userspace: Switching of L3 packets in L2 pipeline

Ports have a new layer3 attribute if they send/receive L3 packets.

The packet_type included in structs dp_packet and flow is considered in
ofproto-dpif. The classical L2 match fields (dl_src, dl_dst, dl_type, and
vlan_tci, vlan_vid, vlan_pcp) now have Ethernet as pre-requisite.

A dummy ethernet header is pushed to L3 packets received from L3 ports
before the the pipeline processing starts. The ethernet header is popped
before sending a packet to a L3 port.

For datapath ports that can receive L2 or L3 packets, the packet_type
becomes part of the flow key for datapath flows and is handled
appropriately in dpif-netdev.

In the 'else' branch in flow_put_on_pmd() function, the additional check
flow_equal(&match.flow, &netdev_flow->flow) was removed, as a) the dpcls
lookup is sufficient to uniquely identify a flow and b) it caused false
negatives because the flow in netdev->flow may not properly masked.

In dpif_netdev_flow_put() we now use the same method for constructing the
netdev_flow_key as the one used when adding the flow to the dplcs to make sure
these always match. The function netdev_flow_key_from_flow() used so far was
not only inefficient but sometimes caused mismatches and subsequent flow
update failures.

The kernel datapath does not support the packet_type match field.
Instead it encodes the packet type implictly by the presence or absence of
the Ethernet attribute in the flow key and mask.
This patch filters the PACKET_TYPE attribute out of netlink flow key and
mask to be sent to the kernel datapath.

Signed-off-by: Lorand Jakab <lojakab@cisco.com>
Signed-off-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: Yi Yang <yi.y.yang@intel.com>
Signed-off-by: Jan Scheurich <jan.scheurich@ericsson.com>
Co-authored-by: Zoltan Balogh <zoltan.balogh@ericsson.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>

Author: 2 people

build-aux/extract-ofp-fields

@@ -39,6 +39,7 @@ FORMATTING = {"decimal":            ("MFS_DECIMAL",      1,   8),
               "TCP flags":          ("MFS_TCP_FLAGS",    2,   2)}
 
 PREREQS = {"none": "MFP_NONE",
+           "Ethernet": "MFP_ETHERNET",
            "ARP": "MFP_ARP",
            "VLAN VID": "MFP_VLAN_VID",
            "IPv4": "MFP_IPV4",

datapath/linux/compat/include/linux/openvswitch.h

@@ -363,6 +363,8 @@ enum ovs_key_attr {
 	/* Only used within kernel data path. */
 	OVS_KEY_ATTR_TUNNEL_INFO,  /* struct ovs_tunnel_info */
 #endif
+
+	OVS_KEY_ATTR_PACKET_TYPE,  /* be32 packet type */
 	__OVS_KEY_ATTR_MAX
 };
 

include/openvswitch/match.h

@@ -116,6 +116,7 @@ void match_set_ct_ipv6_dst(struct match *, const struct in6_addr *);
 void match_set_ct_ipv6_dst_masked(struct match *, const struct in6_addr *,
                                   const struct in6_addr *);
 
+void match_set_packet_type(struct match *, ovs_be32 packet_type);
 void match_set_skb_priority(struct match *, uint32_t skb_priority);
 void match_set_dl_type(struct match *, ovs_be16);
 void match_set_dl_src(struct match *, const struct eth_addr );

