    bool out_nsp_present;
+    bool out_nsp_flow;
+    ovs_be32 out_nsp;           /* outgoing NSH service path */
+
     bool in_key_present;
     bool in_key_flow;
     ovs_be64 in_key;
diff --git a/lib/odp-util.c b/lib/odp-util.c
index e20564f..8cfcca5 100644
--- a/lib/odp-util.c
+++ b/lib/odp-util.c
@@ -801,6 +801,7 @@ tunnel_key_attr_len(int type)
     case OVS_TUNNEL_KEY_ATTR_TTL: return 1;
     case OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT: return 0;
     case OVS_TUNNEL_KEY_ATTR_CSUM: return 0;
+    case OVS_TUNNEL_KEY_ATTR_NSP: return 4;
     case __OVS_TUNNEL_KEY_ATTR_MAX:
         return -1;
     }
@@ -848,6 +849,10 @@ odp_tun_key_from_attr(const struct nlattr *attr, struct flow_tnl *tun)
         case OVS_TUNNEL_KEY_ATTR_CSUM:
             tun->flags |= FLOW_TNL_F_CSUM;
             break;
+        case OVS_TUNNEL_KEY_ATTR_NSP:
+            tun->nsp = nl_attr_get_be32(a);
+            tun->flags |= FLOW_TNL_F_NSP;
+            break;
         default:
             /* Allow this to show up as unexpected, if there are unknown
              * tunnel attribute, eventually resulting in ODP_FIT_TOO_MUCH. */
@@ -891,6 +896,9 @@ tun_key_to_attr(struct ofpbuf *a, const struct flow_tnl *tun_key)
     if (tun_key->flags & FLOW_TNL_F_CSUM) {
         nl_msg_put_flag(a, OVS_TUNNEL_KEY_ATTR_CSUM);
     }
+    if (tun_key->flags & FLOW_TNL_F_NSP) {
+        nl_msg_put_be32(a, OVS_TUNNEL_KEY_ATTR_NSP, tun_key->nsp);
+    }
 
     nl_msg_end_nested(a, tun_key_ofs);
 }
@@ -917,6 +925,7 @@ odp_mask_attr_is_exact(const struct nlattr *ma)
         odp_tun_key_from_attr(ma, &tun_mask);
         if (tun_mask.flags == (FLOW_TNL_F_KEY
                                | FLOW_TNL_F_DONT_FRAGMENT
+                               | FLOW_TNL_F_NSP
                                | FLOW_TNL_F_CSUM)) {
             /* The flags are exact match, check the remaining fields. */
             tun_mask.flags = 0xffff;
@@ -1041,10 +1050,12 @@ format_odp_key_attr(const struct nlattr *a, const struct nlattr *ma,
             memset(&tun_mask, 0, sizeof tun_mask);
             odp_tun_key_from_attr(ma, &tun_mask);
             ds_put_format(ds, "tun_id=%#"PRIx64"/%#"PRIx64
+                          ",nsp=%#"PRIx32"/%#"PRIx32
                           ",src="IP_FMT"/"IP_FMT",dst="IP_FMT"/"IP_FMT
                           ",tos=%#"PRIx8"/%#"PRIx8",ttl=%"PRIu8"/%#"PRIx8
                           ",flags(",
                           ntohll(tun_key.tun_id), ntohll(tun_mask.tun_id),
+                          ntohl(tun_key.nsp), ntohl(tun_mask.nsp),
                           IP_ARGS(tun_key.ip_src), IP_ARGS(tun_mask.ip_src),
                           IP_ARGS(tun_key.ip_dst), IP_ARGS(tun_mask.ip_dst),
                           tun_key.ip_tos, tun_mask.ip_tos,
@@ -1060,9 +1071,11 @@ format_odp_key_attr(const struct nlattr *a, const struct nlattr *ma,
             */
             ds_put_char(ds, ')');
         } else {
-            ds_put_format(ds, "tun_id=0x%"PRIx64",src="IP_FMT",dst="IP_FMT","
+            ds_put_format(ds, "tun_id=0x%"PRIx64",nsp=0x%"PRIx32
+                          ",src="IP_FMT",dst="IP_FMT","
                           "tos=0x%"PRIx8",ttl=%"PRIu8",flags(",
                           ntohll(tun_key.tun_id),
+                          ntohl(tun_key.nsp),
                           IP_ARGS(tun_key.ip_src),
                           IP_ARGS(tun_key.ip_dst),
                           tun_key.ip_tos, tun_key.ip_ttl);
@@ -1621,13 +1634,16 @@ parse_odp_key_mask_attr(const char *s, const struct simap *port_names,
     {
         uint64_t tun_id, tun_id_mask;
         struct flow_tnl tun_key, tun_key_mask;
+        uint32_t nsp, nsp_mask;
         int n = -1;
 
