Skip to contentSkip to conversation

selftests: pmtu: List/flush IPv4 cached routes, improve IPv6 test

PATCH In review Stefano Brivio 2 participants 5 messages Jun 15, 2019 – Jun 17, 2019

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover selftests: pmtu: List/flush IPv4 cached routes, improve IPv6 test

This series introduce a new test, list_flush_ipv4_exception, and improves the existing list_flush_ipv6_exception test by making it as demanding as the IPv4 one.

Stefano Brivio (2):
selftests: pmtu: Introduce list_flush_ipv4_exception test case
selftests: pmtu: Make list_flush_ipv6_exception test more demanding

 tools/testing/selftests/net/pmtu.sh | 84 ++++++++++++++++++++++++++---
 1 file changed, 76 insertions(+), 8 deletions(-)
Signature · 1 line
2.20.1
#1 lore raw

PATCH 1/2 selftests: pmtu: Introduce list_flush_ipv4_exception test case

This test checks that route exceptions can be successfully listed and flushed using ip -6 route {list,flush} cache.

---
 tools/testing/selftests/net/pmtu.sh | 62 +++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)
tools/testing/selftests/net/pmtu.sh +62 -0
index 269e839b747e..6c063b17d7d0 100755
@@ -112,6 +112,10 @@
 # - cleanup_ipv6_exception
 #	Same as above, but use IPv6 transport from A to B
 #
+# - list_flush_ipv4_exception
+#	Using the same topology as in pmtu_ipv4, create exceptions, and check
+#	they are shown when listing exception caches, gone after flushing them
+#
 # - list_flush_ipv6_exception
 #	Using the same topology as in pmtu_ipv6, create exceptions, and check
 #	they are shown when listing exception caches, gone after flushing them
@@ -156,6 +160,7 @@ tests="
 	pmtu_vti6_link_change_mtu	vti6: MTU changes on link changes	0
 	cleanup_ipv4_exception		ipv4: cleanup of cached exceptions	1
 	cleanup_ipv6_exception		ipv6: cleanup of cached exceptions	1
+	list_flush_ipv4_exception	ipv4: list and flush cached exceptions	1
 	list_flush_ipv6_exception	ipv6: list and flush cached exceptions	1"
 
 NS_A="ns-A"
@@ -1207,6 +1212,63 @@ run_test_nh() {
 	USE_NH=no
 }
 
+test_list_flush_ipv4_exception() {
+	setup namespaces routing || return 2
+	trace "${ns_a}"  veth_A-R1    "${ns_r1}" veth_R1-A \
+	      "${ns_r1}" veth_R1-B    "${ns_b}"  veth_B-R1 \
+	      "${ns_a}"  veth_A-R2    "${ns_r2}" veth_R2-A \
+	      "${ns_r2}" veth_R2-B    "${ns_b}"  veth_B-R2
+
+	dst_prefix1="${prefix4}.${b_r1}."
+	dst2="${prefix4}.${b_r2}.1"
+
+	# Set up initial MTU values
+	mtu "${ns_a}"  veth_A-R1 2000
+	mtu "${ns_r1}" veth_R1-A 2000
+	mtu "${ns_r1}" veth_R1-B 1500
+	mtu "${ns_b}"  veth_B-R1 1500
+
+	mtu "${ns_a}"  veth_A-R2 2000
+	mtu "${ns_r2}" veth_R2-A 2000
+	mtu "${ns_r2}" veth_R2-B 1500
+	mtu "${ns_b}"  veth_B-R2 1500
+
+	fail=0
+
+	# Add 100 addresses for veth endpoint on B reached by default A route
+	for i in $(seq 100 199); do
+		run_cmd ${ns_b} ip addr add "${dst_prefix1}${i}" dev veth_B-R1
+	done
+
+	# Create 100 cached route exceptions for path via R1, one via R2. Note
+	# that with IPv4 we need to actually cause a route lookup that matches
+	# the exception caused by ICMP, in order to actually have a cached
+	# route, so we need to ping each destination twice
+	for i in $(seq 100 199); do
+		run_cmd ${ns_a} ping -q -M want -i 0.1 -c 2 -s 1800 "${dst_prefix1}${i}"
+	done
+	run_cmd ${ns_a} ping -q -M want -i 0.1 -c 2 -s 1800 "${dst2}"
+
+	${ns_a} ip route list cache | wc -l
+
+	# Each exception is printed as two lines
+	if [ "$(${ns_a} ip route list cache | wc -l)" -ne 202 ]; then
+		err "  can't list cached exceptions"
+		fail=1
+	fi
+
+	run_cmd ${ns_a} ip route flush cache
+	pmtu1="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst_prefix}1)"
+	pmtu2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst_prefix}2)"
+	if [ -n "${pmtu1}" ] || [ -n "${pmtu2}" ] || \
+	   [ -n "$(${ns_a} ip route list cache)" ]; then
+		err "  can't flush cached exceptions"
+		fail=1
+	fi
+
+	return ${fail}
+}
+
 test_list_flush_ipv6_exception() {
 	setup namespaces routing || return 2
 	trace "${ns_a}"  veth_A-R1    "${ns_r1}" veth_R1-A \
Signature · 1 line
2.20.1
#2 lore raw

PATCH 2/2 selftests: pmtu: Make list_flush_ipv6_exception test more demanding

In reply to Stefano Brivio

Instead of just listing and flushing two cached exceptions, create a relatively big number of them, and count how many are listed. Single netlink dump messages contain approximately 25 entries each, and this way we can make sure the partial dump tracking mechanism is working properly.