include/openvswitch/meta-flow.h

@@ -986,7 +986,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: MAC.
      * Maskable: bitwise.
      * Formatting: Ethernet.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read/write.
      * NXM: NXM_OF_ETH_SRC(2) since v1.1.
      * OXM: OXM_OF_ETH_SRC(4) since OF1.2 and v1.7.
@@ -1002,7 +1002,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: MAC.
      * Maskable: bitwise.
      * Formatting: Ethernet.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read/write.
      * NXM: NXM_OF_ETH_DST(1) since v1.1.
      * OXM: OXM_OF_ETH_DST(3) since OF1.2 and v1.7.
@@ -1021,7 +1021,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: be16.
      * Maskable: no.
      * Formatting: hexadecimal.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read-only.
      * NXM: NXM_OF_ETH_TYPE(3) since v1.1.
      * OXM: OXM_OF_ETH_TYPE(5) since OF1.2 and v1.7.
@@ -1051,7 +1051,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: be16.
      * Maskable: bitwise.
      * Formatting: hexadecimal.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read/write.
      * NXM: NXM_OF_VLAN_TCI(4) since v1.1.
      * OXM: none.
@@ -1067,7 +1067,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: be16 (low 12 bits).
      * Maskable: no.
      * Formatting: decimal.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read/write.
      * NXM: none.
      * OXM: none.
@@ -1085,7 +1085,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: be16 (low 12 bits).
      * Maskable: bitwise.
      * Formatting: decimal.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read/write.
      * NXM: none.
      * OXM: OXM_OF_VLAN_VID(6) since OF1.2 and v1.7.
@@ -1101,7 +1101,7 @@ enum OVS_PACKED_ENUM mf_field_id {
      * Type: u8 (low 3 bits).
      * Maskable: no.
      * Formatting: decimal.
-     * Prerequisites: none.
+     * Prerequisites: Ethernet.
      * Access: read/write.
      * NXM: none.
      * OXM: none.
@@ -1809,6 +1809,7 @@ enum OVS_PACKED_ENUM mf_prereqs {
     MFP_NONE,
 
     /* L2 requirements. */
+    MFP_ETHERNET,
     MFP_ARP,
     MFP_VLAN_VID,
     MFP_IPV4,

lib/dpif-netdev.c

@@ -1918,24 +1918,6 @@ netdev_flow_key_clone(struct netdev_flow_key *dst,
            offsetof(struct netdev_flow_key, mf) + src->len);
 }
 
-/* Slow. */
-static void
-netdev_flow_key_from_flow(struct netdev_flow_key *dst,
-                          const struct flow *src)
-{
-    struct dp_packet packet;
-    uint64_t buf_stub[512 / 8];
-
-    dp_packet_use_stub(&packet, buf_stub, sizeof buf_stub);
-    pkt_metadata_from_flow(&packet.md, src);
-    flow_compose(&packet, src);
-    miniflow_extract(&packet, &dst->mf);
-    dp_packet_uninit(&packet);
-
-    dst->len = netdev_flow_key_size(miniflow_n_values(&dst->mf));
-    dst->hash = 0; /* Not computed yet. */
-}
-
 /* Initialize a netdev_flow_key 'mask' from 'match'. */
 static inline void
 netdev_flow_mask_init(struct netdev_flow_key *mask,
@@ -2411,7 +2393,7 @@ dp_netdev_flow_add(struct dp_netdev_pmd_thread *pmd,
     cmap_insert(&pmd->flow_table, CONST_CAST(struct cmap_node *, &flow->node),
                 dp_netdev_flow_hash(&flow->ufid));
 
-    if (OVS_UNLIKELY(VLOG_IS_DBG_ENABLED())) {
+    if (OVS_UNLIKELY(!VLOG_DROP_DBG((&upcall_rl)))) {
         struct ds ds = DS_EMPTY_INITIALIZER;
         struct ofpbuf key_buf, mask_buf;
         struct odp_flow_key_parms odp_parms = {
@@ -2436,10 +2418,21 @@ dp_netdev_flow_add(struct dp_netdev_pmd_thread *pmd,
         ds_put_cstr(&ds, ", actions:");
         format_odp_actions(&ds, actions, actions_len);
 
-        VLOG_DBG_RL(&upcall_rl, "%s", ds_cstr(&ds));
+        VLOG_DBG("%s", ds_cstr(&ds));
 
         ofpbuf_uninit(&key_buf);
         ofpbuf_uninit(&mask_buf);
+
+        /* Add a printout of the actual match installed. */
+        struct match m;
+        ds_clear(&ds);
+        ds_put_cstr(&ds, "flow match: ");
+        miniflow_expand(&flow->cr.flow.mf, &m.flow);
+        miniflow_expand(&flow->cr.mask->mf, &m.wc.masks);
+        match_format(&m, NULL, &ds, OFP_DEFAULT_PRIORITY);
+
+        VLOG_DBG("%s", ds_cstr(&ds));
+
         ds_destroy(&ds);
     }
 
