1#!/bin/sh
2# SPDX-License-Identifier: GPL-2.0
3#
4# Check that route PMTU values match expectations, and that initial device MTU
5# values are assigned correctly
6#
7# Tests currently implemented:
8#
9# - pmtu_ipv4
10#	Set up two namespaces, A and B, with two paths between them over routers
11#	R1 and R2 (also implemented with namespaces), with different MTUs:
12#
13#	  segment a_r1    segment b_r1		a_r1: 2000
14#	.--------------R1--------------.	b_r1: 1400
15#	A                               B	a_r2: 2000
16#	'--------------R2--------------'	b_r2: 1500
17#	  segment a_r2    segment b_r2
18#
19#	Check that PMTU exceptions with the correct PMTU are created. Then
20#	decrease and increase the MTU of the local link for one of the paths,
21#	A to R1, checking that route exception PMTU changes accordingly over
22#	this path. Also check that locked exceptions are created when an ICMP
23#	message advertising a PMTU smaller than net.ipv4.route.min_pmtu is
24#	received
25#
26# - pmtu_ipv6
27#	Same as pmtu_ipv4, except for locked PMTU tests, using IPv6
28#
29# - pmtu_ipv4_vxlan4_exception
30#	Set up the same network topology as pmtu_ipv4, create a VXLAN tunnel
31#	over IPv4 between A and B, routed via R1. On the link between R1 and B,
32#	set a MTU lower than the VXLAN MTU and the MTU on the link between A and
33#	R1. Send IPv4 packets, exceeding the MTU between R1 and B, over VXLAN
34#	from A to B and check that the PMTU exception is created with the right
35#	value on A
36#
37# - pmtu_ipv6_vxlan4_exception
38#	Same as pmtu_ipv4_vxlan4_exception, but send IPv6 packets from A to B
39#
40# - pmtu_ipv4_vxlan6_exception
41#	Same as pmtu_ipv4_vxlan4_exception, but use IPv6 transport from A to B
42#
43# - pmtu_ipv6_vxlan6_exception
44#	Same as pmtu_ipv4_vxlan6_exception, but send IPv6 packets from A to B
45#
46# - pmtu_ipv4_geneve4_exception
47#	Same as pmtu_ipv4_vxlan4_exception, but using a GENEVE tunnel instead of
48#	VXLAN
49#
50# - pmtu_ipv6_geneve4_exception
51#	Same as pmtu_ipv6_vxlan4_exception, but using a GENEVE tunnel instead of
52#	VXLAN
53#
54# - pmtu_ipv4_geneve6_exception
55#	Same as pmtu_ipv4_vxlan6_exception, but using a GENEVE tunnel instead of
56#	VXLAN
57#
58# - pmtu_ipv6_geneve6_exception
59#	Same as pmtu_ipv6_vxlan6_exception, but using a GENEVE tunnel instead of
60#	VXLAN
61#
62# - pmtu_ipv{4,6}_br_vxlan{4,6}_exception
63#	Set up three namespaces, A, B, and C, with routing between A and B over
64#	R1. R2 is unused in these tests. A has a veth connection to C, and is
65#	connected to B via a VXLAN endpoint, which is directly bridged to C.
66#	MTU on the B-R1 link is lower than other MTUs.
67#
68#	Check that both C and A are able to communicate with B over the VXLAN
69#	tunnel, and that PMTU exceptions with the correct values are created.
70#
71#	                  segment a_r1    segment b_r1            b_r1: 4000
72#	                .--------------R1--------------.    everything
73#	   C---veth     A                               B         else: 5000
74#	        ' bridge                                |
75#	            '---- - - - - - VXLAN - - - - - - - '
76#
77# - pmtu_ipv{4,6}_br_geneve{4,6}_exception
78#	Same as pmtu_ipv{4,6}_br_vxlan{4,6}_exception, with a GENEVE tunnel
79#	instead.
80#
81# - pmtu_ipv{4,6}_ovs_vxlan{4,6}_exception
82#	Set up two namespaces, B, and C, with routing between the init namespace
83#	and B over R1. A and R2 are unused in these tests. The init namespace
84#	has a veth connection to C, and is connected to B via a VXLAN endpoint,
85#	which is handled by Open vSwitch and bridged to C. MTU on the B-R1 link
86#	is lower than other MTUs.
87#
88#	Check that C is able to communicate with B over the VXLAN tunnel, and
89#	that PMTU exceptions with the correct values are created.
90#
91#	                  segment a_r1    segment b_r1            b_r1: 4000
92#	                .--------------R1--------------.    everything
93#	   C---veth    init                             B         else: 5000
94#	        '- ovs                                  |
95#	            '---- - - - - - VXLAN - - - - - - - '
96#
97# - pmtu_ipv{4,6}_ovs_geneve{4,6}_exception
98#	Same as pmtu_ipv{4,6}_ovs_vxlan{4,6}_exception, with a GENEVE tunnel
99#	instead.
100#
101# - pmtu_ipv{4,6}_fou{4,6}_exception
102#	Same as pmtu_ipv4_vxlan4, but using a direct IPv4/IPv6 encapsulation
103#	(FoU) over IPv4/IPv6, instead of VXLAN
104#
105# - pmtu_ipv{4,6}_fou{4,6}_exception
106#	Same as pmtu_ipv4_vxlan4, but using a generic UDP IPv4/IPv6
107#	encapsulation (GUE) over IPv4/IPv6, instead of VXLAN
108#
109# - pmtu_ipv{4,6}_ipv{4,6}_exception
110#	Same as pmtu_ipv4_vxlan4, but using a IPv4/IPv6 tunnel over IPv4/IPv6,
111#	instead of VXLAN
112#
113# - pmtu_vti4_exception
114#	Set up vti tunnel on top of veth, with xfrm states and policies, in two
115#	namespaces with matching endpoints. Check that route exception is not
116#	created if link layer MTU is not exceeded, then exceed it and check that
117#	exception is created with the expected PMTU. The approach described
118#	below for IPv6 doesn't apply here, because, on IPv4, administrative MTU
119#	changes alone won't affect PMTU
120#
121# - pmtu_vti6_exception
122#	Set up vti6 tunnel on top of veth, with xfrm states and policies, in two
123#	namespaces with matching endpoints. Check that route exception is
124#	created by exceeding link layer MTU with ping to other endpoint. Then
125#	decrease and increase MTU of tunnel, checking that route exception PMTU
126#	changes accordingly
127#
128# - pmtu_vti4_default_mtu
129#	Set up vti4 tunnel on top of veth, in two namespaces with matching
130#	endpoints. Check that MTU assigned to vti interface is the MTU of the
131#	lower layer (veth) minus additional lower layer headers (zero, for veth)
132#	minus IPv4 header length
133#
134# - pmtu_vti6_default_mtu
135#	Same as above, for IPv6
136#
137# - pmtu_vti4_link_add_mtu
138#	Set up vti4 interface passing MTU value at link creation, check MTU is
139#	configured, and that link is not created with invalid MTU values
140#
141# - pmtu_vti6_link_add_mtu
142#	Same as above, for IPv6
143#
144# - pmtu_vti6_link_change_mtu
145#	Set up two dummy interfaces with different MTUs, create a vti6 tunnel
146#	and check that configured MTU is used on link creation and changes, and
147#	that MTU is properly calculated instead when MTU is not configured from
148#	userspace
149#
150# - cleanup_ipv4_exception
151#	Similar to pmtu_ipv4_vxlan4_exception, but explicitly generate PMTU
152#	exceptions on multiple CPUs and check that the veth device tear-down
153# 	happens in a timely manner
154#
155# - cleanup_ipv6_exception
156#	Same as above, but use IPv6 transport from A to B
157#
158# - list_flush_ipv4_exception
159#	Using the same topology as in pmtu_ipv4, create exceptions, and check
160#	they are shown when listing exception caches, gone after flushing them
161#
162# - list_flush_ipv6_exception
163#	Using the same topology as in pmtu_ipv6, create exceptions, and check
164#	they are shown when listing exception caches, gone after flushing them
165
166
167# Kselftest framework requirement - SKIP code is 4.