         if (mask && ovs_scan(s, "tunnel(tun_id=%"SCNi64"/%"SCNi64","
+                             "nsp=%"PRIx32"/%"PRIx32","
                              "src="IP_SCAN_FMT"/"IP_SCAN_FMT",dst="IP_SCAN_FMT
                              "/"IP_SCAN_FMT",tos=%"SCNi8"/%"SCNi8","
                              "ttl=%"SCNi8"/%"SCNi8",flags%n",
                              &tun_id, &tun_id_mask,
+                             &nsp, &nsp_mask,
                              IP_SCAN_ARGS(&tun_key.ip_src),
                              IP_SCAN_ARGS(&tun_key_mask.ip_src),
                              IP_SCAN_ARGS(&tun_key.ip_dst),
@@ -1639,6 +1655,8 @@ parse_odp_key_mask_attr(const char *s, const struct simap *port_names,
 
             tun_key.tun_id = htonll(tun_id);
             tun_key_mask.tun_id = htonll(tun_id_mask);
+            tun_key.nsp = htonl(nsp);
+            tun_key_mask.nsp = htonl(nsp_mask);
             res = parse_flags(&s[n], flow_tun_flag_to_string, &flags);
             tun_key.flags = flags;
             tun_key_mask.flags = UINT16_MAX;
@@ -1657,8 +1675,10 @@ parse_odp_key_mask_attr(const char *s, const struct simap *port_names,
             }
             return n;
         } else if (ovs_scan(s, "tunnel(tun_id=%"SCNi64","
+                            "nsp=%"PRIx32","
                             "src="IP_SCAN_FMT",dst="IP_SCAN_FMT
                             ",tos=%"SCNi8",ttl=%"SCNi8",flags%n", &tun_id,
+                            &tun_key.nsp,
                             IP_SCAN_ARGS(&tun_key.ip_src),
                             IP_SCAN_ARGS(&tun_key.ip_dst),
                             &tun_key.ip_tos, &tun_key.ip_ttl, &n)) {
@@ -1666,6 +1686,7 @@ parse_odp_key_mask_attr(const char *s, const struct simap *port_names,
             uint32_t flags;
 
             tun_key.tun_id = htonll(tun_id);
+            tun_key.nsp = htonl(tun_key.nsp);
             res = parse_flags(&s[n], flow_tun_flag_to_string, &flags);
             tun_key.flags = flags;
 
diff --git a/lib/odp-util.h b/lib/odp-util.h
index 7bc64c7..1ad272e 100644
--- a/lib/odp-util.h
+++ b/lib/odp-util.h
@@ -106,6 +106,7 @@ void odp_portno_names_destroy(struct hmap *portno_names);
  *  - OVS_TUNNEL_KEY_ATTR_TTL            1    3      4      8
  *  - OVS_TUNNEL_KEY_ATTR_DONT_FRAGMENT  0    --     4      4
  *  - OVS_TUNNEL_KEY_ATTR_CSUM           0    --     4      4
+ *  - OVS_TUNNEL_KEY_ATTR_NSP            4    --     4      8
  *  OVS_KEY_ATTR_IN_PORT                 4    --     4      8
  *  OVS_KEY_ATTR_SKB_MARK                4    --     4      8
  *  OVS_KEY_ATTR_ETHERNET               12    --     4     16
@@ -117,7 +118,7 @@ void odp_portno_names_destroy(struct hmap *portno_names);
  *  OVS_KEY_ATTR_ICMPV6                  2     2     4      8
  *  OVS_KEY_ATTR_ND                     28    --     4     32
  *  ----------------------------------------------------------
- *  total                                                 208
+ *  total                                                 216
  *
  * We include some slack space in case the calculation isn't quite right or we
  * add another field and forget to adjust this value.
diff --git a/ofproto/tunnel.c b/ofproto/tunnel.c
index 38b782f..d1d629c 100644
--- a/ofproto/tunnel.c
+++ b/ofproto/tunnel.c
@@ -38,11 +38,13 @@ VLOG_DEFINE_THIS_MODULE(tunnel);
 