@@ -2476,8 +2469,7 @@ flow_put_on_pmd(struct dp_netdev_pmd_thread *pmd,
             error = ENOENT;
         }
     } else {
-        if (put->flags & DPIF_FP_MODIFY
-            && flow_equal(&match->flow, &netdev_flow->flow)) {
+        if (put->flags & DPIF_FP_MODIFY) {
             struct dp_netdev_actions *new_actions;
             struct dp_netdev_actions *old_actions;
 
@@ -2519,7 +2511,7 @@ static int
 dpif_netdev_flow_put(struct dpif *dpif, const struct dpif_flow_put *put)
 {
     struct dp_netdev *dp = get_dp_netdev(dpif);
-    struct netdev_flow_key key;
+    struct netdev_flow_key key, mask;
     struct dp_netdev_pmd_thread *pmd;
     struct match match;
     ovs_u128 ufid;
@@ -2548,9 +2540,10 @@ dpif_netdev_flow_put(struct dpif *dpif, const struct dpif_flow_put *put)
     }
 
     /* Must produce a netdev_flow_key for lookup.
-     * This interface is no longer performance critical, since it is not used
-     * for upcall processing any more. */
-    netdev_flow_key_from_flow(&key, &match.flow);
+     * Use the same method as employed to create the key when adding
+     * the flow to the dplcs to make sure they match. */
+    netdev_flow_mask_init(&mask, &match);
+    netdev_flow_key_init_masked(&key, &match.flow, &mask);
 
     if (put->pmd_id == PMD_ID_NULL) {
         if (cmap_count(&dp->poll_threads) == 0) {

lib/dpif-netlink.c

@@ -1636,7 +1636,7 @@ dpif_netlink_encode_execute(int dp_ifindex, const struct dpif_execute *d_exec,
                       dp_packet_size(d_exec->packet));
 
     key_ofs = nl_msg_start_nested(buf, OVS_PACKET_ATTR_KEY);
-    odp_key_from_pkt_metadata(buf, &d_exec->packet->md);
+    odp_key_from_dp_packet(buf, d_exec->packet);
     nl_msg_end_nested(buf, key_ofs);
 
     nl_msg_put_unspec(buf, OVS_PACKET_ATTR_ACTIONS,
@@ -2984,6 +2984,38 @@ dpif_netlink_flow_from_ofpbuf(struct dpif_netlink_flow *flow,
     return 0;
 }
 
+
+/*
+ * If PACKET_TYPE attribute is present in 'data', it filters PACKET_TYPE out,
+ * then puts 'data' to 'buf'.
+ */
+static void
+put_exclude_packet_type(struct ofpbuf *buf, uint16_t type,
+                        const struct nlattr *data, uint16_t data_len)
+{
+    const struct nlattr *packet_type;
+
+    packet_type = nl_attr_find__(data, data_len, OVS_KEY_ATTR_PACKET_TYPE);
+
+    if (packet_type) {
+        /* exclude PACKET_TYPE Netlink attribute. */
+        ovs_assert(NLA_ALIGN(packet_type->nla_len) == NL_A_U32_SIZE);
+        size_t packet_type_len = NL_A_U32_SIZE;
+        size_t first_chunk_size = (uint8_t *)packet_type - (uint8_t *)data;
+        size_t second_chunk_size = data_len - first_chunk_size
+                                   - packet_type_len;
+        uint8_t *first_attr = NULL;
+        struct nlattr *next_attr = nl_attr_next(packet_type);
+
+        first_attr = nl_msg_put_unspec_uninit(buf, type,
+                                              data_len - packet_type_len);
+        memcpy(first_attr, data, first_chunk_size);
+        memcpy(first_attr + first_chunk_size, next_attr, second_chunk_size);
+    } else {
+        nl_msg_put_unspec(buf, type, data, data_len);
+    }
+}
+
 /* Appends to 'buf' (which must initially be empty) a "struct ovs_header"
  * followed by Netlink attributes corresponding to 'flow'. */
 static void
@@ -3010,13 +3042,12 @@ dpif_netlink_flow_to_ofpbuf(const struct dpif_netlink_flow *flow,
     }
     if (!flow->ufid_terse || !flow->ufid_present) {
         if (flow->key_len) {
-            nl_msg_put_unspec(buf, OVS_FLOW_ATTR_KEY,
-                              flow->key, flow->key_len);
+            put_exclude_packet_type(buf, OVS_FLOW_ATTR_KEY, flow->key,
+                                           flow->key_len);
         }
-
         if (flow->mask_len) {
-            nl_msg_put_unspec(buf, OVS_FLOW_ATTR_MASK,
-                              flow->mask, flow->mask_len);
+            put_exclude_packet_type(buf, OVS_FLOW_ATTR_MASK, flow->mask,
+                                           flow->mask_len);
         }
         if (flow->actions || flow->actions_len) {
             nl_msg_put_unspec(buf, OVS_FLOW_ATTR_ACTIONS,