168ksft_skip=4
169
170PAUSE_ON_FAIL=no
171VERBOSE=0
172TRACING=0
173
174# Some systems don't have a ping6 binary anymore
175which ping6 > /dev/null 2>&1 && ping6=$(which ping6) || ping6=$(which ping)
176
177#               Name                          Description                  re-run with nh
178tests="
179	pmtu_ipv4_exception		ipv4: PMTU exceptions			1
180	pmtu_ipv6_exception		ipv6: PMTU exceptions			1
181	pmtu_ipv4_vxlan4_exception	IPv4 over vxlan4: PMTU exceptions	1
182	pmtu_ipv6_vxlan4_exception	IPv6 over vxlan4: PMTU exceptions	1
183	pmtu_ipv4_vxlan6_exception	IPv4 over vxlan6: PMTU exceptions	1
184	pmtu_ipv6_vxlan6_exception	IPv6 over vxlan6: PMTU exceptions	1
185	pmtu_ipv4_geneve4_exception	IPv4 over geneve4: PMTU exceptions	1
186	pmtu_ipv6_geneve4_exception	IPv6 over geneve4: PMTU exceptions	1
187	pmtu_ipv4_geneve6_exception	IPv4 over geneve6: PMTU exceptions	1
188	pmtu_ipv6_geneve6_exception	IPv6 over geneve6: PMTU exceptions	1
189	pmtu_ipv4_br_vxlan4_exception	IPv4, bridged vxlan4: PMTU exceptions	1
190	pmtu_ipv6_br_vxlan4_exception	IPv6, bridged vxlan4: PMTU exceptions	1
191	pmtu_ipv4_br_vxlan6_exception	IPv4, bridged vxlan6: PMTU exceptions	1
192	pmtu_ipv6_br_vxlan6_exception	IPv6, bridged vxlan6: PMTU exceptions	1
193	pmtu_ipv4_br_geneve4_exception	IPv4, bridged geneve4: PMTU exceptions	1
194	pmtu_ipv6_br_geneve4_exception	IPv6, bridged geneve4: PMTU exceptions	1
195	pmtu_ipv4_br_geneve6_exception	IPv4, bridged geneve6: PMTU exceptions	1
196	pmtu_ipv6_br_geneve6_exception	IPv6, bridged geneve6: PMTU exceptions	1
197	pmtu_ipv4_ovs_vxlan4_exception	IPv4, OVS vxlan4: PMTU exceptions	1
198	pmtu_ipv6_ovs_vxlan4_exception	IPv6, OVS vxlan4: PMTU exceptions	1
199	pmtu_ipv4_ovs_vxlan6_exception	IPv4, OVS vxlan6: PMTU exceptions	1
200	pmtu_ipv6_ovs_vxlan6_exception	IPv6, OVS vxlan6: PMTU exceptions	1
201	pmtu_ipv4_ovs_geneve4_exception	IPv4, OVS geneve4: PMTU exceptions	1
202	pmtu_ipv6_ovs_geneve4_exception	IPv6, OVS geneve4: PMTU exceptions	1
203	pmtu_ipv4_ovs_geneve6_exception	IPv4, OVS geneve6: PMTU exceptions	1
204	pmtu_ipv6_ovs_geneve6_exception	IPv6, OVS geneve6: PMTU exceptions	1
205	pmtu_ipv4_fou4_exception	IPv4 over fou4: PMTU exceptions		1
206	pmtu_ipv6_fou4_exception	IPv6 over fou4: PMTU exceptions		1
207	pmtu_ipv4_fou6_exception	IPv4 over fou6: PMTU exceptions		1
208	pmtu_ipv6_fou6_exception	IPv6 over fou6: PMTU exceptions		1
209	pmtu_ipv4_gue4_exception	IPv4 over gue4: PMTU exceptions		1
210	pmtu_ipv6_gue4_exception	IPv6 over gue4: PMTU exceptions		1
211	pmtu_ipv4_gue6_exception	IPv4 over gue6: PMTU exceptions		1
212	pmtu_ipv6_gue6_exception	IPv6 over gue6: PMTU exceptions		1
213	pmtu_ipv4_ipv4_exception	IPv4 over IPv4: PMTU exceptions		1
214	pmtu_ipv6_ipv4_exception	IPv6 over IPv4: PMTU exceptions		1
215	pmtu_ipv4_ipv6_exception	IPv4 over IPv6: PMTU exceptions		1
216	pmtu_ipv6_ipv6_exception	IPv6 over IPv6: PMTU exceptions		1
217	pmtu_vti6_exception		vti6: PMTU exceptions			0
218	pmtu_vti4_exception		vti4: PMTU exceptions			0
219	pmtu_vti4_default_mtu		vti4: default MTU assignment		0
220	pmtu_vti6_default_mtu		vti6: default MTU assignment		0
221	pmtu_vti4_link_add_mtu		vti4: MTU setting on link creation	0
222	pmtu_vti6_link_add_mtu		vti6: MTU setting on link creation	0
223	pmtu_vti6_link_change_mtu	vti6: MTU changes on link changes	0
224	cleanup_ipv4_exception		ipv4: cleanup of cached exceptions	1
225	cleanup_ipv6_exception		ipv6: cleanup of cached exceptions	1
226	list_flush_ipv4_exception	ipv4: list and flush cached exceptions	1
227	list_flush_ipv6_exception	ipv6: list and flush cached exceptions	1"
228
229NS_A="ns-A"
230NS_B="ns-B"
231NS_C="ns-C"
232NS_R1="ns-R1"
233NS_R2="ns-R2"
234ns_a="ip netns exec ${NS_A}"
235ns_b="ip netns exec ${NS_B}"
236ns_c="ip netns exec ${NS_C}"
237ns_r1="ip netns exec ${NS_R1}"
238ns_r2="ip netns exec ${NS_R2}"
239
240# Addressing and routing for tests with routers: four network segments, with
241# index SEGMENT between 1 and 4, a common prefix (PREFIX4 or PREFIX6) and an
242# identifier ID, which is 1 for hosts (A and B), 2 for routers (R1 and R2).
243# Addresses are:
244# - IPv4: PREFIX4.SEGMENT.ID (/24)
245# - IPv6: PREFIX6:SEGMENT::ID (/64)
246prefix4="10.0"
247prefix6="fc00"
248a_r1=1
249a_r2=2
250b_r1=3
251b_r2=4
252#	ns	peer	segment
253routing_addrs="
254	A	R1	${a_r1}
255	A	R2	${a_r2}
256	B	R1	${b_r1}
257	B	R2	${b_r2}
258"
259# Traffic from A to B goes through R1 by default, and through R2, if destined to
260# B's address on the b_r2 segment.
261# Traffic from B to A goes through R1.
262#	ns	destination		gateway
263routes="
264	A	default			${prefix4}.${a_r1}.2
265	A	${prefix4}.${b_r2}.1	${prefix4}.${a_r2}.2
266	B	default			${prefix4}.${b_r1}.2
267
268	A	default			${prefix6}:${a_r1}::2
269	A	${prefix6}:${b_r2}::1	${prefix6}:${a_r2}::2
270	B	default			${prefix6}:${b_r1}::2
271"
272
273USE_NH="no"
274#	ns	family	nh id	   destination		gateway
275nexthops="
276	A	4	41	${prefix4}.${a_r1}.2	veth_A-R1
277	A	4	42	${prefix4}.${a_r2}.2	veth_A-R2
278	B	4	41	${prefix4}.${b_r1}.2	veth_B-R1
279
280	A	6	61	${prefix6}:${a_r1}::2	veth_A-R1
281	A	6	62	${prefix6}:${a_r2}::2	veth_A-R2
282	B	6	61	${prefix6}:${b_r1}::2	veth_B-R1
283"
284
285# nexthop id correlates to id in nexthops config above
286#	ns    family	prefix			nh id
287routes_nh="
288	A	4	default			41
289	A	4	${prefix4}.${b_r2}.1	42
290	B	4	default			41
291
292	A	6	default			61
293	A	6	${prefix6}:${b_r2}::1	62
294	B	6	default			61
295"
296
297veth4_a_addr="192.168.1.1"
298veth4_b_addr="192.168.1.2"
299veth4_c_addr="192.168.2.10"
300veth4_mask="24"
301veth6_a_addr="fd00:1::a"
302veth6_b_addr="fd00:1::b"
303veth6_c_addr="fd00:2::c"
304veth6_mask="64"
305
306tunnel4_a_addr="192.168.2.1"
307tunnel4_b_addr="192.168.2.2"
308tunnel4_mask="24"
309tunnel6_a_addr="fd00:2::a"
310tunnel6_b_addr="fd00:2::b"
311tunnel6_mask="64"
312
313dummy6_0_prefix="fc00:1000::"
314dummy6_1_prefix="fc00:1001::"
315dummy6_mask="64"
316
317err_buf=
318tcpdump_pids=
319
320err() {
321	err_buf="${err_buf}${1}
322"
323}
324
325err_flush() {
326	echo -n "${err_buf}"
327	err_buf=
328}
329
330run_cmd() {
331	cmd="$*"
332
333	if [ "$VERBOSE" = "1" ]; then
334		printf "    COMMAND: $cmd\n"
335	fi
336
337	out="$($cmd 2>&1)"
338	rc=$?
339	if [ "$VERBOSE" = "1" -a -n "$out" ]; then
340		echo "    $out"
341		echo
342	fi
343
344	return $rc
345}
346
347# Find the auto-generated name for this namespace
348nsname() {
349	eval echo \$NS_$1
350}
351
352setup_fou_or_gue() {
353	outer="${1}"
354	inner="${2}"
355	encap="${3}"
356
357	if [ "${outer}" = "4" ]; then
358		modprobe fou || return 2
359		a_addr="${prefix4}.${a_r1}.1"
360		b_addr="${prefix4}.${b_r1}.1"
361		if [ "${inner}" = "4" ]; then
362			type="ipip"
363			ipproto="4"
364		else
365			type="sit"
366			ipproto="41"
367		fi
368	else
369		modprobe fou6 || return 2
370		a_addr="${prefix6}:${a_r1}::1"
371		b_addr="${prefix6}:${b_r1}::1"
372		if [ "${inner}" = "4" ]; then
373			type="ip6tnl"
374			mode="mode ipip6"
375			ipproto="4 -6"
376		else
377			type="ip6tnl"
378			mode="mode ip6ip6"
379			ipproto="41 -6"
380		fi
381	fi
382
383	run_cmd ${ns_a} ip fou add port 5555 ipproto ${ipproto} || return 2
384	run_cmd ${ns_a} ip link add ${encap}_a type ${type} ${mode} local ${a_addr} remote ${b_addr} encap ${encap} encap-sport auto encap-dport 5556 || return 2
385
386	run_cmd ${ns_b} ip fou add port 5556 ipproto ${ipproto}
387	run_cmd ${ns_b} ip link add ${encap}_b type ${type} ${mode} local ${b_addr} remote ${a_addr} encap ${encap} encap-sport auto encap-dport 5555
388
389	if [ "${inner}" = "4" ]; then
390		run_cmd ${ns_a} ip addr add ${tunnel4_a_addr}/${tunnel4_mask} dev ${encap}_a
391		run_cmd ${ns_b} ip addr add ${tunnel4_b_addr}/${tunnel4_mask} dev ${encap}_b
392	else
393		run_cmd ${ns_a} ip addr add ${tunnel6_a_addr}/${tunnel6_mask} dev ${encap}_a
394		run_cmd ${ns_b} ip addr add ${tunnel6_b_addr}/${tunnel6_mask} dev ${encap}_b
395	fi
396
397	run_cmd ${ns_a} ip link set ${encap}_a up
398	run_cmd ${ns_b} ip link set ${encap}_b up
399}
400
401setup_fou44() {
402	setup_fou_or_gue 4 4 fou
403}
404
405setup_fou46() {
406	setup_fou_or_gue 4 6 fou
407}
