Skip to contentSkip to conversation

net: Multipath hashing on inner L3

PATCH In review Stephen Suryaputra 2 participants 5 messages Jul 6, 2019 – Jul 8, 2019

Subsystems (5)
lore
Share & tools
lore
#0 lore raw

v2 cover net: Multipath hashing on inner L3

This series extends commit 363887a2cdfe ("ipv4: Support multipath hashing on inner IP pkts for GRE tunnel") to include support when the outer L3 is IPv6 and to consider the case where the inner L3 is different version from the outer L3, such as IPv6 tunneled by IPv4 GRE or vice versa. It also includes kselftest scripts to test the use cases.

v2: Clarify the commit messages in the commits in this series to use the
term tunneled by IPv4 GRE or by IPv6 GRE so that it's clear which one is the inner and which one is the outer (per David Miller).

Stephen Suryaputra (3):
ipv4: Multipath hashing on inner L3 needs to consider inner IPv6 pkts
ipv6: Support multipath hashing on inner IP pkts
selftests: forwarding: Test multipath hashing on inner IP pkts for GRE
tunnel

 Documentation/networking/ip-sysctl.txt        |   1 +
 net/ipv4/route.c                              |  21 +-
 net/ipv6/route.c                              |  36 +++
 .../net/forwarding/gre_inner_v4_multipath.sh  | 305 +++++++++++++++++
 .../net/forwarding/gre_inner_v6_multipath.sh  | 306 ++++++++++++++++++
 .../forwarding/ip6gre_inner_v4_multipath.sh   | 304 +++++++++++++++++
 .../forwarding/ip6gre_inner_v6_multipath.sh   | 305 +++++++++++++++++
 7 files changed, 1274 insertions(+), 4 deletions(-)
 create mode 100755 tools/testing/selftests/net/forwarding/gre_inner_v4_multipath.sh
 create mode 100755 tools/testing/selftests/net/forwarding/gre_inner_v6_multipath.sh
 create mode 100755 tools/testing/selftests/net/forwarding/ip6gre_inner_v4_multipath.sh
 create mode 100755 tools/testing/selftests/net/forwarding/ip6gre_inner_v6_multipath.sh
Signature · 1 line
2.17.1
#1 lore raw

PATCH v2 1/3 ipv4: Multipath hashing on inner L3 needs to consider inner IPv6 pkts

Commit 363887a2cdfe ("ipv4: Support multipath hashing on inner IP pkts for GRE tunnel") supports multipath policy value of 2, Layer 3 or inner Layer 3 if present, but it only considers inner IPv4. There is a use case of IPv6 is tunneled by IPv4 GRE, thus add the ability to hash on inner IPv6 addresses.

Fixes 363887a2cdfe ("ipv4: Support multipath hashing on inner IP pkts for GRE tunnel")
---
 net/ipv4/route.c | 21 +++++++++++++++++----
 1 file changed, 17 insertions(+), 4 deletions(-)
net/ipv4/route.c +17 -4
index dc1f510a7c81..abaa7f9371e5 100644
@@ -1964,17 +1964,30 @@ int fib_multipath_hash(const struct net *net, const struct flowi4 *fl4,
 		break;
 	case 2:
 		memset(&hash_keys, 0, sizeof(hash_keys));
-		hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
 		/* skb is currently provided only when forwarding */
 		if (skb) {
 			struct flow_keys keys;
 
 			skb_flow_dissect_flow_keys(skb, &keys, 0);
-
-			hash_keys.addrs.v4addrs.src = keys.addrs.v4addrs.src;
-			hash_keys.addrs.v4addrs.dst = keys.addrs.v4addrs.dst;
+			/* Inner can be v4 or v6 */
+			if (keys.control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+				hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+				hash_keys.addrs.v4addrs.src = keys.addrs.v4addrs.src;
+				hash_keys.addrs.v4addrs.dst = keys.addrs.v4addrs.dst;
+			} else if (keys.control.addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+				hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+				hash_keys.addrs.v6addrs.src = keys.addrs.v6addrs.src;
+				hash_keys.addrs.v6addrs.dst = keys.addrs.v6addrs.dst;
+				hash_keys.tags.flow_label = keys.tags.flow_label;
+				hash_keys.basic.ip_proto = keys.basic.ip_proto;
+			} else {
+				/* Same as case 0 */
+				hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+				ip_multipath_l3_keys(skb, &hash_keys);
+			}
 		} else {
 			/* Same as case 0 */
+			hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
 			hash_keys.addrs.v4addrs.src = fl4->saddr;
 			hash_keys.addrs.v4addrs.dst = fl4->daddr;
 		}
Signature · 1 line
2.17.1
#2 lore raw

