hash: Add symmetric L3/L4 hash functions for multipath, bundle hashing.

Signed-off-by: Jeroen van Bemmel <jvb127@gmail.com>
[blp@nicira.com made code style fixes, expanded documentation]
Signed-off-by: Ben Pfaff <blp@nicira.com>

8 files changed, 132 insertions, 6 deletions

diff --git a/AUTHORS b/AUTHORS
index 14574a11c..93d7dc5e5 100644
--- a/AUTHORS
+++ b/AUTHORS
@@ -86,6 +86,7 @@ Jason Kölker            jason@koelker.net
 Jasper Capel            jasper@capel.tv
 Jean Tourrilhes         jt@hpl.hp.com
 Jeremy Stribling        strib@nicira.com
+Jeroen van Bemmel       jvb127@gmail.com
 Jesse Gross             jesse@nicira.com
 Jing Ai                 jinga@google.com
 Joe Perches             joe@perches.com
diff --git a/NEWS b/NEWS
index 59f25527f..b08919b5b 100644
--- a/NEWS
+++ b/NEWS
@@ -11,6 +11,7 @@ Post-v2.4.0
      * OpenFlow 1.4+ OFPMP_TABLE_DESC is now implemented.
    - Support for matching and generating options with Geneve tunnels.
    - Support Multicast Listener Discovery (MLDv1 and MLDv2).
+   - Add 'symmetric_l3l4' and 'symmetric_l3l4+udp' hash functions.
 
 
 v2.4.0 - xx xxx xxxx
diff --git a/include/openflow/nicira-ext.h b/include/openflow/nicira-ext.h
index 22325aa40..9154800a4 100644
--- a/include/openflow/nicira-ext.h
+++ b/include/openflow/nicira-ext.h
@@ -109,7 +109,28 @@ enum nx_hash_fields {
      *  - NXM_OF_IP_SRC / NXM_OF_IP_DST
      *  - NXM_OF_TCP_SRC / NXM_OF_TCP_DST
      */
-    NX_HASH_FIELDS_SYMMETRIC_L4
+    NX_HASH_FIELDS_SYMMETRIC_L4,
+
+    /* L3+L4 only, including the following fields:
+     *
+     *  - NXM_OF_IP_PROTO
+     *  - NXM_OF_IP_SRC / NXM_OF_IP_DST
+     *  - NXM_OF_SCTP_SRC / NXM_OF_SCTP_DST
+     *  - NXM_OF_TCP_SRC / NXM_OF_TCP_DST
+     */
+    NX_HASH_FIELDS_SYMMETRIC_L3L4,
+
+    /* L3+L4 only with UDP ports, including the following fields:
+     *
+     *  - NXM_OF_IP_PROTO
+     *  - NXM_OF_IP_SRC / NXM_OF_IP_DST
+     *  - NXM_OF_SCTP_SRC / NXM_OF_SCTP_DST
+     *  - NXM_OF_TCP_SRC / NXM_OF_TCP_DST
+     *  - NXM_OF_UDP_SRC / NXM_OF_UDP_DST
+     */
+    NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP
+
+
 };
 
 /* This command enables or disables an Open vSwitch extension that allows a
diff --git a/lib/bundle.c b/lib/bundle.c
index ee8079c81..f1b1478e5 100644
--- a/lib/bundle.c
+++ b/lib/bundle.c
@@ -186,6 +186,10 @@ bundle_parse__(const char *s, char **save_ptr,
         bundle->fields = NX_HASH_FIELDS_ETH_SRC;
     } else if (!strcasecmp(fields, "symmetric_l4")) {
         bundle->fields = NX_HASH_FIELDS_SYMMETRIC_L4;
+    } else if (!strcasecmp(fields, "symmetric_l3l4")) {
+        bundle->fields = NX_HASH_FIELDS_SYMMETRIC_L3L4;
+    } else if (!strcasecmp(fields, "symmetric_l3l4+udp")) {
+        bundle->fields = NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP;
     } else {
         return xasprintf("%s: unknown fields `%s'", s, fields);
     }
diff --git a/lib/flow.c b/lib/flow.c
index 6bfe73864..366922820 100644
--- a/lib/flow.c
+++ b/lib/flow.c
@@ -1347,6 +1347,40 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis)
     return jhash_bytes(&fields, sizeof fields, basis);
 }
 