408
409setup_fou64() {
410	setup_fou_or_gue 6 4 fou
411}
412
413setup_fou66() {
414	setup_fou_or_gue 6 6 fou
415}
416
417setup_gue44() {
418	setup_fou_or_gue 4 4 gue
419}
420
421setup_gue46() {
422	setup_fou_or_gue 4 6 gue
423}
424
425setup_gue64() {
426	setup_fou_or_gue 6 4 gue
427}
428
429setup_gue66() {
430	setup_fou_or_gue 6 6 gue
431}
432
433setup_ipvX_over_ipvY() {
434	inner=${1}
435	outer=${2}
436
437	if [ "${outer}" -eq 4 ]; then
438		a_addr="${prefix4}.${a_r1}.1"
439		b_addr="${prefix4}.${b_r1}.1"
440		if [ "${inner}" -eq 4 ]; then
441			type="ipip"
442			mode="ipip"
443		else
444			type="sit"
445			mode="ip6ip"
446		fi
447	else
448		a_addr="${prefix6}:${a_r1}::1"
449		b_addr="${prefix6}:${b_r1}::1"
450		type="ip6tnl"
451		if [ "${inner}" -eq 4 ]; then
452			mode="ipip6"
453		else
454			mode="ip6ip6"
455		fi
456	fi
457
458	run_cmd ${ns_a} ip link add ip_a type ${type} local ${a_addr} remote ${b_addr} mode ${mode} || return 2
459	run_cmd ${ns_b} ip link add ip_b type ${type} local ${b_addr} remote ${a_addr} mode ${mode}
460
461	run_cmd ${ns_a} ip link set ip_a up
462	run_cmd ${ns_b} ip link set ip_b up
463
464	if [ "${inner}" = "4" ]; then
465		run_cmd ${ns_a} ip addr add ${tunnel4_a_addr}/${tunnel4_mask} dev ip_a
466		run_cmd ${ns_b} ip addr add ${tunnel4_b_addr}/${tunnel4_mask} dev ip_b
467	else
468		run_cmd ${ns_a} ip addr add ${tunnel6_a_addr}/${tunnel6_mask} dev ip_a
469		run_cmd ${ns_b} ip addr add ${tunnel6_b_addr}/${tunnel6_mask} dev ip_b
470	fi
471}
472
473setup_ip4ip4() {
474	setup_ipvX_over_ipvY 4 4
475}
476
477setup_ip6ip4() {
478	setup_ipvX_over_ipvY 6 4
479}
480
481setup_ip4ip6() {
482	setup_ipvX_over_ipvY 4 6
483}
484
485setup_ip6ip6() {
486	setup_ipvX_over_ipvY 6 6
487}
488
489setup_namespaces() {
490	for n in ${NS_A} ${NS_B} ${NS_C} ${NS_R1} ${NS_R2}; do
491		ip netns add ${n} || return 1
492
493		# Disable DAD, so that we don't have to wait to use the
494		# configured IPv6 addresses
495		ip netns exec ${n} sysctl -q net/ipv6/conf/default/accept_dad=0
496	done
497}
498
499setup_veth() {
500	run_cmd ${ns_a} ip link add veth_a type veth peer name veth_b || return 1
501	run_cmd ${ns_a} ip link set veth_b netns ${NS_B}
502
503	run_cmd ${ns_a} ip addr add ${veth4_a_addr}/${veth4_mask} dev veth_a
504	run_cmd ${ns_b} ip addr add ${veth4_b_addr}/${veth4_mask} dev veth_b
505
506	run_cmd ${ns_a} ip addr add ${veth6_a_addr}/${veth6_mask} dev veth_a
507	run_cmd ${ns_b} ip addr add ${veth6_b_addr}/${veth6_mask} dev veth_b
508
509	run_cmd ${ns_a} ip link set veth_a up
510	run_cmd ${ns_b} ip link set veth_b up
511}
512
513setup_vti() {
514	proto=${1}
515	veth_a_addr="${2}"
516	veth_b_addr="${3}"
517	vti_a_addr="${4}"
518	vti_b_addr="${5}"
519	vti_mask=${6}
520
521	[ ${proto} -eq 6 ] && vti_type="vti6" || vti_type="vti"
522
523	run_cmd ${ns_a} ip link add vti${proto}_a type ${vti_type} local ${veth_a_addr} remote ${veth_b_addr} key 10 || return 1
524	run_cmd ${ns_b} ip link add vti${proto}_b type ${vti_type} local ${veth_b_addr} remote ${veth_a_addr} key 10
525
526	run_cmd ${ns_a} ip addr add ${vti_a_addr}/${vti_mask} dev vti${proto}_a
527	run_cmd ${ns_b} ip addr add ${vti_b_addr}/${vti_mask} dev vti${proto}_b
528
529	run_cmd ${ns_a} ip link set vti${proto}_a up
530	run_cmd ${ns_b} ip link set vti${proto}_b up
531}
532
533setup_vti4() {
534	setup_vti 4 ${veth4_a_addr} ${veth4_b_addr} ${tunnel4_a_addr} ${tunnel4_b_addr} ${tunnel4_mask}
535}
536
537setup_vti6() {
538	setup_vti 6 ${veth6_a_addr} ${veth6_b_addr} ${tunnel6_a_addr} ${tunnel6_b_addr} ${tunnel6_mask}
539}
540
541setup_vxlan_or_geneve() {
542	type="${1}"
543	a_addr="${2}"
544	b_addr="${3}"
545	opts="${4}"
546	br_if_a="${5}"
547
548	if [ "${type}" = "vxlan" ]; then
549		opts="${opts} ttl 64 dstport 4789"
550		opts_a="local ${a_addr}"
551		opts_b="local ${b_addr}"
552	else
553		opts_a=""
554		opts_b=""
555	fi
556
557	run_cmd ${ns_a} ip link add ${type}_a type ${type} id 1 ${opts_a} remote ${b_addr} ${opts} || return 1
558	run_cmd ${ns_b} ip link add ${type}_b type ${type} id 1 ${opts_b} remote ${a_addr} ${opts}
559
560	if [ -n "${br_if_a}" ]; then
561		run_cmd ${ns_a} ip addr add ${tunnel4_a_addr}/${tunnel4_mask} dev ${br_if_a}
562		run_cmd ${ns_a} ip addr add ${tunnel6_a_addr}/${tunnel6_mask} dev ${br_if_a}
563		run_cmd ${ns_a} ip link set ${type}_a master ${br_if_a}
564	else
565		run_cmd ${ns_a} ip addr add ${tunnel4_a_addr}/${tunnel4_mask} dev ${type}_a
566		run_cmd ${ns_a} ip addr add ${tunnel6_a_addr}/${tunnel6_mask} dev ${type}_a
567	fi
568
569	run_cmd ${ns_b} ip addr add ${tunnel4_b_addr}/${tunnel4_mask} dev ${type}_b
570	run_cmd ${ns_b} ip addr add ${tunnel6_b_addr}/${tunnel6_mask} dev ${type}_b
571
572	run_cmd ${ns_a} ip link set ${type}_a up
573	run_cmd ${ns_b} ip link set ${type}_b up
574}
575
576setup_geneve4() {
577	setup_vxlan_or_geneve geneve ${prefix4}.${a_r1}.1  ${prefix4}.${b_r1}.1  "df set"
578}
579
580setup_vxlan4() {
581	setup_vxlan_or_geneve vxlan  ${prefix4}.${a_r1}.1  ${prefix4}.${b_r1}.1  "df set"
582}
583
584setup_geneve6() {
585	setup_vxlan_or_geneve geneve ${prefix6}:${a_r1}::1 ${prefix6}:${b_r1}::1 ""
586}
587
588setup_vxlan6() {
589	setup_vxlan_or_geneve vxlan  ${prefix6}:${a_r1}::1 ${prefix6}:${b_r1}::1 ""
590}
591
592setup_bridged_geneve4() {
593	setup_vxlan_or_geneve geneve ${prefix4}.${a_r1}.1  ${prefix4}.${b_r1}.1  "df set" "br0"
594}
595
596setup_bridged_vxlan4() {
597	setup_vxlan_or_geneve vxlan  ${prefix4}.${a_r1}.1  ${prefix4}.${b_r1}.1  "df set" "br0"
598}
599
600setup_bridged_geneve6() {
601	setup_vxlan_or_geneve geneve ${prefix6}:${a_r1}::1 ${prefix6}:${b_r1}::1 "" "br0"
602}
603
604setup_bridged_vxlan6() {
605	setup_vxlan_or_geneve vxlan  ${prefix6}:${a_r1}::1 ${prefix6}:${b_r1}::1 "" "br0"
606}
607
608setup_xfrm() {
609	proto=${1}
610	veth_a_addr="${2}"
611	veth_b_addr="${3}"
612
613	run_cmd ${ns_a} ip -${proto} xfrm state add src ${veth_a_addr} dst ${veth_b_addr} spi 0x1000 proto esp aead 'rfc4106(gcm(aes))' 0x0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f 128 mode tunnel || return 1
614	run_cmd ${ns_a} ip -${proto} xfrm state add src ${veth_b_addr} dst ${veth_a_addr} spi 0x1001 proto esp aead 'rfc4106(gcm(aes))' 0x0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f 128 mode tunnel
615	run_cmd ${ns_a} ip -${proto} xfrm policy add dir out mark 10 tmpl src ${veth_a_addr} dst ${veth_b_addr} proto esp mode tunnel
616	run_cmd ${ns_a} ip -${proto} xfrm policy add dir in mark 10 tmpl src ${veth_b_addr} dst ${veth_a_addr} proto esp mode tunnel
617
618	run_cmd ${ns_b} ip -${proto} xfrm state add src ${veth_a_addr} dst ${veth_b_addr} spi 0x1000 proto esp aead 'rfc4106(gcm(aes))' 0x0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f 128 mode tunnel
619	run_cmd ${ns_b} ip -${proto} xfrm state add src ${veth_b_addr} dst ${veth_a_addr} spi 0x1001 proto esp aead 'rfc4106(gcm(aes))' 0x0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f0f 128 mode tunnel
620	run_cmd ${ns_b} ip -${proto} xfrm policy add dir out mark 10 tmpl src ${veth_b_addr} dst ${veth_a_addr} proto esp mode tunnel
621	run_cmd ${ns_b} ip -${proto} xfrm policy add dir in mark 10 tmpl src ${veth_a_addr} dst ${veth_b_addr} proto esp mode tunnel
622}
623
624setup_xfrm4() {
625	setup_xfrm 4 ${veth4_a_addr} ${veth4_b_addr}
626}
627
628setup_xfrm6() {
629	setup_xfrm 6 ${veth6_a_addr} ${veth6_b_addr}
630}
631
632setup_routing_old() {
633	for i in ${routes}; do
634		[ "${ns}" = "" ]	&& ns="${i}"		&& continue
635		[ "${addr}" = "" ]	&& addr="${i}"		&& continue
636		[ "${gw}" = "" ]	&& gw="${i}"
637
638		ns_name="$(nsname ${ns})"
639
640		ip -n ${ns_name} route add ${addr} via ${gw}
641
642		ns=""; addr=""; gw=""
643	done
644}
645
646setup_routing_new() {
647	for i in ${nexthops}; do
648		[ "${ns}" = "" ]	&& ns="${i}"		&& continue
649		[ "${fam}" = "" ]	&& fam="${i}"		&& continue
650		[ "${nhid}" = "" ]	&& nhid="${i}"		&& continue
651		[ "${gw}" = "" ]	&& gw="${i}"		&& continue
652		[ "${dev}" = "" ]	&& dev="${i}"
653
654		ns_name="$(nsname ${ns})"
655
656		ip -n ${ns_name} -${fam} nexthop add id ${nhid} via ${gw} dev ${dev}
657
658		ns=""; fam=""; nhid=""; gw=""; dev=""
659