PATCH v2 2/3 ipv6: Support multipath hashing on inner IP pkts

In reply to Stephen Suryaputra

Make the same support as commit 363887a2cdfe ("ipv4: Support multipath hashing on inner IP pkts for GRE tunnel") for outer IPv6. The hashing considers both IPv4 and IPv6 pkts when they are tunneled by IPv6 GRE.

---
 Documentation/networking/ip-sysctl.txt |  1 +
 net/ipv6/route.c                       | 36 ++++++++++++++++++++++++++
 2 files changed, 37 insertions(+)
Documentation/networking/ip-sysctl.txt +1 -0
index f0e6d1f53485..48c79e78817b 100644
@@ -1473,6 +1473,7 @@ fib_multipath_hash_policy - INTEGER
 	Possible values:
 	0 - Layer 3 (source and destination addresses plus flow label)
 	1 - Layer 4 (standard 5-tuple)
+	2 - Layer 3 or inner Layer 3 if present
 
 anycast_src_echo_reply - BOOLEAN
 	Controls the use of anycast addresses as source addresses for ICMPv6
net/ipv6/route.c +36 -0
index 39361f57351a..4d2e6b31a8d6 100644
@@ -2370,6 +2370,42 @@ u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
 			hash_keys.basic.ip_proto = fl6->flowi6_proto;
 		}
 		break;
+	case 2:
+		memset(&hash_keys, 0, sizeof(hash_keys));
+		hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+		if (skb) {
+			struct flow_keys keys;
+
+			if (!flkeys) {
+				skb_flow_dissect_flow_keys(skb, &keys, 0);
+				flkeys = &keys;
+			}
+
+			/* Inner can be v4 or v6 */
+			if (flkeys->control.addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+				hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS;
+				hash_keys.addrs.v4addrs.src = flkeys->addrs.v4addrs.src;
+				hash_keys.addrs.v4addrs.dst = flkeys->addrs.v4addrs.dst;
+			} else if (flkeys->control.addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+				hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+				hash_keys.addrs.v6addrs.src = flkeys->addrs.v6addrs.src;
+				hash_keys.addrs.v6addrs.dst = flkeys->addrs.v6addrs.dst;
+				hash_keys.tags.flow_label = flkeys->tags.flow_label;
+				hash_keys.basic.ip_proto = flkeys->basic.ip_proto;
+			} else {
+				/* Same as case 0 */
+				hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+				ip6_multipath_l3_keys(skb, &hash_keys, flkeys);
+			}
+		} else {
+			/* Same as case 0 */
+			hash_keys.control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
+			hash_keys.addrs.v6addrs.src = fl6->saddr;
+			hash_keys.addrs.v6addrs.dst = fl6->daddr;
+			hash_keys.tags.flow_label = (__force u32)flowi6_get_flowlabel(fl6);
+			hash_keys.basic.ip_proto = fl6->flowi6_proto;
+		}
+		break;
 	}
 	mhash = flow_hash_from_keys(&hash_keys);
 
Signature · 1 line
2.17.1
#3 lore raw

PATCH v2 3/3 selftests: forwarding: Test multipath hashing on inner IP pkts for GRE tunnel

In reply to Stephen Suryaputra

Add selftest scripts for multipath hashing on inner IP pkts when there is a single GRE tunnel but there are multiple underlay routes to reach the other end of the tunnel.