While at it, also ensure that no cached routes can be listed after flush, and remove 'sleep 1' calls, they are not actually needed.

---
 tools/testing/selftests/net/pmtu.sh | 22 ++++++++++++++--------
 1 file changed, 14 insertions(+), 8 deletions(-)
tools/testing/selftests/net/pmtu.sh +14 -8
index 6c063b17d7d0..8998a7c03d3e 100755
@@ -1276,7 +1276,7 @@ test_list_flush_ipv6_exception() {
 	      "${ns_a}"  veth_A-R2    "${ns_r2}" veth_R2-A \
 	      "${ns_r2}" veth_R2-B    "${ns_b}"  veth_B-R2
 
-	dst1="${prefix6}:${b_r1}::1"
+	dst_prefix1="${prefix6}:${b_r1}::"
 	dst2="${prefix6}:${b_r2}::1"
 
 	# Set up initial MTU values
@@ -1292,20 +1292,26 @@ test_list_flush_ipv6_exception() {
 
 	fail=0
 
-	# Create route exceptions
-	run_cmd ${ns_a} ${ping6} -q -M want -i 0.1 -w 1 -s 1800 ${dst1}
-	run_cmd ${ns_a} ${ping6} -q -M want -i 0.1 -w 1 -s 1800 ${dst2}
+	# Add 100 addresses for veth endpoint on B reached by default A route
+	for i in $(seq 100 199); do
+		run_cmd ${ns_b} ip addr add "${dst_prefix1}${i}" dev veth_B-R1
+	done
 
-	if [ "$(${ns_a} ip -6 route list cache | wc -l)" -ne 2 ]; then
+	# Create 100 cached route exceptions for path via R1, one via R2
+	for i in $(seq 100 199); do
+		run_cmd ${ns_a} ping -q -M want -i 0.1 -w 1 -s 1800 "${dst_prefix1}${i}"
+	done
+	run_cmd ${ns_a} ping -q -M want -i 0.1 -w 1 -s 1800 "${dst2}"
+	if [ "$(${ns_a} ip -6 route list cache | wc -l)" -ne 101 ]; then
 		err "  can't list cached exceptions"
 		fail=1
 	fi
 
 	run_cmd ${ns_a} ip -6 route flush cache
-	sleep 1
-	pmtu1="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst1})"
+	pmtu1="$(route_get_dst_pmtu_from_exception "${ns_a}" "${dst_prefix1}100")"
 	pmtu2="$(route_get_dst_pmtu_from_exception "${ns_a}" ${dst2})"
-	if [ -n "${pmtu1}" ] || [ -n "${pmtu2}" ]; then
+	if [ -n "${pmtu1}" ] || [ -n "${pmtu2}" ] || \
+	   [ -n "$(${ns_a} ip -6 route list cache)" ]; then
 		err "  can't flush cached exceptions"
 		fail=1
 	fi
Signature · 1 line
2.20.1
maintainer
#3 lore raw
In reply to Stefano Brivio

From: Stefano Brivio <[email protected]>
Date: Sat, 15 Jun 2019 03:38:16 +0200

Quoted 3 lines

This series introduce a new test, list_flush_ipv4_exception, and improves the existing list_flush_ipv6_exception test by making it as demanding as the IPv4 one.

I suspect this will need a respin because semantics are still being discussed and I seem to recall a mention of there being some conflict with some of David A's changes.

#4 lore raw
Quoted 8 lines · On Sun, 16 Jun 2019 20:45:52 -0700 (PDT) David Miller <[email protected]>…

From: Stefano Brivio <[email protected]>
Date: Sat, 15 Jun 2019 03:38:16 +0200

Quoted 3 lines

This series introduce a new test, list_flush_ipv4_exception, and improves the existing list_flush_ipv6_exception test by making it as demanding as the IPv4 one.

I suspect this will need a respin because semantics are still being discussed

Maybe not a respin, because we're discussing netlink semantics and how many past versions of iproute2 need to work, whereas user interface and expectations of fixed, recent kernel/iproute2 are untouched.

Anyway, sure, it doesn't make sense to merge this before the fix is final -- I'll resend then.

This prompts some questions though (answer this quick survey and win a patch for netdev-FAQ.rst): when (and against which tree) do tests that are fixed by a recent patch need to be submitted? Is it a problem if the test is merged before the fix? Would a "dependency" note help?

Quoted 2 lines

and I seem to recall a mention of there being some conflict with some of David A's changes.

That was for e28799e52a0a ("selftests: pmtu: Introduce list_flush_ipv6_exception test case") on top of 438a9a856ba4 ("selftests: pmtu:
Add support for routing via nexthop objects"), but you already fixed the conflict.

That test case, by the way, will also fail until we agree on the fix.

Signature · 1 line
Stefano