660	done
661
662	for i in ${routes_nh}; do
663		[ "${ns}" = "" ]	&& ns="${i}"		&& continue
664		[ "${fam}" = "" ]	&& fam="${i}"		&& continue
665		[ "${addr}" = "" ]	&& addr="${i}"		&& continue
666		[ "${nhid}" = "" ]	&& nhid="${i}"
667
668		ns_name="$(nsname ${ns})"
669
670		ip -n ${ns_name} -${fam} route add ${addr} nhid ${nhid}
671
672		ns=""; fam=""; addr=""; nhid=""
673	done
674}
675
676setup_routing() {
677	for i in ${NS_R1} ${NS_R2}; do
678		ip netns exec ${i} sysctl -q net/ipv4/ip_forward=1
679		ip netns exec ${i} sysctl -q net/ipv6/conf/all/forwarding=1
680	done
681
682	for i in ${routing_addrs}; do
683		[ "${ns}" = "" ]	&& ns="${i}"		&& continue
684		[ "${peer}" = "" ]	&& peer="${i}"		&& continue
685		[ "${segment}" = "" ]	&& segment="${i}"
686
687		ns_name="$(nsname ${ns})"
688		peer_name="$(nsname ${peer})"
689		if="veth_${ns}-${peer}"
690		ifpeer="veth_${peer}-${ns}"
691
692		# Create veth links
693		ip link add ${if} up netns ${ns_name} type veth peer name ${ifpeer} netns ${peer_name} || return 1
694		ip -n ${peer_name} link set dev ${ifpeer} up
695
696		# Add addresses
697		ip -n ${ns_name}   addr add ${prefix4}.${segment}.1/24  dev ${if}
698		ip -n ${ns_name}   addr add ${prefix6}:${segment}::1/64 dev ${if}
699
700		ip -n ${peer_name} addr add ${prefix4}.${segment}.2/24  dev ${ifpeer}
701		ip -n ${peer_name} addr add ${prefix6}:${segment}::2/64 dev ${ifpeer}
702
703		ns=""; peer=""; segment=""
704	done
705
706	if [ "$USE_NH" = "yes" ]; then
707		setup_routing_new
708	else
709		setup_routing_old
710	fi
711
712	return 0
713}
714
715setup_bridge() {
716	run_cmd ${ns_a} ip link add br0 type bridge || return 2
717	run_cmd ${ns_a} ip link set br0 up
718
719	run_cmd ${ns_c} ip link add veth_C-A type veth peer name veth_A-C
720	run_cmd ${ns_c} ip link set veth_A-C netns ns-A
721
722	run_cmd ${ns_a} ip link set veth_A-C up
723	run_cmd ${ns_c} ip link set veth_C-A up
724	run_cmd ${ns_c} ip addr add ${veth4_c_addr}/${veth4_mask} dev veth_C-A
725	run_cmd ${ns_c} ip addr add ${veth6_c_addr}/${veth6_mask} dev veth_C-A
726	run_cmd ${ns_a} ip link set veth_A-C master br0
727}
728
729setup_ovs_vxlan_or_geneve() {
730	type="${1}"
731	a_addr="${2}"
732	b_addr="${3}"
733
734	if [ "${type}" = "vxlan" ]; then
735		opts="${opts} ttl 64 dstport 4789"
736		opts_b="local ${b_addr}"
737	fi
738
739	run_cmd ovs-vsctl add-port ovs_br0 ${type}_a -- \
740		set interface ${type}_a type=${type} \
741		options:remote_ip=${b_addr} options:key=1 options:csum=true || return 1
742
743	run_cmd ${ns_b} ip link add ${type}_b type ${type} id 1 ${opts_b} remote ${a_addr} ${opts} || return 1
744
745	run_cmd ${ns_b} ip addr add ${tunnel4_b_addr}/${tunnel4_mask} dev ${type}_b
746	run_cmd ${ns_b} ip addr add ${tunnel6_b_addr}/${tunnel6_mask} dev ${type}_b
747
748	run_cmd ${ns_b} ip link set ${type}_b up
749}
750
751setup_ovs_geneve4() {
752	setup_ovs_vxlan_or_geneve geneve ${prefix4}.${a_r1}.1  ${prefix4}.${b_r1}.1
753}
754
755setup_ovs_vxlan4() {
756	setup_ovs_vxlan_or_geneve vxlan  ${prefix4}.${a_r1}.1  ${prefix4}.${b_r1}.1
757}
758
759setup_ovs_geneve6() {
760	setup_ovs_vxlan_or_geneve geneve ${prefix6}:${a_r1}::1 ${prefix6}:${b_r1}::1
761}
762
763setup_ovs_vxlan6() {
764	setup_ovs_vxlan_or_geneve vxlan  ${prefix6}:${a_r1}::1 ${prefix6}:${b_r1}::1
765}
766
767setup_ovs_bridge() {
768	run_cmd ovs-vsctl add-br ovs_br0 || return 2
769	run_cmd ip link set ovs_br0 up
770
771	run_cmd ${ns_c} ip link add veth_C-A type veth peer name veth_A-C
772	run_cmd ${ns_c} ip link set veth_A-C netns 1
773
774	run_cmd         ip link set veth_A-C up
775	run_cmd ${ns_c} ip link set veth_C-A up
776	run_cmd ${ns_c} ip addr add ${veth4_c_addr}/${veth4_mask} dev veth_C-A
777	run_cmd ${ns_c} ip addr add ${veth6_c_addr}/${veth6_mask} dev veth_C-A
778	run_cmd ovs-vsctl add-port ovs_br0 veth_A-C
779
780	# Move veth_A-R1 to init
781	run_cmd ${ns_a} ip link set veth_A-R1 netns 1
782	run_cmd ip addr add ${prefix4}.${a_r1}.1/${veth4_mask} dev veth_A-R1
783	run_cmd ip addr add ${prefix6}:${a_r1}::1/${veth6_mask} dev veth_A-R1
784	run_cmd ip link set veth_A-R1 up
785	run_cmd ip route add ${prefix4}.${b_r1}.1 via ${prefix4}.${a_r1}.2
786	run_cmd ip route add ${prefix6}:${b_r1}::1 via ${prefix6}:${a_r1}::2
787}
788
789setup() {
790	[ "$(id -u)" -ne 0 ] && echo "  need to run as root" && return $ksft_skip
791
792	cleanup
793	for arg do
794		eval setup_${arg} || { echo "  ${arg} not supported"; return 1; }
795	done
796}
797
798trace() {
799	[ $TRACING -eq 0 ] && return
800
801	for arg do
802		[ "${ns_cmd}" = "" ] && ns_cmd="${arg}" && continue
803		${ns_cmd} tcpdump -s 0 -i "${arg}" -w "${name}_${arg}.pcap" 2> /dev/null &
804		tcpdump_pids="${tcpdump_pids} $!"
805		ns_cmd=
806	done
807	sleep 1
808}
809
810cleanup() {
811	for pid in ${tcpdump_pids}; do
812		kill ${pid}
813	done
814	tcpdump_pids=
815
816	for n in ${NS_A} ${NS_B} ${NS_C} ${NS_R1} ${NS_R2}; do
817		ip netns del ${n} 2> /dev/null
818	done
819
820	ip link del veth_A-C			2>/dev/null
821	ip link del veth_A-R1			2>/dev/null
822	ovs-vsctl --if-exists del-port vxlan_a	2>/dev/null
823	ovs-vsctl --if-exists del-br ovs_br0	2>/dev/null
824}
825
826mtu() {
827	ns_cmd="${1}"
828	dev="${2}"
829	mtu="${3}"
830
831	${ns_cmd} ip link set dev ${dev} mtu ${mtu}
832}
833
834mtu_parse() {
835	input="${1}"
836
837	next=0
838	for i in ${input}; do
839		[ ${next} -eq 1 -a "${i}" = "lock" ] && next=2 && continue
840		[ ${next} -eq 1 ] && echo "${i}" && return
841		[ ${next} -eq 2 ] && echo "lock ${i}" && return
842		[ "${i}" = "mtu" ] && next=1
843	done
844}
845
846link_get() {
847	ns_cmd="${1}"
848	name="${2}"
849
850	${ns_cmd} ip link show dev "${name}"
851}
852
853link_get_mtu() {
854	ns_cmd="${1}"
855	name="${2}"
856
857	mtu_parse "$(link_get "${ns_cmd}" ${name})"
858}
859
860route_get_dst_exception() {
861	ns_cmd="${1}"
862	dst="${2}"
863
864	${ns_cmd} ip route get "${dst}"
865}
866
867route_get_dst_pmtu_from_exception() {
868	ns_cmd="${1}"
869	dst="${2}"
870
871	mtu_parse "$(route_get_dst_exception "${ns_cmd}" ${dst})"
872}
873
874check_pmtu_value() {
875	expected="${1}"
876	value="${2}"
877	event="${3}"
878
879	[ "${expected}" = "any" ] && [ -n "${value}" ] && return 0
880	[ "${value}" = "${expected}" ] && return 0
881	[ -z "${value}" ] &&    err "  PMTU exception wasn't created after ${event}" && return 1
882	[ -z "${expected}" ] && err "  PMTU exception shouldn't exist after ${event}" && return 1
883	err "  found PMTU exception with incorrect MTU ${value}, expected ${expected}, after ${event}"
884	return 1
885}
886
887test_pmtu_ipvX() {
888	family=${1}
889
890	setup namespaces routing || return 2
891	trace "${ns_a}"  veth_A-R1    "${ns_r1}" veth_R1-A \
892	      "${ns_r1}" veth_R1-B    "${ns_b}"  veth_B-R1 \
893	      "${ns_a}"  veth_A-R2    "${ns_r2}" veth_R2-A \
894	      "${ns_r2}" veth_R2-B    "${ns_b}"  veth_B-R2
895
896	if [ ${family} -eq 4 ]; then
897		ping=ping
898		dst1="${prefix4}.${b_r1}.1"
899		dst2="${prefix4}.${b_r2}.1"
900	else
901		ping=${ping6}
902		dst1="${prefix6}:${b_r1}::1"
903		dst2="${prefix6}:${b_r2}::1"
904	fi
905
906	# Set up initial MTU values
907	mtu "${ns_a}"  veth_A-R1 2000
908	mtu "${ns_r1}" veth_R1-A 2000
909	mtu "${ns_r1}" veth_R1-B 1400
910	mtu "${ns_b}"  veth_B-R1 1400
911
912	mtu "${ns_a}"  veth_A-R2 2000
913	mtu "${ns_r2}" veth_R2-A 2000
914	mtu "${ns_r2}" veth_R2-B 1500
915	mtu "${ns_b}"  veth_B-R2 1500
916
917	# Create route exceptions
918	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s 1800 ${dst1}
919	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s 1800 ${dst2}
920
921	# Check that exceptions have been created with the correct PMTU
922	pmtu_1="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst1})"
923	check_pmtu_value "1400" "${pmtu_1}" "exceeding MTU" || return 1
924	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
925	check_pmtu_value "1500" "${pmtu_2}" "exceeding MTU" || return 1
926
927	# Decrease local MTU below PMTU, check for PMTU decrease in route exception
928	mtu "${ns_a}"  veth_A-R1 1300
929	mtu "${ns_r1}" veth_R1-A 1300
930	pmtu_1="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst1})"