Four cases are covered in these scripts:
- IPv4 inner, IPv4 outer
- IPv6 inner, IPv4 outer
- IPv4 inner, IPv6 outer
- IPv6 inner, IPv6 outer

---
 .../net/forwarding/gre_inner_v4_multipath.sh  | 305 +++++++++++++++++
 .../net/forwarding/gre_inner_v6_multipath.sh  | 306 ++++++++++++++++++
 .../forwarding/ip6gre_inner_v4_multipath.sh   | 304 +++++++++++++++++
 .../forwarding/ip6gre_inner_v6_multipath.sh   | 305 +++++++++++++++++
 4 files changed, 1220 insertions(+)
 create mode 100755 tools/testing/selftests/net/forwarding/gre_inner_v4_multipath.sh
 create mode 100755 tools/testing/selftests/net/forwarding/gre_inner_v6_multipath.sh
 create mode 100755 tools/testing/selftests/net/forwarding/ip6gre_inner_v4_multipath.sh
 create mode 100755 tools/testing/selftests/net/forwarding/ip6gre_inner_v6_multipath.sh
tools/testing/selftests/net/forwarding/gre_inner_v4_multipath.sh new file +305 -0
new file mode 100755
index 000000000000..e4009f658003
@@ -0,0 +1,305 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+
+# Test traffic distribution when there are multiple routes between an IPv4
+# GRE tunnel. The tunnel carries IPv4 traffic between multiple hosts.
+# Multiple routes are in the underlay network. With the default multipath
+# policy, SW2 will only look at the outer IP addresses, hence only a single
+# route would be used.
+#
+# +-------------------------+
+# | H1                      |
+# |               $h1 +     |
+# | 192.0.3.{2-62}/24 |     |
+# +-------------------|-----+
+#                     |
+# +-------------------|------------------------+
+# | SW1               |                        |
+# |              $ol1 +                        |
+# |      192.0.3.1/24                          |
+# |                                            |
+# |  + g1 (gre)                                |
+# |    loc=192.0.2.65                          |
+# |    rem=192.0.2.66 --.                      |
+# |    tos=inherit      |                      |
+# |                     v                      |
+# |                     + $ul1                 |
+# |                     | 192.0.2.129/28       |
+# +---------------------|----------------------+
+#                       |
+# +---------------------|----------------------+
+# | SW2                 |                      |
+# |               $ul21 +                      |
+# |      192.0.2.130/28                        |
+# |                   |                        |
+# !   ________________|_____                   |
+# |  /                      \                  |
+# |  |                      |                  |
+# |  + $ul22.111 (vlan)     + $ul22.222 (vlan) |
+# |  | 192.0.2.145/28       | 192.0.2.161/28   |
+# |  |                      |                  |
+# +--|----------------------|------------------+
+#    |                      |
+# +--|----------------------|------------------+
+# |  |                      |                  |
+# |  + $ul32.111 (vlan)     + $ul32.222 (vlan) |
+# |  | 192.0.2.146/28       | 192.0.2.162/28   |
+# |  |                      |                  |
+# |  \______________________/                  |
+# |                   |                        |
+# |                   |                        |
+# |               $ul31 +                      |
+# |      192.0.2.177/28 |                  SW3 |
+# +---------------------|----------------------+