 struct tnl_match {
     ovs_be64 in_key;
+    ovs_be32 in_nsp;
     ovs_be32 ip_src;
     ovs_be32 ip_dst;
     odp_port_t odp_port;
     uint32_t pkt_mark;
     bool in_key_flow;
+    bool in_nsp_flow;
     bool ip_src_flow;
     bool ip_dst_flow;
 };
@@ -76,17 +78,21 @@ static struct ovs_rwlock rwlock = OVS_RWLOCK_INITIALIZER;
  *       (ip_dst_flow == false) or arrange for the destination IP to be matched
  *       as tunnel.ip_dst in the OpenFlow flow (ip_dst_flow == true).
  *
+ *     - in_nsp: A vport may match a specific NSH service path (in_nsp_flow ==
+ *       false) or arrange for the service path to be matched as tunnel.in_nsp
+ *       in the OpenFlow flow (in_nsp_flow == true).
+ *
  *     - ip_src: A vport may match a specific IP source address (ip_src_flow ==
  *       false, ip_src != 0), wildcard all source addresses (ip_src_flow ==
  *       false, ip_src == 0), or arrange for the IP source address to be
  *       handled in the OpenFlow flow table (ip_src_flow == true).
  *
- * Thus, there are 2 * 2 * 3 == 12 possible ways a vport can match against a
- * tunnel packet.  We number the possibilities for each field in increasing
- * order as listed in each bullet above.  We order the 12 overall combinations
- * in lexicographic order considering in_key first, then ip_dst, then
- * ip_src. */
-#define N_MATCH_TYPES (2 * 2 * 3)
+ * Thus, there are 2 * 2 * 2 * 3 == 24 possible ways a vport can match
+ * against a tunnel packet.  We number the possibilities for each field in
+ * increasing order as listed in each bullet above.  We order the 24 overall
+ * combinations in lexicographic order considering in_key first, then ip_dst,
+ * then in_nsp, then ip_src. */
+#define N_MATCH_TYPES (2 * 2 * 2 * 3)
 
 /* The three possibilities (see above) for vport ip_src matches. */
 enum ip_src_type {
@@ -101,6 +107,17 @@ enum ip_src_type {
 static struct hmap *tnl_match_maps[N_MATCH_TYPES] OVS_GUARDED_BY(rwlock);
 static struct hmap **tnl_match_map(const struct tnl_match *);
 
+/* The tnl_match_maps_bm is a bitmap for tnl_match_maps showing places which
+ * are filled */
+static uint64_t tnl_match_maps_bm OVS_GUARDED_BY(rwlock);
+static void tnl_match_maps_bm_set(unsigned int);
+static void tnl_match_maps_bm_reset(unsigned int);
+static bool tnl_match_maps_bm_check(unsigned int);
+
+static unsigned int tnl_match_m_to_idx(const struct tnl_match *);
+static void tnl_match_idx_to_m(const struct flow *, unsigned int, 
+    struct tnl_match *);
+
 static struct hmap ofport_map__ = HMAP_INITIALIZER(&ofport_map__);
 static struct hmap *ofport_map OVS_GUARDED_BY(rwlock) = &ofport_map__;
 
