@@ -884,7 +884,12 @@ netdev_srv6_build_header(const struct netdev *netdev,
goto out;
}
- srh = netdev_tnl_ip_build_header(data, params, IPPROTO_ROUTING, 0);
+ /* Writes the netdev_srv6_flowlabel enum value to the ipv6
+ * flowlabel field. It must later be replaced by a valid value
+ * in the header push. */
+ srh = netdev_tnl_ip_build_header(data, params, IPPROTO_ROUTING,
+ htonl(tnl_cfg->srv6_flowlabel));
+
srh->rt_hdr.segments_left = nr_segs - 1;
srh->rt_hdr.type = IPV6_SRCRT_TYPE_4;
srh->rt_hdr.hdrlen = 2 * nr_segs;
@@ -923,10 +928,34 @@ netdev_srv6_push_header(const struct netdev *netdev OVS_UNUSED,
struct dp_packet *packet,
const struct ovs_action_push_tnl *data)
{
+ struct ovs_16aligned_ip6_hdr *inner_ip6, *outer_ip6;
+ enum netdev_srv6_flowlabel srv6_flowlabel;
+ ovs_be32 ipv6_label = 0;
int ip_tot_size;
+ uint32_t flow;
- netdev_tnl_push_ip_header(packet, data->header, data->header_len,
- &ip_tot_size, 0);
+ inner_ip6 = dp_packet_l3(packet);
+ outer_ip6 = netdev_tnl_ipv6_hdr((void *) data->header);
+ srv6_flowlabel = ntohl(get_16aligned_be32(&outer_ip6->ip6_flow)) &
+ IPV6_LABEL_MASK;
+
+ switch (srv6_flowlabel) {
+ case SRV6_FLOWLABEL_COPY:
+ flow = ntohl(get_16aligned_be32(&inner_ip6->ip6_flow));
+ ipv6_label = (flow >> 28) == 6 ? htonl(flow & IPV6_LABEL_MASK) : 0;
+ break;
+
+ case SRV6_FLOWLABEL_ZERO:
+ ipv6_label = 0;
+ break;
+
+ case SRV6_FLOWLABEL_COMPUTE:
+ ipv6_label = htonl(dp_packet_get_rss_hash(packet) & IPV6_LABEL_MASK);
+ break;
+ }
+
+ netdev_tnl_push_ip_header(packet, data->header,
+ data->header_len, &ip_tot_size, ipv6_label);
}
struct dp_packet *
@@ -792,6 +792,14 @@ set_tunnel_config(struct netdev *dev_, const struct smap *args, char **errp)
name, node->value);
break;
}
+ } else if (!strcmp(node->key, "srv6_flowlabel")) {
+ if (!strcmp(node->value, "zero")) {
+ tnl_cfg.srv6_flowlabel = SRV6_FLOWLABEL_ZERO;
+ } else if (!strcmp(node->value, "compute")) {
+ tnl_cfg.srv6_flowlabel = SRV6_FLOWLABEL_COMPUTE;
+ } else {
+ tnl_cfg.srv6_flowlabel = SRV6_FLOWLABEL_COPY;
+ }
} else if (!strcmp(node->key, "payload_type")) {
if (!strcmp(node->value, "mpls")) {
tnl_cfg.payload_ethertype = htons(ETH_TYPE_MPLS);
@@ -97,6 +97,17 @@ enum netdev_pt_mode {
NETDEV_PT_LEGACY_L3,
};
+enum netdev_srv6_flowlabel {
+ /* Copy the flowlabel of inner packet. */
+ SRV6_FLOWLABEL_COPY,
+
+ /* Simply set flowlabel to 0. */
+ SRV6_FLOWLABEL_ZERO,
+
+ /* Set flowlabel to a hash over L3/L4 fields of the inner packet. */
+ SRV6_FLOWLABEL_COMPUTE,
+};
+
/* Configuration specific to tunnels. */
struct netdev_tunnel_config {
ovs_be64 in_key;
@@ -144,6 +155,7 @@ struct netdev_tunnel_config {
uint8_t srv6_num_segs;
#define SRV6_MAX_SEGS 6
struct in6_addr srv6_segs[SRV6_MAX_SEGS];
+ enum netdev_srv6_flowlabel srv6_flowlabel;
};
void netdev_run(void);
@@ -1,5 +1,91 @@
AT_BANNER([tunnel_push_pop_ipv6])
+AT_SETUP([tunnel_push_pop_ipv6 - srv6])
+
+OVS_VSWITCHD_START([add-port br0 p0 -- set Interface p0 type=dummy ofport_request=1 other-config:hwaddr=aa:55:aa:55:00:00 options:pcap=p0.pcap])
+AT_CHECK([ovs-vsctl add-br int-br1 -- set bridge int-br1 datapath_type=dummy], [0])