+#                       |
+# +---------------------|----------------------+
+# |                     + $ul4                 |
+# |                     ^ 192.0.2.178/28       |
+# |                     |                      |
+# |  + g2 (gre)         |                      |
+# |    loc=192.0.2.66   |                      |
+# |    rem=192.0.2.65 --'                      |
+# |    tos=inherit                             |
+# |                                            |
+# |               $ol4 +                       |
+# |       192.0.4.1/24 |                   SW4 |
+# +--------------------|-----------------------+
+#                      |
+# +--------------------|---------+
+# |                    |         |
+# |                $h2 +         |
+# |  192.0.4.{2-62}/24        H2 |
+# +------------------------------+
+
+ALL_TESTS="
+	ping_ipv4
+	multipath_ipv4
+"
+
+NUM_NETIFS=10
+source lib.sh
+
+h1_create()
+{
+	simple_if_init $h1 192.0.3.2/24
+	ip route add vrf v$h1 192.0.4.0/24 via 192.0.3.1
+}
+
+h1_destroy()
+{
+	ip route del vrf v$h1 192.0.4.0/24 via 192.0.3.1
+	simple_if_fini $h1 192.0.3.2/24
+}
+
+sw1_create()
+{
+	simple_if_init $ol1 192.0.3.1/24
+	__simple_if_init $ul1 v$ol1 192.0.2.129/28
+
+	tunnel_create g1 gre 192.0.2.65 192.0.2.66 tos inherit dev v$ol1
+	__simple_if_init g1 v$ol1 192.0.2.65/32
+	ip route add vrf v$ol1 192.0.2.66/32 via 192.0.2.130
+
+	ip route add vrf v$ol1 192.0.4.0/24 nexthop dev g1
+}
+
+sw1_destroy()
+{
+	ip route del vrf v$ol1 192.0.4.0/24
+
+	ip route del vrf v$ol1 192.0.2.66/32
+	__simple_if_fini g1 192.0.2.65/32
+	tunnel_destroy g1
+
+	__simple_if_fini $ul1 192.0.2.129/28
+	simple_if_fini $ol1 192.0.3.1/24
+}
+
+sw2_create()
+{
+	simple_if_init $ul21 192.0.2.130/28
+	__simple_if_init $ul22 v$ul21
+	vlan_create $ul22 111 v$ul21 192.0.2.145/28
+	vlan_create $ul22 222 v$ul21 192.0.2.161/28
+
+	ip route add vrf v$ul21 192.0.2.65/32 via 192.0.2.129
+	ip route add vrf v$ul21 192.0.2.66/32 \
+	   nexthop via 192.0.2.146 \
+	   nexthop via 192.0.2.162
+}
+
+sw2_destroy()
+{
+	ip route del vrf v$ul21 192.0.2.66/32
+	ip route del vrf v$ul21 192.0.2.65/32
+
+	vlan_destroy $ul22 222
+	vlan_destroy $ul22 111
+	__simple_if_fini $ul22
+	simple_if_fini $ul21 192.0.2.130/28
+}
+
+sw3_create()
+{
+	simple_if_init $ul31 192.0.2.177/28
+	__simple_if_init $ul32 v$ul31
+	vlan_create $ul32 111 v$ul31 192.0.2.146/28
+	vlan_create $ul32 222 v$ul31 192.0.2.162/28
+
+	ip route add vrf v$ul31 192.0.2.66/32 via 192.0.2.178
+	ip route add vrf v$ul31 192.0.2.65/32 \
+	   nexthop via 192.0.2.145 \
+	   nexthop via 192.0.2.161
+
+	tc qdisc add dev $ul32 clsact
+	tc filter add dev $ul32 ingress pref 111 prot 802.1Q \
+	   flower vlan_id 111 action pass
+	tc filter add dev $ul32 ingress pref 222 prot 802.1Q \
+	   flower vlan_id 222 action pass
+}
+
+sw3_destroy()
+{
+	tc qdisc del dev $ul32 clsact
+
+	ip route del vrf v$ul31 192.0.2.65/32
+	ip route del vrf v$ul31 192.0.2.66/32
+
+	vlan_destroy $ul32 222
+	vlan_destroy $ul32 111
+	__simple_if_fini $ul32
+	simple_if_fini $ul31 192.0.2.177/28
+}
+
+sw4_create()
+{
+	simple_if_init $ol4 192.0.4.1/24
+	__simple_if_init $ul4 v$ol4 192.0.2.178/28
+
+	tunnel_create g2 gre 192.0.2.66 192.0.2.65 tos inherit dev v$ol4
+	__simple_if_init g2 v$ol4 192.0.2.66/32
+	ip route add vrf v$ol4 192.0.2.65/32 via 192.0.2.177
+
+	ip route add vrf v$ol4 192.0.3.0/24 nexthop dev g2