@@ -131,6 +148,7 @@ tnl_port_add__(const struct ofport_dpif *ofport, const struct netdev *netdev,
     struct tnl_port *existing_port;
     struct tnl_port *tnl_port;
     struct hmap **map;
+    unsigned int idx;
 
     cfg = netdev_get_tunnel_config(netdev);
     ovs_assert(cfg);
@@ -141,14 +159,17 @@ tnl_port_add__(const struct ofport_dpif *ofport, const struct netdev *netdev,
     tnl_port->change_seq = seq_read(connectivity_seq_get());
 
     tnl_port->match.in_key = cfg->in_key;
+    tnl_port->match.in_nsp = cfg->in_nsp;
     tnl_port->match.ip_src = cfg->ip_src;
     tnl_port->match.ip_dst = cfg->ip_dst;
     tnl_port->match.ip_src_flow = cfg->ip_src_flow;
     tnl_port->match.ip_dst_flow = cfg->ip_dst_flow;
     tnl_port->match.pkt_mark = cfg->ipsec ? IPSEC_MARK : 0;
     tnl_port->match.in_key_flow = cfg->in_key_flow;
+    tnl_port->match.in_nsp_flow = cfg->in_nsp_flow;
     tnl_port->match.odp_port = odp_port;
 
+    idx = tnl_match_m_to_idx(&tnl_port->match);
     map = tnl_match_map(&tnl_port->match);
     existing_port = tnl_find_exact(&tnl_port->match, *map);
     if (existing_port) {
@@ -169,6 +190,7 @@ tnl_port_add__(const struct ofport_dpif *ofport, const struct netdev *netdev,
     if (!*map) {
         *map = xmalloc(sizeof **map);
         hmap_init(*map);
+        tnl_match_maps_bm_set(idx);
     }
     hmap_insert(*map, &tnl_port->match_node, tnl_hash(&tnl_port->match));
     tnl_port_mod_log(tnl_port, "adding");
@@ -219,6 +241,7 @@ static void
 tnl_port_del__(const struct ofport_dpif *ofport) OVS_REQ_WRLOCK(rwlock)
 {
     struct tnl_port *tnl_port;
+    unsigned int idx;
 
     if (!ofport) {
         return;
@@ -229,10 +252,12 @@ tnl_port_del__(const struct ofport_dpif *ofport) OVS_REQ_WRLOCK(rwlock)
         struct hmap **map;
 
         tnl_port_mod_log(tnl_port, "removing");
+        idx = tnl_match_m_to_idx(&tnl_port->match);
         map = tnl_match_map(&tnl_port->match);
         hmap_remove(*map, &tnl_port->match_node);
         if (hmap_is_empty(*map)) {
             hmap_destroy(*map);
+            tnl_match_maps_bm_reset(idx);
             free(*map);
             *map = NULL;
         }
@@ -379,6 +404,10 @@ tnl_port_send(const struct ofport_dpif *ofport, struct flow *flow,
         flow->tunnel.tun_id = cfg->out_key;
     }
 
+    if (!cfg->out_nsp_flow) {
+        flow->tunnel.nsp = cfg->out_nsp;
+    }
+
     if (cfg->ttl_inherit && is_ip_any(flow)) {
         wc->masks.nw_ttl = 0xff;
         flow->tunnel.ip_ttl = flow->nw_ttl;
@@ -406,6 +435,7 @@ tnl_port_send(const struct ofport_dpif *ofport, struct flow *flow,
 
     flow->tunnel.flags = (cfg->dont_fragment ? FLOW_TNL_F_DONT_FRAGMENT : 0)
         | (cfg->csum ? FLOW_TNL_F_CSUM : 0)
+        | (cfg->out_nsp_present ? FLOW_TNL_F_NSP : 0)
         | (cfg->out_key_present ? FLOW_TNL_F_KEY : 0);
 
     if (pre_flow_str) {
@@ -468,46 +498,22 @@ tnl_find_exact(struct tnl_match *match, struct hmap *map)
 static struct tnl_port *
 tnl_find(const struct flow *flow) OVS_REQ_RDLOCK(rwlock)
 {
-    enum ip_src_type ip_src;
-    int in_key_flow;
-    int ip_dst_flow;
     int i;
 