+/* Hashes 'flow' based on its L3 through L4 protocol information */
+uint32_t
+flow_hash_symmetric_l3l4(const struct flow *flow, uint32_t basis,
+                         bool inc_udp_ports)
+{
+    uint32_t hash = basis;
+
+    /* UDP source and destination port are also taken into account. */
+    if (flow->dl_type == htons(ETH_TYPE_IP)) {
+        hash = hash_add(hash,
+                        (OVS_FORCE uint32_t) (flow->nw_src ^ flow->nw_dst));
+    } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) {
+        /* IPv6 addresses are 64-bit aligned inside struct flow. */
+        const uint64_t *a = ALIGNED_CAST(uint64_t *, flow->ipv6_src.s6_addr);
+        const uint64_t *b = ALIGNED_CAST(uint64_t *, flow->ipv6_dst.s6_addr);
+
+        for (int i = 0; i < 4; i++) {
+            hash = hash_add64(hash, a[i] ^ b[i]);
+        }
+    } else {
+        /* Cannot hash non-IP flows */
+        return 0;
+    }
+
+    hash = hash_add(hash, flow->nw_proto);
+    if (flow->nw_proto == IPPROTO_TCP || flow->nw_proto == IPPROTO_SCTP ||
+         (inc_udp_ports && flow->nw_proto == IPPROTO_UDP)) {
+        hash = hash_add(hash,
+                        (OVS_FORCE uint16_t) (flow->tp_src ^ flow->tp_dst));
+    }
+
+    return hash_finish(hash, basis);
+}
+
 /* Initialize a flow with random fields that matter for nx_hash_fields. */
 void
 flow_random_hash_fields(struct flow *flow)
@@ -1414,6 +1448,30 @@ flow_mask_hash_fields(const struct flow *flow, struct flow_wildcards *wc,
         wc->masks.vlan_tci |= htons(VLAN_VID_MASK | VLAN_CFI);
         break;
 
+    case NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP:
+        if (is_ip_any(flow) && flow->nw_proto == IPPROTO_UDP) {
+            memset(&wc->masks.tp_src, 0xff, sizeof wc->masks.tp_src);
+            memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst);
+        }
+        /* no break */
+    case NX_HASH_FIELDS_SYMMETRIC_L3L4:
+        if (flow->dl_type == htons(ETH_TYPE_IP)) {
+            memset(&wc->masks.nw_src, 0xff, sizeof wc->masks.nw_src);
+            memset(&wc->masks.nw_dst, 0xff, sizeof wc->masks.nw_dst);
+        } else if (flow->dl_type == htons(ETH_TYPE_IPV6)) {
+            memset(&wc->masks.ipv6_src, 0xff, sizeof wc->masks.ipv6_src);
+            memset(&wc->masks.ipv6_dst, 0xff, sizeof wc->masks.ipv6_dst);
+        } else {
+            break; /* non-IP flow */
+        }
+
+        memset(&wc->masks.nw_proto, 0xff, sizeof wc->masks.nw_proto);
+        if (flow->nw_proto == IPPROTO_TCP || flow->nw_proto == IPPROTO_SCTP) {
+            memset(&wc->masks.tp_src, 0xff, sizeof wc->masks.tp_src);
+            memset(&wc->masks.tp_dst, 0xff, sizeof wc->masks.tp_dst);
+        }
+        break;
+
     default:
         OVS_NOT_REACHED();
     }
@@ -1431,6 +1489,13 @@ flow_hash_fields(const struct flow *flow, enum nx_hash_fields fields,
 
     case NX_HASH_FIELDS_SYMMETRIC_L4:
         return flow_hash_symmetric_l4(flow, basis);
+
+    case NX_HASH_FIELDS_SYMMETRIC_L3L4:
+        return flow_hash_symmetric_l3l4(flow, basis, false);
+
+    case NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP:
+        return flow_hash_symmetric_l3l4(flow, basis, true);
+
     }
 
     OVS_NOT_REACHED();
@@ -1443,6 +1508,8 @@ flow_hash_fields_to_str(enum nx_hash_fields fields)
     switch (fields) {
     case NX_HASH_FIELDS_ETH_SRC: return "eth_src";
     case NX_HASH_FIELDS_SYMMETRIC_L4: return "symmetric_l4";
+    case NX_HASH_FIELDS_SYMMETRIC_L3L4: return "symmetric_l3l4";
+    case NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP: return "symmetric_l3l4+udp";
     default: return "<unknown>";
     }
 }
@@ -1452,7 +1519,9 @@ bool
 flow_hash_fields_valid(enum nx_hash_fields fields)
 {
     return fields == NX_HASH_FIELDS_ETH_SRC
-        || fields == NX_HASH_FIELDS_SYMMETRIC_L4;
+        || fields == NX_HASH_FIELDS_SYMMETRIC_L4
+        || fields == NX_HASH_FIELDS_SYMMETRIC_L3L4
+        || fields == NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP;
 }
 