+}
+
+sw4_destroy()
+{
+	ip route del vrf v$ol4 192.0.3.0/24
+
+	ip route del vrf v$ol4 192.0.2.65/32
+	__simple_if_fini g2 192.0.2.66/32
+	tunnel_destroy g2
+
+	__simple_if_fini $ul4 192.0.2.178/28
+	simple_if_fini $ol4 192.0.4.1/24
+}
+
+h2_create()
+{
+	simple_if_init $h2 192.0.4.2/24
+	ip route add vrf v$h2 192.0.3.0/24 via 192.0.4.1
+}
+
+h2_destroy()
+{
+	ip route del vrf v$h2 192.0.3.0/24 via 192.0.4.1
+	simple_if_fini $h2 192.0.4.2/24
+}
+
+setup_prepare()
+{
+	h1=${NETIFS[p1]}
+
+	ol1=${NETIFS[p2]}
+	ul1=${NETIFS[p3]}
+
+	ul21=${NETIFS[p4]}
+	ul22=${NETIFS[p5]}
+
+	ul32=${NETIFS[p6]}
+	ul31=${NETIFS[p7]}
+
+	ul4=${NETIFS[p8]}
+	ol4=${NETIFS[p9]}
+
+	h2=${NETIFS[p10]}
+
+	vrf_prepare
+	h1_create
+	sw1_create
+	sw2_create
+	sw3_create
+	sw4_create
+	h2_create
+
+	forwarding_enable
+}
+
+cleanup()
+{
+	pre_cleanup
+
+	forwarding_restore
+
+	h2_destroy
+	sw4_destroy
+	sw3_destroy
+	sw2_destroy
+	sw1_destroy
+	h1_destroy
+	vrf_cleanup
+}
+
+multipath4_test()
+{
+	local what=$1; shift
+	local weight1=$1; shift
+	local weight2=$1; shift
+
+	sysctl_set net.ipv4.fib_multipath_hash_policy 2
+	ip route replace vrf v$ul21 192.0.2.66/32 \
+	   nexthop via 192.0.2.146 weight $weight1 \
+	   nexthop via 192.0.2.162 weight $weight2
+
+	local t0_111=$(tc_rule_stats_get $ul32 111 ingress)
+	local t0_222=$(tc_rule_stats_get $ul32 222 ingress)
+
+	ip vrf exec v$h1 \
+	   $MZ $h1 -q -p 64 -A "192.0.3.2-192.0.3.62" -B "192.0.4.2-192.0.4.62" \
+	       -d 1msec -c 50 -t udp "sp=1024,dp=1024"
+	sleep 1
+
+	local t1_111=$(tc_rule_stats_get $ul32 111 ingress)
+	local t1_222=$(tc_rule_stats_get $ul32 222 ingress)
+
+	local d111=$((t1_111 - t0_111))
+	local d222=$((t1_222 - t0_222))
+	multipath_eval "$what" $weight1 $weight2 $d111 $d222
+
+	ip route replace vrf v$ul21 192.0.2.66/32 \
+	   nexthop via 192.0.2.146 \
+	   nexthop via 192.0.2.162
+	sysctl_restore net.ipv4.fib_multipath_hash_policy
+}
+
+ping_ipv4()
+{
+	ping_test $h1 192.0.4.2
+}
+
+multipath_ipv4()
+{
+	log_info "Running IPv4 over GRE over IPv4 multipath tests"
+	multipath4_test "ECMP" 1 1
+	multipath4_test "Weighted MP 2:1" 2 1
+	multipath4_test "Weighted MP 11:45" 11 45
+}
+
+trap cleanup EXIT
+
+setup_prepare
+setup_wait
+tests_run
+
+exit $EXIT_STATUS
tools/testing/selftests/net/forwarding/gre_inner_v6_multipath.sh new file +306 -0

306 lines not on this page · raw message

tools/testing/selftests/net/forwarding/ip6gre_inner_v4_multipath.sh new file +304 -0

304 lines not on this page · raw message

tools/testing/selftests/net/forwarding/ip6gre_inner_v6_multipath.sh new file +305 -0

305 lines not on this page · raw message

Signature · 1 line
2.17.1
maintainer
#4 lore raw
In reply to Stephen Suryaputra

From: Stephen Suryaputra <[email protected]>
Date: Sat, 6 Jul 2019 10:55:16 -0400

Quoted 9 lines

This series extends commit 363887a2cdfe ("ipv4: Support multipath hashing on inner IP pkts for GRE tunnel") to include support when the outer L3 is IPv6 and to consider the case where the inner L3 is different version from the outer L3, such as IPv6 tunneled by IPv4 GRE or vice versa. It also includes kselftest scripts to test the use cases.

v2: Clarify the commit messages in the commits in this series to use the
term tunneled by IPv4 GRE or by IPv6 GRE so that it's clear which one is the inner and which one is the outer (per David Miller).

Series applied, thanks.