931	check_pmtu_value "1300" "${pmtu_1}" "decreasing local MTU" || return 1
932	# Second exception shouldn't be modified
933	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
934	check_pmtu_value "1500" "${pmtu_2}" "changing local MTU on a link not on this path" || return 1
935
936	# Increase MTU, check for PMTU increase in route exception
937	mtu "${ns_a}"  veth_A-R1 1700
938	mtu "${ns_r1}" veth_R1-A 1700
939	pmtu_1="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst1})"
940	check_pmtu_value "1700" "${pmtu_1}" "increasing local MTU" || return 1
941	# Second exception shouldn't be modified
942	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
943	check_pmtu_value "1500" "${pmtu_2}" "changing local MTU on a link not on this path" || return 1
944
945	# Skip PMTU locking tests for IPv6
946	[ $family -eq 6 ] && return 0
947
948	# Decrease remote MTU on path via R2, get new exception
949	mtu "${ns_r2}" veth_R2-B 400
950	mtu "${ns_b}"  veth_B-R2 400
951	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s 1400 ${dst2}
952	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
953	check_pmtu_value "lock 552" "${pmtu_2}" "exceeding MTU, with MTU < min_pmtu" || return 1
954
955	# Decrease local MTU below PMTU
956	mtu "${ns_a}"  veth_A-R2 500
957	mtu "${ns_r2}" veth_R2-A 500
958	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
959	check_pmtu_value "500" "${pmtu_2}" "decreasing local MTU" || return 1
960
961	# Increase local MTU
962	mtu "${ns_a}"  veth_A-R2 1500
963	mtu "${ns_r2}" veth_R2-A 1500
964	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
965	check_pmtu_value "1500" "${pmtu_2}" "increasing local MTU" || return 1
966
967	# Get new exception
968	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s 1400 ${dst2}
969	pmtu_2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
970	check_pmtu_value "lock 552" "${pmtu_2}" "exceeding MTU, with MTU < min_pmtu" || return 1
971}
972
973test_pmtu_ipv4_exception() {
974	test_pmtu_ipvX 4
975}
976
977test_pmtu_ipv6_exception() {
978	test_pmtu_ipvX 6
979}
980
981test_pmtu_ipvX_over_vxlanY_or_geneveY_exception() {
982	type=${1}
983	family=${2}
984	outer_family=${3}
985	ll_mtu=4000
986
987	if [ ${outer_family} -eq 4 ]; then
988		setup namespaces routing ${type}4 || return 2
989		#                      IPv4 header   UDP header   VXLAN/GENEVE header   Ethernet header
990		exp_mtu=$((${ll_mtu} - 20          - 8          - 8                   - 14))
991	else
992		setup namespaces routing ${type}6 || return 2
993		#                      IPv6 header   UDP header   VXLAN/GENEVE header   Ethernet header
994		exp_mtu=$((${ll_mtu} - 40          - 8          - 8                   - 14))
995	fi
996
997	trace "${ns_a}" ${type}_a    "${ns_b}"  ${type}_b \
998	      "${ns_a}" veth_A-R1    "${ns_r1}" veth_R1-A \
999	      "${ns_b}" veth_B-R1    "${ns_r1}" veth_R1-B
1000
1001	if [ ${family} -eq 4 ]; then
1002		ping=ping
1003		dst=${tunnel4_b_addr}
1004	else
1005		ping=${ping6}
1006		dst=${tunnel6_b_addr}
1007	fi
1008
1009	# Create route exception by exceeding link layer MTU
1010	mtu "${ns_a}"  veth_A-R1 $((${ll_mtu} + 1000))
1011	mtu "${ns_r1}" veth_R1-A $((${ll_mtu} + 1000))
1012	mtu "${ns_b}"  veth_B-R1 ${ll_mtu}
1013	mtu "${ns_r1}" veth_R1-B ${ll_mtu}
1014
1015	mtu "${ns_a}" ${type}_a $((${ll_mtu} + 1000))
1016	mtu "${ns_b}" ${type}_b $((${ll_mtu} + 1000))
1017	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s $((${ll_mtu} + 500)) ${dst}
1018
1019	# Check that exception was created
1020	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst})"
1021	check_pmtu_value ${exp_mtu} "${pmtu}" "exceeding link layer MTU on ${type} interface"
1022}
1023
1024test_pmtu_ipv4_vxlan4_exception() {
1025	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception vxlan  4 4
1026}
1027
1028test_pmtu_ipv6_vxlan4_exception() {
1029	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception vxlan  6 4
1030}
1031
1032test_pmtu_ipv4_geneve4_exception() {
1033	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception geneve 4 4
1034}
1035
1036test_pmtu_ipv6_geneve4_exception() {
1037	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception geneve 6 4
1038}
1039
1040test_pmtu_ipv4_vxlan6_exception() {
1041	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception vxlan  4 6
1042}
1043
1044test_pmtu_ipv6_vxlan6_exception() {
1045	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception vxlan  6 6
1046}
1047
1048test_pmtu_ipv4_geneve6_exception() {
1049	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception geneve 4 6
1050}
1051
1052test_pmtu_ipv6_geneve6_exception() {
1053	test_pmtu_ipvX_over_vxlanY_or_geneveY_exception geneve 6 6
1054}
1055
1056test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception() {
1057	type=${1}
1058	family=${2}
1059	outer_family=${3}
1060	ll_mtu=4000
1061
1062	if [ ${outer_family} -eq 4 ]; then
1063		setup namespaces routing bridge bridged_${type}4 || return 2
1064		#                      IPv4 header   UDP header   VXLAN/GENEVE header   Ethernet header
1065		exp_mtu=$((${ll_mtu} - 20          - 8          - 8                   - 14))
1066	else
1067		setup namespaces routing bridge bridged_${type}6 || return 2
1068		#                      IPv6 header   UDP header   VXLAN/GENEVE header   Ethernet header
1069		exp_mtu=$((${ll_mtu} - 40          - 8          - 8                   - 14))
1070	fi
1071
1072	trace "${ns_a}" ${type}_a    "${ns_b}"  ${type}_b \
1073	      "${ns_a}" veth_A-R1    "${ns_r1}" veth_R1-A \
1074	      "${ns_b}" veth_B-R1    "${ns_r1}" veth_R1-B \
1075	      "${ns_a}" br0          "${ns_a}"  veth-A-C  \
1076	      "${ns_c}" veth_C-A
1077
1078	if [ ${family} -eq 4 ]; then
1079		ping=ping
1080		dst=${tunnel4_b_addr}
1081	else
1082		ping=${ping6}
1083		dst=${tunnel6_b_addr}
1084	fi
1085
1086	# Create route exception by exceeding link layer MTU
1087	mtu "${ns_a}"  veth_A-R1 $((${ll_mtu} + 1000))
1088	mtu "${ns_a}"  br0       $((${ll_mtu} + 1000))
1089	mtu "${ns_a}"  veth_A-C  $((${ll_mtu} + 1000))
1090	mtu "${ns_c}"  veth_C-A  $((${ll_mtu} + 1000))
1091	mtu "${ns_r1}" veth_R1-A $((${ll_mtu} + 1000))
1092	mtu "${ns_b}"  veth_B-R1 ${ll_mtu}
1093	mtu "${ns_r1}" veth_R1-B ${ll_mtu}
1094
1095	mtu "${ns_a}" ${type}_a $((${ll_mtu} + 1000))
1096	mtu "${ns_b}" ${type}_b $((${ll_mtu} + 1000))
1097
1098	run_cmd ${ns_c} ${ping} -q -M want -i 0.1 -c 10 -s $((${ll_mtu} + 500)) ${dst} || return 1
1099	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1  -s $((${ll_mtu} + 500)) ${dst} || return 1
1100
1101	# Check that exceptions were created
1102	pmtu="$(route_get_dst_pmtu_from_exception "${ns_c}" ${dst})"
1103	check_pmtu_value ${exp_mtu} "${pmtu}" "exceeding link layer MTU on bridged ${type} interface"
1104	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst})"
1105	check_pmtu_value ${exp_mtu} "${pmtu}" "exceeding link layer MTU on locally bridged ${type} interface"
1106}
1107
1108test_pmtu_ipv4_br_vxlan4_exception() {
1109	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception vxlan  4 4
1110}
1111
1112test_pmtu_ipv6_br_vxlan4_exception() {
1113	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception vxlan  6 4
1114}
1115
1116test_pmtu_ipv4_br_geneve4_exception() {
1117	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception geneve 4 4
1118}
1119
1120test_pmtu_ipv6_br_geneve4_exception() {
1121	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception geneve 6 4
1122}
1123
1124test_pmtu_ipv4_br_vxlan6_exception() {
1125	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception vxlan  4 6
1126}
1127
1128test_pmtu_ipv6_br_vxlan6_exception() {
1129	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception vxlan  6 6
1130}
1131
1132test_pmtu_ipv4_br_geneve6_exception() {
1133	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception geneve 4 6
1134}
1135
1136test_pmtu_ipv6_br_geneve6_exception() {
1137	test_pmtu_ipvX_over_bridged_vxlanY_or_geneveY_exception geneve 6 6
1138}
1139
1140test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception() {
1141	type=${1}
1142	family=${2}
1143	outer_family=${3}
1144	ll_mtu=4000
1145