+AT_CHECK([ovs-vsctl add-br int-br2 -- set bridge int-br2 datapath_type=dummy], [0])
+AT_CHECK([ovs-vsctl add-br int-br3 -- set bridge int-br3 datapath_type=dummy], [0])
+AT_CHECK([ovs-vsctl add-port int-br1 t1 -- set Interface t1 type=srv6 \
+ options:remote_ip=2001:cafe::91 ofport_request=2 \
+ options:srv6_flowlabel=copy \
+ ], [0])
+AT_CHECK([ovs-vsctl add-port int-br2 t2 -- set Interface t2 type=srv6 \
+ options:remote_ip=2001:cafe::92 ofport_request=3 \
+ options:srv6_flowlabel=zero \
+ ], [0])
+AT_CHECK([ovs-vsctl add-port int-br3 t3 -- set Interface t3 type=srv6 \
+ options:remote_ip=2001:cafe::93 ofport_request=4 \
+ options:srv6_flowlabel=compute \
+ ], [0])
+
+dnl First setup dummy interface IP address, then add the route
+dnl so that tnl-port table can get valid IP address for the device.
+AT_CHECK([ovs-appctl netdev-dummy/ip6addr br0 2001:cafe::88/24], [0], [OK
+])
+AT_CHECK([ovs-appctl ovs/route/add 2001:cafe::0/24 br0], [0], [OK
+])
+AT_CHECK([ovs-appctl tnl/neigh/set br0 2001:cafe::91 aa:55:aa:55:00:01], [0], [OK
+])
+AT_CHECK([ovs-appctl tnl/neigh/set br0 2001:cafe::92 aa:55:aa:55:00:02], [0], [OK
+])
+AT_CHECK([ovs-appctl tnl/neigh/set br0 2001:cafe::93 aa:55:aa:55:00:03], [0], [OK
+])
+AT_CHECK([ovs-ofctl add-flow br0 action=1])
+AT_CHECK([ovs-ofctl add-flow int-br1 action=2])
+AT_CHECK([ovs-ofctl add-flow int-br2 action=3])
+AT_CHECK([ovs-ofctl add-flow int-br3 action=4])
+
+dnl Check "srv6_flowlabel=copy".
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br1 'in_port(2),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.2,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br1 'in_port(2),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.3,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br1 'in_port(2),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::2,label=2,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br1 'in_port(2),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::3,label=3,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-ofctl parse-pcap p0.pcap | tail -n 4 | grep -o 'ipv6_label=0x[[0-9a-f]]*'], [0], [dnl
+ipv6_label=0x00000
+ipv6_label=0x00000
+ipv6_label=0x00002
+ipv6_label=0x00003
+])
+
+dnl Check "srv6_flowlabel=zero".
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br2 'in_port(3),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.2,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br2 'in_port(3),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.3,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br2 'in_port(3),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::2,label=2,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br2 'in_port(3),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::3,label=3,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-ofctl parse-pcap p0.pcap | tail -n 4 | grep -o 'ipv6_label=0x[[0-9a-f]]*'], [0], [dnl
+ipv6_label=0x00000
+ipv6_label=0x00000
+ipv6_label=0x00000
+ipv6_label=0x00000
+])
+
+dnl dnl Check "srv6_flowlabel=compute" for different flows.