 /* Returns a hash value for the bits of 'flow' that are active based on
diff --git a/lib/flow.h b/lib/flow.h
index a1c6e9751..97790859b 100644
--- a/lib/flow.h
+++ b/lib/flow.h
@@ -341,6 +341,8 @@ bool flow_wildcards_equal(const struct flow_wildcards *,
                           const struct flow_wildcards *);
 uint32_t flow_hash_5tuple(const struct flow *flow, uint32_t basis);
 uint32_t flow_hash_symmetric_l4(const struct flow *flow, uint32_t basis);
+uint32_t flow_hash_symmetric_l3l4(const struct flow *flow, uint32_t basis,
+                         bool inc_udp_ports );
 
 /* Initialize a flow with random fields that matter for nx_hash_fields. */
 void flow_random_hash_fields(struct flow *);
diff --git a/lib/multipath.c b/lib/multipath.c
index 355ef4b47..0a58c062c 100644
--- a/lib/multipath.c
+++ b/lib/multipath.c
@@ -162,6 +162,10 @@ multipath_parse__(struct ofpact_multipath *mp, const char *s_, char *s)
         mp->fields = NX_HASH_FIELDS_ETH_SRC;
     } else if (!strcasecmp(fields, "symmetric_l4")) {
         mp->fields = NX_HASH_FIELDS_SYMMETRIC_L4;
+    } else if (!strcasecmp(fields, "symmetric_l3l4")) {
+        mp->fields = NX_HASH_FIELDS_SYMMETRIC_L3L4;
+    } else if (!strcasecmp(fields, "symmetric_l3l4+udp")) {
+        mp->fields = NX_HASH_FIELDS_SYMMETRIC_L3L4_UDP;
     } else {
         return xasprintf("%s: unknown fields `%s'", s_, fields);
     }
diff --git a/utilities/ovs-ofctl.8.in b/utilities/ovs-ofctl.8.in
index 0a40efa1f..fc5f2921c 100644
--- a/utilities/ovs-ofctl.8.in
+++ b/utilities/ovs-ofctl.8.in
@@ -1619,8 +1619,32 @@ numbered 0 through \fIn_links\fR minus 1, and stores the link into
 \fIdst\fB[\fIstart\fB..\fIend\fB]\fR, which must be an NXM field as
 described above.
 .IP
-Currently, \fIfields\fR must be either \fBeth_src\fR or
-\fBsymmetric_l4\fR and \fIalgorithm\fR must be one of \fBmodulo_n\fR,
+\fIfields\fR must be one of the following:
+.RS
+.IP \fBeth_src\fR
+Hashes Ethernet source address only.
+.IP \fBsymmetric_l4\fR
+Hashes Ethernet source, destination, and type, VLAN ID, IPv4/IPv6
+source, destination, and protocol, and TCP or SCTP (but not UDP)
+ports.  The hash is computed so that pairs of corresponding flows in
+each direction hash to the same value, in environments where L2 paths
+are the same in each direction.  UDP ports are not included in the
+hash to support protocols such as VXLAN that use asymmetric ports in
+each direction.
+.IP \fBsymmetric_l3l4\fR
+Hashes IPv4/IPv6 source, destination, and protocol, and TCP or SCTP
+(but not UDP) ports.  Like \fBsymmetric_l4\fR, this is a symmetric
+hash, but by excluding L2 headers it is more effective in environments
+with asymmetric L2 paths (e.g. paths involving VRRP IP addresses on a
+router).  Not an effective hash function for protocols other than IPv4
+and IPv6, which hash to a constant zero.
+.IP \fBsymmetric_l3l4+udp\fR
+Like \fBsymmetric_l3l4+udp\fR, but UDP ports are included in the hash.
+This is a more effective hash when asymmetric UDP protocols such as
+VXLAN are not a consideration.
+.RE
+.IP
+\fIalgorithm\fR must be one of \fBmodulo_n\fR,
 \fBhash_threshold\fR, \fBhrw\fR, and \fBiter_hash\fR.  Only
 the \fBiter_hash\fR algorithm uses \fIarg\fR.
 .IP
@@ -1633,8 +1657,8 @@ slaves represented as \fIslave_type\fR.  Currently the only supported
 \fIslave_type\fR is \fBofport\fR.  Thus, each \fIs1\fR through \fIsN\fR should
 be an OpenFlow port number. Outputs to the selected slave.
 .IP
-Currently, \fIfields\fR must be either \fBeth_src\fR or \fBsymmetric_l4\fR and
-\fIalgorithm\fR must be one of \fBhrw\fR and \fBactive_backup\fR.
+Currently, \fIfields\fR must be either \fBeth_src\fR, \fBsymmetric_l4\fR, \fBsymmetric_l3l4\fR, or \fBsymmetric_l3l4+udp\fR, 
+and \fIalgorithm\fR must be one of \fBhrw\fR and \fBactive_backup\fR.
 .IP
 Example: \fBbundle(eth_src,0,hrw,ofport,slaves:4,8)\fR uses an Ethernet source
 hash with basis 0, to select between OpenFlow ports 4 and 8 using the Highest