1146	if [ ${outer_family} -eq 4 ]; then
1147		setup namespaces routing ovs_bridge ovs_${type}4 || return 2
1148		#                      IPv4 header   UDP header   VXLAN/GENEVE header   Ethernet header
1149		exp_mtu=$((${ll_mtu} - 20          - 8          - 8                   - 14))
1150	else
1151		setup namespaces routing ovs_bridge ovs_${type}6 || return 2
1152		#                      IPv6 header   UDP header   VXLAN/GENEVE header   Ethernet header
1153		exp_mtu=$((${ll_mtu} - 40          - 8          - 8                   - 14))
1154	fi
1155
1156	if [ "${type}" = "vxlan" ]; then
1157		tun_a="vxlan_sys_4789"
1158	elif [ "${type}" = "geneve" ]; then
1159		tun_a="genev_sys_6081"
1160	fi
1161
1162	trace ""        "${tun_a}"  "${ns_b}"  ${type}_b \
1163	      ""        veth_A-R1   "${ns_r1}" veth_R1-A \
1164	      "${ns_b}" veth_B-R1   "${ns_r1}" veth_R1-B \
1165	      ""        ovs_br0     ""         veth-A-C  \
1166	      "${ns_c}" veth_C-A
1167
1168	if [ ${family} -eq 4 ]; then
1169		ping=ping
1170		dst=${tunnel4_b_addr}
1171	else
1172		ping=${ping6}
1173		dst=${tunnel6_b_addr}
1174	fi
1175
1176	# Create route exception by exceeding link layer MTU
1177	mtu ""         veth_A-R1 $((${ll_mtu} + 1000))
1178	mtu ""         ovs_br0   $((${ll_mtu} + 1000))
1179	mtu ""         veth_A-C  $((${ll_mtu} + 1000))
1180	mtu "${ns_c}"  veth_C-A  $((${ll_mtu} + 1000))
1181	mtu "${ns_r1}" veth_R1-A $((${ll_mtu} + 1000))
1182	mtu "${ns_b}"  veth_B-R1 ${ll_mtu}
1183	mtu "${ns_r1}" veth_R1-B ${ll_mtu}
1184
1185	mtu ""        ${tun_a}  $((${ll_mtu} + 1000))
1186	mtu "${ns_b}" ${type}_b $((${ll_mtu} + 1000))
1187
1188	run_cmd ${ns_c} ${ping} -q -M want -i 0.1 -c 20 -s $((${ll_mtu} + 500)) ${dst} || return 1
1189
1190	# Check that exceptions were created
1191	pmtu="$(route_get_dst_pmtu_from_exception "${ns_c}" ${dst})"
1192	check_pmtu_value ${exp_mtu} "${pmtu}" "exceeding link layer MTU on Open vSwitch ${type} interface"
1193}
1194
1195test_pmtu_ipv4_ovs_vxlan4_exception() {
1196	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception vxlan  4 4
1197}
1198
1199test_pmtu_ipv6_ovs_vxlan4_exception() {
1200	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception vxlan  6 4
1201}
1202
1203test_pmtu_ipv4_ovs_geneve4_exception() {
1204	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception geneve 4 4
1205}
1206
1207test_pmtu_ipv6_ovs_geneve4_exception() {
1208	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception geneve 6 4
1209}
1210
1211test_pmtu_ipv4_ovs_vxlan6_exception() {
1212	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception vxlan  4 6
1213}
1214
1215test_pmtu_ipv6_ovs_vxlan6_exception() {
1216	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception vxlan  6 6
1217}
1218
1219test_pmtu_ipv4_ovs_geneve6_exception() {
1220	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception geneve 4 6
1221}
1222
1223test_pmtu_ipv6_ovs_geneve6_exception() {
1224	test_pmtu_ipvX_over_ovs_vxlanY_or_geneveY_exception geneve 6 6
1225}
1226
1227test_pmtu_ipvX_over_fouY_or_gueY() {
1228	inner_family=${1}
1229	outer_family=${2}
1230	encap=${3}
1231	ll_mtu=4000
1232
1233	setup namespaces routing ${encap}${outer_family}${inner_family} || return 2
1234	trace "${ns_a}" ${encap}_a   "${ns_b}"  ${encap}_b \
1235	      "${ns_a}" veth_A-R1    "${ns_r1}" veth_R1-A \
1236	      "${ns_b}" veth_B-R1    "${ns_r1}" veth_R1-B
1237
1238	if [ ${inner_family} -eq 4 ]; then
1239		ping=ping
1240		dst=${tunnel4_b_addr}
1241	else
1242		ping=${ping6}
1243		dst=${tunnel6_b_addr}
1244	fi
1245
1246	if [ "${encap}" = "gue" ]; then
1247		encap_overhead=4
1248	else
1249		encap_overhead=0
1250	fi
1251
1252	if [ ${outer_family} -eq 4 ]; then
1253		#                      IPv4 header   UDP header
1254		exp_mtu=$((${ll_mtu} - 20          - 8         - ${encap_overhead}))
1255	else
1256		#                      IPv6 header   Option 4   UDP header
1257		exp_mtu=$((${ll_mtu} - 40          - 8        - 8       - ${encap_overhead}))
1258	fi
1259
1260	# Create route exception by exceeding link layer MTU
1261	mtu "${ns_a}"  veth_A-R1 $((${ll_mtu} + 1000))
1262	mtu "${ns_r1}" veth_R1-A $((${ll_mtu} + 1000))
1263	mtu "${ns_b}"  veth_B-R1 ${ll_mtu}
1264	mtu "${ns_r1}" veth_R1-B ${ll_mtu}
1265
1266	mtu "${ns_a}" ${encap}_a $((${ll_mtu} + 1000))
1267	mtu "${ns_b}" ${encap}_b $((${ll_mtu} + 1000))
1268	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s $((${ll_mtu} + 500)) ${dst}
1269
1270	# Check that exception was created
1271	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst})"
1272	check_pmtu_value ${exp_mtu} "${pmtu}" "exceeding link layer MTU on ${encap} interface"
1273}
1274
1275test_pmtu_ipv4_fou4_exception() {
1276	test_pmtu_ipvX_over_fouY_or_gueY 4 4 fou
1277}
1278
1279test_pmtu_ipv6_fou4_exception() {
1280	test_pmtu_ipvX_over_fouY_or_gueY 6 4 fou
1281}
1282
1283test_pmtu_ipv4_fou6_exception() {
1284	test_pmtu_ipvX_over_fouY_or_gueY 4 6 fou
1285}
1286
1287test_pmtu_ipv6_fou6_exception() {
1288	test_pmtu_ipvX_over_fouY_or_gueY 6 6 fou
1289}
1290
1291test_pmtu_ipv4_gue4_exception() {
1292	test_pmtu_ipvX_over_fouY_or_gueY 4 4 gue
1293}
1294
1295test_pmtu_ipv6_gue4_exception() {
1296	test_pmtu_ipvX_over_fouY_or_gueY 6 4 gue
1297}
1298
1299test_pmtu_ipv4_gue6_exception() {
1300	test_pmtu_ipvX_over_fouY_or_gueY 4 6 gue
1301}
1302
1303test_pmtu_ipv6_gue6_exception() {
1304	test_pmtu_ipvX_over_fouY_or_gueY 6 6 gue
1305}
1306
1307test_pmtu_ipvX_over_ipvY_exception() {
1308	inner=${1}
1309	outer=${2}
1310	ll_mtu=4000
1311
1312	setup namespaces routing ip${inner}ip${outer} || return 2
1313
1314	trace "${ns_a}" ip_a         "${ns_b}"  ip_b  \
1315	      "${ns_a}" veth_A-R1    "${ns_r1}" veth_R1-A \
1316	      "${ns_b}" veth_B-R1    "${ns_r1}" veth_R1-B
1317
1318	if [ ${inner} -eq 4 ]; then
1319		ping=ping
1320		dst=${tunnel4_b_addr}
1321	else
1322		ping=${ping6}
1323		dst=${tunnel6_b_addr}
1324	fi
1325
1326	if [ ${outer} -eq 4 ]; then
1327		#                      IPv4 header
1328		exp_mtu=$((${ll_mtu} - 20))
1329	else
1330		#                      IPv6 header   Option 4
1331		exp_mtu=$((${ll_mtu} - 40          - 8))
1332	fi
1333
1334	# Create route exception by exceeding link layer MTU
1335	mtu "${ns_a}"  veth_A-R1 $((${ll_mtu} + 1000))
1336	mtu "${ns_r1}" veth_R1-A $((${ll_mtu} + 1000))
1337	mtu "${ns_b}"  veth_B-R1 ${ll_mtu}
1338	mtu "${ns_r1}" veth_R1-B ${ll_mtu}
1339
1340	mtu "${ns_a}" ip_a $((${ll_mtu} + 1000)) || return
1341	mtu "${ns_b}" ip_b $((${ll_mtu} + 1000)) || return
1342	run_cmd ${ns_a} ${ping} -q -M want -i 0.1 -w 1 -s $((${ll_mtu} + 500)) ${dst}
1343
1344	# Check that exception was created
1345	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst})"
1346	check_pmtu_value ${exp_mtu} "${pmtu}" "exceeding link layer MTU on ip${inner}ip${outer} interface"
1347}
1348
1349test_pmtu_ipv4_ipv4_exception() {
1350	test_pmtu_ipvX_over_ipvY_exception 4 4
1351}
1352
1353test_pmtu_ipv6_ipv4_exception() {
1354	test_pmtu_ipvX_over_ipvY_exception 6 4
1355}
1356
1357test_pmtu_ipv4_ipv6_exception() {
1358	test_pmtu_ipvX_over_ipvY_exception 4 6
1359}
1360
1361test_pmtu_ipv6_ipv6_exception() {
1362	test_pmtu_ipvX_over_ipvY_exception 6 6
1363}
1364
1365test_pmtu_vti4_exception() {
1366	setup namespaces veth vti4 xfrm4 || return 2
1367	trace "${ns_a}" veth_a    "${ns_b}" veth_b \
1368	      "${ns_a}" vti4_a    "${ns_b}" vti4_b
1369
1370	veth_mtu=1500
1371	vti_mtu=$((veth_mtu - 20))
1372
1373	#                                SPI   SN   IV  ICV   pad length   next header
1374	esp_payload_rfc4106=$((vti_mtu - 4   - 4  - 8 - 16  - 1          - 1))
1375	ping_payload=$((esp_payload_rfc4106 - 28))
1376
1377	mtu "${ns_a}" veth_a ${veth_mtu}
1378	mtu "${ns_b}" veth_b ${veth_mtu}
1379	mtu "${ns_a}" vti4_a ${vti_mtu}
1380	mtu "${ns_b}" vti4_b ${vti_mtu}
1381
1382	# Send DF packet without exceeding link layer MTU, check that no
1383	# exception is created
1384	run_cmd ${ns_a} ping -q -M want -i 0.1 -w 1 -s ${ping_payload} ${tunnel4_b_addr}
1385	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${tunnel4_b_addr})"
1386	check_pmtu_value "" "${pmtu}" "sending packet smaller than PMTU (IP payload length ${esp_payload_rfc4106})" || return 1