-    i = 0;
-    for (in_key_flow = 0; in_key_flow < 2; in_key_flow++) {
-        for (ip_dst_flow = 0; ip_dst_flow < 2; ip_dst_flow++) {
-            for (ip_src = 0; ip_src < 3; ip_src++) {
-                struct hmap *map = tnl_match_maps[i];
-
-                if (map) {
-                    struct tnl_port *tnl_port;
-                    struct tnl_match match;
-
-                    memset(&match, 0, sizeof match);
-
-                    /* The apparent mix-up of 'ip_dst' and 'ip_src' below is
-                     * correct, because "struct tnl_match" is expressed in
-                     * terms of packets being sent out, but we are using it
-                     * here as a description of how to treat received
-                     * packets. */
-                    match.in_key = in_key_flow ? 0 : flow->tunnel.tun_id;
-                    match.ip_src = (ip_src == IP_SRC_CFG
-                                    ? flow->tunnel.ip_dst
-                                    : 0);
-                    match.ip_dst = ip_dst_flow ? 0 : flow->tunnel.ip_src;
-                    match.odp_port = flow->in_port.odp_port;
-                    match.pkt_mark = flow->pkt_mark;
-                    match.in_key_flow = in_key_flow;
-                    match.ip_dst_flow = ip_dst_flow;
-                    match.ip_src_flow = ip_src == IP_SRC_FLOW;
-
-                    tnl_port = tnl_find_exact(&match, map);
-                    if (tnl_port) {
-                        return tnl_port;
-                    }
-                }
+    for (i = 0; i < N_MATCH_TYPES; i++) {
+        if (tnl_match_maps_bm_check(i)) {
+            struct hmap *map = tnl_match_maps[i];
+
+            if (map) {
+                struct tnl_port *tnl_port;
+                struct tnl_match match;
 
-                i++;
+                memset(&match, 0, sizeof match);
+                tnl_match_idx_to_m(flow, i, &match);
+                tnl_port = tnl_find_exact(&match, map);
+                if (tnl_port) {
+                    return tnl_port;
+                }
             }
         }
     }
@@ -515,10 +521,58 @@ tnl_find(const struct flow *flow) OVS_REQ_RDLOCK(rwlock)
     return NULL;
 }
 
-/* Returns a pointer to the 'tnl_match_maps' element corresponding to 'm''s
- * matching criteria. */
-static struct hmap **
-tnl_match_map(const struct tnl_match *m)
+/* Coverts a index to corresponding matching criteria 'm'. */
+static void
+tnl_match_idx_to_m(const struct flow *flow, unsigned int idx,
+                   struct tnl_match *m)
+{
+    enum ip_src_type ip_src;
+    bool in_key_flow;
+    bool ip_dst_flow;
+    bool in_nsp_flow;
+
+    if (!m)
+        return;
+
+    in_key_flow = (idx < (N_MATCH_TYPES / 2)) ? false : true;
+
+    if (idx >= (N_MATCH_TYPES / 2))
+        idx -= (N_MATCH_TYPES / 2);
+
+    ip_dst_flow = (idx < (N_MATCH_TYPES / (2 * 2))) ? false : true;
+
+    if (idx >= (N_MATCH_TYPES / (2 * 2)))
+        idx -= (N_MATCH_TYPES / (2 * 2));
+
+    in_nsp_flow = (idx < (N_MATCH_TYPES / (2 * 2 * 2))) ? false : true;
+
+    if (idx >= (N_MATCH_TYPES / (2 * 2 * 2)))
+        idx -= (N_MATCH_TYPES / (2 * 2 * 2));
+
+    ip_src = idx;
+
+    /* The apparent mix-up of 'ip_dst' and 'ip_src' below is
+     * correct, because "struct tnl_match" is expressed in
+     * terms of packets being sent out, but we are using it
+     * here as a description of how to treat received
+     * packets. */
+    m->in_key = in_key_flow ? 0 : flow->tunnel.tun_id;
+    m->ip_src = (ip_src == IP_SRC_CFG
+                    ? flow->tunnel.ip_dst
+                    : 0);
+    m->in_nsp = in_nsp_flow ? 0 : flow->tunnel.nsp;
+    m->ip_dst = ip_dst_flow ? 0 : flow->tunnel.ip_src;
+    m->odp_port = flow->in_port.odp_port;
+    m->pkt_mark = flow->pkt_mark;
+    m->in_key_flow = in_key_flow;
+    m->ip_dst_flow = ip_dst_flow;
+    m->in_nsp_flow = in_nsp_flow;
+    m->ip_src_flow = ip_src == IP_SRC_FLOW;
+}
+
+/* Returns a index corresponding to 'm''s matching criteria. */
+static unsigned int
+tnl_match_m_to_idx(const struct tnl_match *m)
 {
     enum ip_src_type ip_src;
 