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.2,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.3,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::2,label=2,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::3,label=3,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-ofctl parse-pcap p0.pcap | tail -n 4 | grep -o 'ipv6_label=0x[[0-9a-f]]*'| sort | uniq -c | wc -l], [0], [dnl
+4
+])
+
+dnl dnl Check "srv6_flowlabel=compute" for same IPv4/TCP flow.
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.2,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x0800),ipv4(src=10.0.0.1,dst=10.0.0.2,proto=6,tos=0,ttl=64,frag=no),tcp(src=100,dst=200),tcp_flags(0x002)'])
+AT_CHECK([ovs-ofctl parse-pcap p0.pcap | tail -n 2 | grep -o 'ipv6_label=0x[[0-9a-f]]*' | sort | uniq -c | wc -l], [0], [dnl
+1
+])
+
+dnl dnl Check "srv6_flowlabel=compute" for same IPv6/TCP flow.
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::2,label=2,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x001)'])
+AT_CHECK([ovs-appctl netdev-dummy/receive int-br3 'in_port(4),eth(src=f8:bc:12:44:34:b6,dst=aa:55:aa:55:00:00),eth_type(0x86dd),ipv6(src=2001:beef::1,dst=2001:beef::2,label=3,proto=6,tclass=0x0,hlimit=64),tcp(src=100,dst=200),tcp_flags(0x002)'])
+AT_CHECK([ovs-ofctl parse-pcap p0.pcap | tail -n 2 | grep -o 'ipv6_label=0x[[0-9a-f]]*' | sort | uniq -c | wc -l], [0], [dnl
+1
+])
+
+OVS_VSWITCHD_STOP
+AT_CLEANUP
+
AT_SETUP([tunnel_push_pop_ipv6 - ip6gre])
OVS_VSWITCHD_START([add-port br0 p0 -- set Interface p0 type=dummy ofport_request=1 other-config:hwaddr=aa:55:aa:55:00:00])
@@ -3287,6 +3287,32 @@
<ref column="options" key="remote_ip"/>.
</p>
</column>
+ <column name="options" key="srv6_flowlabel"
+ type='{"type": "string",
+ "enum": ["set", ["zero", "copy", "compute"]]}'>
+ <p>
+ Optional.
+ This option controls how flowlabel in outer IPv6 header is
+ configured. It gives the benefit of IPv6 flow label based
+ load balancing, which is supported by some popular vendor
+ appliances. Like net.ipv6.seg6_flowlabel sysconfig, it is
+ one of the three values below:
+ </p>
+ <ul>
+ <li>
+ By default, or if this option is <code>copy</code>, copy the
+ flowlabel of inner IPv6 header to the flowlabel of outer IPv6
+ header. If inner header is not IPv6, it is set to 0.
+ </li>
+ <li>
+ If this option is <code>zero</code>, simply set flowlabel to 0.
+ </li>
+ <li>
+ If this option is <code>compute</code>, set flowlabel to a hash
+ over the L3/L4 fields of the inner packet.
+ </li>
+ </ul>
+ </column>
</group>
<group title="Patch Options">
It supports flowlabel based load balancing by controlling the flowlabel of outer IPv6 header, which is already implemented in Linux kernel as seg6_flowlabel sysctl [1]. [1]: https://docs.kernel.org/networking/seg6-sysctl.html Signed-off-by: Nobuhiro MIKI <nmiki@yahoo-corp.jp> --- lib/netdev-native-tnl.c | 35 ++++++++++++-- lib/netdev-vport.c | 8 ++++ lib/netdev.h | 12 +++++ tests/tunnel-push-pop-ipv6.at | 86 +++++++++++++++++++++++++++++++++++ vswitchd/vswitch.xml | 26 +++++++++++ 5 files changed, 164 insertions(+), 3 deletions(-)