1387
1388	# Now exceed link layer MTU by one byte, check that exception is created
1389	# with the right PMTU value
1390	run_cmd ${ns_a} ping -q -M want -i 0.1 -w 1 -s $((ping_payload + 1)) ${tunnel4_b_addr}
1391	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${tunnel4_b_addr})"
1392	check_pmtu_value "${esp_payload_rfc4106}" "${pmtu}" "exceeding PMTU (IP payload length $((esp_payload_rfc4106 + 1)))"
1393}
1394
1395test_pmtu_vti6_exception() {
1396	setup namespaces veth vti6 xfrm6 || return 2
1397	trace "${ns_a}" veth_a    "${ns_b}" veth_b \
1398	      "${ns_a}" vti6_a    "${ns_b}" vti6_b
1399	fail=0
1400
1401	# Create route exception by exceeding link layer MTU
1402	mtu "${ns_a}" veth_a 4000
1403	mtu "${ns_b}" veth_b 4000
1404	mtu "${ns_a}" vti6_a 5000
1405	mtu "${ns_b}" vti6_b 5000
1406	run_cmd ${ns_a} ${ping6} -q -i 0.1 -w 1 -s 60000 ${tunnel6_b_addr}
1407
1408	# Check that exception was created
1409	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${tunnel6_b_addr})"
1410	check_pmtu_value any "${pmtu}" "creating tunnel exceeding link layer MTU" || return 1
1411
1412	# Decrease tunnel MTU, check for PMTU decrease in route exception
1413	mtu "${ns_a}" vti6_a 3000
1414	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${tunnel6_b_addr})"
1415	check_pmtu_value "3000" "${pmtu}" "decreasing tunnel MTU" || fail=1
1416
1417	# Increase tunnel MTU, check for PMTU increase in route exception
1418	mtu "${ns_a}" vti6_a 9000
1419	pmtu="$(route_get_dst_pmtu_from_exception "${ns_a}" ${tunnel6_b_addr})"
1420	check_pmtu_value "9000" "${pmtu}" "increasing tunnel MTU" || fail=1
1421
1422	return ${fail}
1423}
1424
1425test_pmtu_vti4_default_mtu() {
1426	setup namespaces veth vti4 || return 2
1427
1428	# Check that MTU of vti device is MTU of veth minus IPv4 header length
1429	veth_mtu="$(link_get_mtu "${ns_a}" veth_a)"
1430	vti4_mtu="$(link_get_mtu "${ns_a}" vti4_a)"
1431	if [ $((veth_mtu - vti4_mtu)) -ne 20 ]; then
1432		err "  vti MTU ${vti4_mtu} is not veth MTU ${veth_mtu} minus IPv4 header length"
1433		return 1
1434	fi
1435}
1436
1437test_pmtu_vti6_default_mtu() {
1438	setup namespaces veth vti6 || return 2
1439
1440	# Check that MTU of vti device is MTU of veth minus IPv6 header length
1441	veth_mtu="$(link_get_mtu "${ns_a}" veth_a)"
1442	vti6_mtu="$(link_get_mtu "${ns_a}" vti6_a)"
1443	if [ $((veth_mtu - vti6_mtu)) -ne 40 ]; then
1444		err "  vti MTU ${vti6_mtu} is not veth MTU ${veth_mtu} minus IPv6 header length"
1445		return 1
1446	fi
1447}
1448
1449test_pmtu_vti4_link_add_mtu() {
1450	setup namespaces || return 2
1451
1452	run_cmd ${ns_a} ip link add vti4_a type vti local ${veth4_a_addr} remote ${veth4_b_addr} key 10
1453	[ $? -ne 0 ] && err "  vti not supported" && return 2
1454	run_cmd ${ns_a} ip link del vti4_a
1455
1456	fail=0
1457
1458	min=68
1459	max=$((65535 - 20))
1460	# Check invalid values first
1461	for v in $((min - 1)) $((max + 1)); do
1462		run_cmd ${ns_a} ip link add vti4_a mtu ${v} type vti local ${veth4_a_addr} remote ${veth4_b_addr} key 10
1463		# This can fail, or MTU can be adjusted to a proper value
1464		[ $? -ne 0 ] && continue
1465		mtu="$(link_get_mtu "${ns_a}" vti4_a)"
1466		if [ ${mtu} -lt ${min} -o ${mtu} -gt ${max} ]; then
1467			err "  vti tunnel created with invalid MTU ${mtu}"
1468			fail=1
1469		fi
1470		run_cmd ${ns_a} ip link del vti4_a
1471	done
1472
1473	# Now check valid values
1474	for v in ${min} 1300 ${max}; do
1475		run_cmd ${ns_a} ip link add vti4_a mtu ${v} type vti local ${veth4_a_addr} remote ${veth4_b_addr} key 10
1476		mtu="$(link_get_mtu "${ns_a}" vti4_a)"
1477		run_cmd ${ns_a} ip link del vti4_a
1478		if [ "${mtu}" != "${v}" ]; then
1479			err "  vti MTU ${mtu} doesn't match configured value ${v}"
1480			fail=1
1481		fi
1482	done
1483
1484	return ${fail}
1485}
1486
1487test_pmtu_vti6_link_add_mtu() {
1488	setup namespaces || return 2
1489
1490	run_cmd ${ns_a} ip link add vti6_a type vti6 local ${veth6_a_addr} remote ${veth6_b_addr} key 10
1491	[ $? -ne 0 ] && err "  vti6 not supported" && return 2
1492	run_cmd ${ns_a} ip link del vti6_a
1493
1494	fail=0
1495
1496	min=68			# vti6 can carry IPv4 packets too
1497	max=$((65535 - 40))
1498	# Check invalid values first
1499	for v in $((min - 1)) $((max + 1)); do
1500		run_cmd ${ns_a} ip link add vti6_a mtu ${v} type vti6 local ${veth6_a_addr} remote ${veth6_b_addr} key 10
1501		# This can fail, or MTU can be adjusted to a proper value
1502		[ $? -ne 0 ] && continue
1503		mtu="$(link_get_mtu "${ns_a}" vti6_a)"
1504		if [ ${mtu} -lt ${min} -o ${mtu} -gt ${max} ]; then
1505			err "  vti6 tunnel created with invalid MTU ${v}"
1506			fail=1
1507		fi
1508		run_cmd ${ns_a} ip link del vti6_a
1509	done
1510
1511	# Now check valid values
1512	for v in 68 1280 1300 $((65535 - 40)); do
1513		run_cmd ${ns_a} ip link add vti6_a mtu ${v} type vti6 local ${veth6_a_addr} remote ${veth6_b_addr} key 10
1514		mtu="$(link_get_mtu "${ns_a}" vti6_a)"
1515		run_cmd ${ns_a} ip link del vti6_a
1516		if [ "${mtu}" != "${v}" ]; then
1517			err "  vti6 MTU ${mtu} doesn't match configured value ${v}"
1518			fail=1
1519		fi
1520	done
1521
1522	return ${fail}
1523}
1524
1525test_pmtu_vti6_link_change_mtu() {
1526	setup namespaces || return 2
1527
1528	run_cmd ${ns_a} ip link add dummy0 mtu 1500 type dummy
1529	[ $? -ne 0 ] && err "  dummy not supported" && return 2
1530	run_cmd ${ns_a} ip link add dummy1 mtu 3000 type dummy
1531	run_cmd ${ns_a} ip link set dummy0 up
1532	run_cmd ${ns_a} ip link set dummy1 up
1533
1534	run_cmd ${ns_a} ip addr add ${dummy6_0_prefix}1/${dummy6_mask} dev dummy0
1535	run_cmd ${ns_a} ip addr add ${dummy6_1_prefix}1/${dummy6_mask} dev dummy1
1536
1537	fail=0
1538
1539	# Create vti6 interface bound to device, passing MTU, check it
1540	run_cmd ${ns_a} ip link add vti6_a mtu 1300 type vti6 remote ${dummy6_0_prefix}2 local ${dummy6_0_prefix}1
1541	mtu="$(link_get_mtu "${ns_a}" vti6_a)"
1542	if [ ${mtu} -ne 1300 ]; then
1543		err "  vti6 MTU ${mtu} doesn't match configured value 1300"
1544		fail=1
1545	fi
1546
1547	# Move to another device with different MTU, without passing MTU, check
1548	# MTU is adjusted
1549	run_cmd ${ns_a} ip link set vti6_a type vti6 remote ${dummy6_1_prefix}2 local ${dummy6_1_prefix}1
1550	mtu="$(link_get_mtu "${ns_a}" vti6_a)"
1551	if [ ${mtu} -ne $((3000 - 40)) ]; then
1552		err "  vti MTU ${mtu} is not dummy MTU 3000 minus IPv6 header length"
1553		fail=1
1554	fi
1555
1556	# Move it back, passing MTU, check MTU is not overridden
1557	run_cmd ${ns_a} ip link set vti6_a mtu 1280 type vti6 remote ${dummy6_0_prefix}2 local ${dummy6_0_prefix}1
1558	mtu="$(link_get_mtu "${ns_a}" vti6_a)"
1559	if [ ${mtu} -ne 1280 ]; then
1560		err "  vti6 MTU ${mtu} doesn't match configured value 1280"
1561		fail=1
1562	fi
1563
1564	return ${fail}
1565}
1566
1567check_command() {
1568	cmd=${1}
1569
1570	if ! which ${cmd} > /dev/null 2>&1; then
1571		err "  missing required command: '${cmd}'"
1572		return 1
1573	fi
1574	return 0
1575}
1576
1577test_cleanup_vxlanX_exception() {
1578	outer="${1}"
1579	encap="vxlan"
1580	ll_mtu=4000
1581
1582	check_command taskset || return 2
1583	cpu_list=$(grep -m 2 processor /proc/cpuinfo | cut -d ' ' -f 2)
1584
1585	setup namespaces routing ${encap}${outer} || return 2
1586	trace "${ns_a}" ${encap}_a   "${ns_b}"  ${encap}_b \
1587	      "${ns_a}" veth_A-R1    "${ns_r1}" veth_R1-A \
1588	      "${ns_b}" veth_B-R1    "${ns_r1}" veth_R1-B
1589
1590	# Create route exception by exceeding link layer MTU
1591	mtu "${ns_a}"  veth_A-R1 $((${ll_mtu} + 1000))
1592	mtu "${ns_r1}" veth_R1-A $((${ll_mtu} + 1000))
1593	mtu "${ns_b}"  veth_B-R1 ${ll_mtu}
1594	mtu "${ns_r1}" veth_R1-B ${ll_mtu}
1595
1596	mtu "${ns_a}" ${encap}_a $((${ll_mtu} + 1000))
1597	mtu "${ns_b}" ${encap}_b $((${ll_mtu} + 1000))
1598
1599	# Fill exception cache for multiple CPUs (2)
1600	# we can always use inner IPv4 for that
1601	for cpu in ${cpu_list}; do
1602		run_cmd taskset --cpu-list ${cpu} ${ns_a} ping -q -M want -i 0.1 -w 1 -s $((${ll_mtu} + 500)) ${tunnel4_b_addr}
1603	done
1604
1605	${ns_a} ip link del dev veth_A-R1 &
1606	iplink_pid=$!