@@ -526,7 +580,42 @@ tnl_match_map(const struct tnl_match *m)
               : m->ip_src ? IP_SRC_CFG
               : IP_SRC_ANY);
 
-    return &tnl_match_maps[6 * m->in_key_flow + 3 * m->ip_dst_flow + ip_src];
+    return (m->in_key_flow * (N_MATCH_TYPES / 2) +
+            m->ip_dst_flow * (N_MATCH_TYPES / (2 * 2)) +
+            m->in_nsp_flow * (N_MATCH_TYPES / (2 * 2 * 2)) +
+            ip_src);
+}
+
+/* Returns a pointer to the 'tnl_match_maps' element corresponding to 'm''s
+ * matching criteria. */
+static struct hmap **
+tnl_match_map(const struct tnl_match *m)
+{
+    return &tnl_match_maps[tnl_match_m_to_idx(m)];
+}
+
+/* Set the bit corresponding to idx in tnl_match_maps_bm */
+static void
+tnl_match_maps_bm_set(unsigned int idx)
+{
+    if (idx < N_MATCH_TYPES)
+        tnl_match_maps_bm |= 1 << idx;
+}
+
+/* Reset the bit corresponding to idx in tnl_match_maps_bm */
+static void
+tnl_match_maps_bm_reset(unsigned int idx)
+{
+    if (idx < N_MATCH_TYPES)
+        tnl_match_maps_bm &= ~(1 << idx);
+}
+
+/* Return the bit corresponding to idx in tnl_match_maps_bm */
+static bool
+tnl_match_maps_bm_check(unsigned int idx)
+{
+    if (idx < N_MATCH_TYPES)
+        return (tnl_match_maps_bm & (1 << idx)) != 0;
 }
 
 static void
@@ -548,6 +637,12 @@ tnl_match_fmt(const struct tnl_match *match, struct ds *ds)
         ds_put_format(ds, ", key=%#"PRIx64, ntohll(match->in_key));
     }
 
+    if (match->in_nsp_flow) {
+        ds_put_cstr(ds, ", nsp=flow");
+    } else {
+        ds_put_format(ds, ", nsp=%#"PRIx32, ntohl(match->in_nsp));
+    }
+
     ds_put_format(ds, ", dp port=%"PRIu32, match->odp_port);
     ds_put_format(ds, ", pkt mark=%"PRIu32, match->pkt_mark);
 }
@@ -591,6 +686,19 @@ tnl_port_fmt(const struct tnl_port *tnl_port) OVS_REQ_RDLOCK(rwlock)
         }
     }
 
+    if (cfg->out_nsp != cfg->in_nsp ||
+        cfg->out_nsp_present != cfg->in_nsp_present ||
+        cfg->out_nsp_flow != cfg->in_nsp_flow) {
+        ds_put_cstr(&ds, ", out_nsp=");
+        if (!cfg->out_nsp_present) {
+            ds_put_cstr(&ds, "none");
+        } else if (cfg->out_nsp_flow) {
+            ds_put_cstr(&ds, "flow");
+        } else {
+            ds_put_format(&ds, "%#"PRIx32, ntohl(cfg->out_nsp));
+        }
+    }
+
     if (cfg->ttl_inherit) {
         ds_put_cstr(&ds, ", ttl=inherit");
     } else {
-- 
1.7.9.5




More information about the dev mailing list