lib/dpif.c

@@ -1765,7 +1765,7 @@ log_execute_message(struct dpif *dpif, const struct dpif_execute *execute,
         packet = ofp_packet_to_string(dp_packet_data(execute->packet),
                                       dp_packet_size(execute->packet),
                                       execute->packet->packet_type);
-        odp_key_from_pkt_metadata(&md, &execute->packet->md);
+        odp_key_from_dp_packet(&md, execute->packet);
         ds_put_format(&ds, "%s: %sexecute ",
                       dpif_name(dpif),
                       (subexecute ? "sub-"

lib/match.c

@@ -476,6 +476,13 @@ match_set_ct_ipv6_dst_masked(struct match *match, const struct in6_addr *dst,
     match->wc.masks.ct_ipv6_dst = *mask;
 }
 
+void
+match_set_packet_type(struct match *match, ovs_be32 packet_type)
+{
+    match->flow.packet_type = packet_type;
+    match->wc.masks.packet_type = OVS_BE32_MAX;
+}
+
 void
 match_set_dl_type(struct match *match, ovs_be16 dl_type)
 {
@@ -1249,6 +1256,22 @@ match_format(const struct match *match,
         format_be16_masked(s, "ct_tp_dst", f->ct_tp_dst, wc->masks.ct_tp_dst);
     }
 
+    if (wc->masks.packet_type) {
+        if (pt_ns_type_be(wc->masks.packet_type) == 0) {
+            ds_put_format(s, "packet_type=(%u,*),",
+                          pt_ns(f->packet_type));
+        } else if (pt_ns_type_be(wc->masks.packet_type) == OVS_BE16_MAX) {
+            ds_put_format(s, "packet_type=(%u,%#"PRIx16"),",
+                          pt_ns(f->packet_type),
+                          pt_ns_type(f->packet_type));
+        } else {
+            ds_put_format(s, "packet_type=(%u,%#"PRIx16"/%#"PRIx16"),",
+                          pt_ns(f->packet_type),
+                          pt_ns_type(f->packet_type),
+                          pt_ns_type(wc->masks.packet_type));
+        }
+    }
+
     if (wc->masks.dl_type) {
         skip_type = true;
         if (f->dl_type == htons(ETH_TYPE_IP)) {
@@ -1361,8 +1384,10 @@ match_format(const struct match *match,
                           ntohs(wc->masks.vlans[i].tci));
         }
     }
+
     format_eth_masked(s, "dl_src", f->dl_src, wc->masks.dl_src);
     format_eth_masked(s, "dl_dst", f->dl_dst, wc->masks.dl_dst);
+
     if (!skip_type && wc->masks.dl_type) {
         ds_put_format(s, "%sdl_type=%s0x%04"PRIx16",",
                       colors.param, colors.end, ntohs(f->dl_type));

lib/meta-flow.c

@@ -404,6 +404,8 @@ mf_are_prereqs_ok__(const struct mf_field *mf, const struct flow *flow,
     switch (mf->prereqs) {
     case MFP_NONE:
         return true;
+    case MFP_ETHERNET:
+        return is_ethernet(flow, wc);
     case MFP_ARP:
         return (flow->dl_type == htons(ETH_TYPE_ARP) ||
                 flow->dl_type == htons(ETH_TYPE_RARP));

lib/netdev-vport.c

@@ -97,9 +97,13 @@ netdev_vport_is_patch(const struct netdev *netdev)
 bool
 netdev_vport_is_layer3(const struct netdev *dev)
 {
-    const char *type = netdev_get_type(dev);
+    if (is_vport_class(netdev_get_class(dev))) {
+        struct netdev_vport *vport = netdev_vport_cast(dev);
+
+        return vport->tnl_cfg.is_layer3;
+    }
 
-    return (!strcmp("lisp", type));
+    return false;
 }
 
 static bool