1607	sleep 1
1608	if [ "$(cat /proc/${iplink_pid}/cmdline 2>/dev/null | tr -d '\0')" = "iplinkdeldevveth_A-R1" ]; then
1609		err "  can't delete veth device in a timely manner, PMTU dst likely leaked"
1610		return 1
1611	fi
1612}
1613
1614test_cleanup_ipv6_exception() {
1615	test_cleanup_vxlanX_exception 6
1616}
1617
1618test_cleanup_ipv4_exception() {
1619	test_cleanup_vxlanX_exception 4
1620}
1621
1622run_test() {
1623	(
1624	tname="$1"
1625	tdesc="$2"
1626
1627	unset IFS
1628
1629	if [ "$VERBOSE" = "1" ]; then
1630		printf "\n##########################################################################\n\n"
1631	fi
1632
1633	eval test_${tname}
1634	ret=$?
1635
1636	if [ $ret -eq 0 ]; then
1637		printf "TEST: %-60s  [ OK ]\n" "${tdesc}"
1638	elif [ $ret -eq 1 ]; then
1639		printf "TEST: %-60s  [FAIL]\n" "${tdesc}"
1640		if [ "${PAUSE_ON_FAIL}" = "yes" ]; then
1641			echo
1642			echo "Pausing. Hit enter to continue"
1643			read a
1644		fi
1645		err_flush
1646		exit 1
1647	elif [ $ret -eq 2 ]; then
1648		printf "TEST: %-60s  [SKIP]\n" "${tdesc}"
1649		err_flush
1650	fi
1651
1652	return $ret
1653	)
1654	ret=$?
1655	[ $ret -ne 0 ] && exitcode=1
1656
1657	return $ret
1658}
1659
1660run_test_nh() {
1661	tname="$1"
1662	tdesc="$2"
1663
1664	USE_NH=yes
1665	run_test "${tname}" "${tdesc} - nexthop objects"
1666	USE_NH=no
1667}
1668
1669test_list_flush_ipv4_exception() {
1670	setup namespaces routing || return 2
1671	trace "${ns_a}"  veth_A-R1    "${ns_r1}" veth_R1-A \
1672	      "${ns_r1}" veth_R1-B    "${ns_b}"  veth_B-R1 \
1673	      "${ns_a}"  veth_A-R2    "${ns_r2}" veth_R2-A \
1674	      "${ns_r2}" veth_R2-B    "${ns_b}"  veth_B-R2
1675
1676	dst_prefix1="${prefix4}.${b_r1}."
1677	dst2="${prefix4}.${b_r2}.1"
1678
1679	# Set up initial MTU values
1680	mtu "${ns_a}"  veth_A-R1 2000
1681	mtu "${ns_r1}" veth_R1-A 2000
1682	mtu "${ns_r1}" veth_R1-B 1500
1683	mtu "${ns_b}"  veth_B-R1 1500
1684
1685	mtu "${ns_a}"  veth_A-R2 2000
1686	mtu "${ns_r2}" veth_R2-A 2000
1687	mtu "${ns_r2}" veth_R2-B 1500
1688	mtu "${ns_b}"  veth_B-R2 1500
1689
1690	fail=0
1691
1692	# Add 100 addresses for veth endpoint on B reached by default A route
1693	for i in $(seq 100 199); do
1694		run_cmd ${ns_b} ip addr add "${dst_prefix1}${i}" dev veth_B-R1
1695	done
1696
1697	# Create 100 cached route exceptions for path via R1, one via R2. Note
1698	# that with IPv4 we need to actually cause a route lookup that matches
1699	# the exception caused by ICMP, in order to actually have a cached
1700	# route, so we need to ping each destination twice
1701	for i in $(seq 100 199); do
1702		run_cmd ${ns_a} ping -q -M want -i 0.1 -c 2 -s 1800 "${dst_prefix1}${i}"
1703	done
1704	run_cmd ${ns_a} ping -q -M want -i 0.1 -c 2 -s 1800 "${dst2}"
1705
1706	if [ "$(${ns_a} ip -oneline route list cache | wc -l)" -ne 101 ]; then
1707		err "  can't list cached exceptions"
1708		fail=1
1709	fi
1710
1711	run_cmd ${ns_a} ip route flush cache
1712	pmtu1="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst_prefix}1)"
1713	pmtu2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst_prefix}2)"
1714	if [ -n "${pmtu1}" ] || [ -n "${pmtu2}" ] || \
1715	   [ -n "$(${ns_a} ip route list cache)" ]; then
1716		err "  can't flush cached exceptions"
1717		fail=1
1718	fi
1719
1720	return ${fail}
1721}
1722
1723test_list_flush_ipv6_exception() {
1724	setup namespaces routing || return 2
1725	trace "${ns_a}"  veth_A-R1    "${ns_r1}" veth_R1-A \
1726	      "${ns_r1}" veth_R1-B    "${ns_b}"  veth_B-R1 \
1727	      "${ns_a}"  veth_A-R2    "${ns_r2}" veth_R2-A \
1728	      "${ns_r2}" veth_R2-B    "${ns_b}"  veth_B-R2
1729
1730	dst_prefix1="${prefix6}:${b_r1}::"
1731	dst2="${prefix6}:${b_r2}::1"
1732
1733	# Set up initial MTU values
1734	mtu "${ns_a}"  veth_A-R1 2000
1735	mtu "${ns_r1}" veth_R1-A 2000
1736	mtu "${ns_r1}" veth_R1-B 1500
1737	mtu "${ns_b}"  veth_B-R1 1500
1738
1739	mtu "${ns_a}"  veth_A-R2 2000
1740	mtu "${ns_r2}" veth_R2-A 2000
1741	mtu "${ns_r2}" veth_R2-B 1500
1742	mtu "${ns_b}"  veth_B-R2 1500
1743
1744	fail=0
1745
1746	# Add 100 addresses for veth endpoint on B reached by default A route
1747	for i in $(seq 100 199); do
1748		run_cmd ${ns_b} ip addr add "${dst_prefix1}${i}" dev veth_B-R1
1749	done
1750
1751	# Create 100 cached route exceptions for path via R1, one via R2
1752	for i in $(seq 100 199); do
1753		run_cmd ${ns_a} ping -q -M want -i 0.1 -w 1 -s 1800 "${dst_prefix1}${i}"
1754	done
1755	run_cmd ${ns_a} ping -q -M want -i 0.1 -w 1 -s 1800 "${dst2}"
1756	if [ "$(${ns_a} ip -oneline -6 route list cache | wc -l)" -ne 101 ]; then
1757		err "  can't list cached exceptions"
1758		fail=1
1759	fi
1760
1761	run_cmd ${ns_a} ip -6 route flush cache
1762	pmtu1="$(route_get_dst_pmtu_from_exception "${ns_a}" "${dst_prefix1}100")"
1763	pmtu2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
1764	if [ -n "${pmtu1}" ] || [ -n "${pmtu2}" ] || \
1765	   [ -n "$(${ns_a} ip -6 route list cache)" ]; then
1766		err "  can't flush cached exceptions"
1767		fail=1
1768	fi
1769
1770	return ${fail}
1771}
1772
1773usage() {
1774	echo
1775	echo "$0 [OPTIONS] [TEST]..."
1776	echo "If no TEST argument is given, all tests will be run."
1777	echo
1778	echo "Options"
1779	echo "  --trace: capture traffic to TEST_INTERFACE.pcap"
1780	echo
1781	echo "Available tests${tests}"
1782	exit 1
1783}
1784
1785################################################################################
1786#
1787exitcode=0
1788desc=0
1789
1790while getopts :ptv o
1791do
1792	case $o in
1793	p) PAUSE_ON_FAIL=yes;;
1794	v) VERBOSE=1;;
1795	t) if which tcpdump > /dev/null 2>&1; then
1796		TRACING=1
1797	   else
1798		echo "=== tcpdump not available, tracing disabled"
1799	   fi
1800	   ;;
1801	*) usage;;
1802	esac
1803done
1804shift $(($OPTIND-1))
1805
1806IFS="
1807"
1808
1809for arg do
1810	# Check first that all requested tests are available before running any
1811	command -v > /dev/null "test_${arg}" || { echo "=== Test ${arg} not found"; usage; }
1812done
1813
1814trap cleanup EXIT
1815
1816# start clean
1817cleanup
1818
1819HAVE_NH=no
1820ip nexthop ls >/dev/null 2>&1
1821[ $? -eq 0 ] && HAVE_NH=yes
1822
1823name=""
1824desc=""
1825rerun_nh=0
1826for t in ${tests}; do
1827	[ "${name}" = "" ]	&& name="${t}"	&& continue
1828	[ "${desc}" = "" ]	&& desc="${t}"	&& continue
1829
1830	if [ "${HAVE_NH}" = "yes" ]; then
1831		rerun_nh="${t}"
1832	fi
1833
1834	run_this=1
1835	for arg do
1836		[ "${arg}" != "${arg#--*}" ] && continue
1837		[ "${arg}" = "${name}" ] && run_this=1 && break
1838		run_this=0
1839	done
1840	if [ $run_this -eq 1 ]; then
1841		run_test "${name}" "${desc}"
1842		# if test was skipped no need to retry with nexthop objects
1843		[ $? -eq 2 ] && rerun_nh=0
1844
1845		if [ "${rerun_nh}" = "1" ]; then
1846			run_test_nh "${name}" "${desc}"
1847		fi
1848	fi
1849	name=""
1850	desc=""
1851	rerun_nh=0
1852done
1853
1854exit ${exitcode}
1855