netdev - [PATCH v1] selftests: openvswitch: retry instead of sleep

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240708134451.3489802-1-amorenoz@redhat.com>
Date: Mon,  8 Jul 2024 15:44:49 +0200
From: Adrian Moreno <amorenoz@...hat.com>
To: netdev@...r.kernel.org
Cc: aconole@...hat.com,
	Adrian Moreno <amorenoz@...hat.com>,
	Pravin B Shelar <pshelar@....org>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>,
	Paolo Abeni <pabeni@...hat.com>,
	Shuah Khan <shuah@...nel.org>,
	dev@...nvswitch.org,
	linux-kselftest@...r.kernel.org,
	linux-kernel@...r.kernel.org
Subject: [PATCH v1] selftests: openvswitch: retry instead of sleep

There are a couple of places where the test script "sleep"s to wait for
some external condition to be met.

This is error prone, specially in slow systems (identified in CI by
"KSFT_MACHINE_SLOW=yes").

To fix this, add a "ovs_wait" function that tries to execute a command
a few times until it succeeds. The timeout used is set to 5s for
"normal" systems and doubled if a slow CI machine is detected.

This should make the following work:

$ vng --build  \
    --config tools/testing/selftests/net/config \
    --config kernel/configs/debug.config

$ vng --run . --user root -- "make -C tools/testing/selftests/ \
    KSFT_MACHINE_SLOW=yes TARGETS=net/openvswitch run_tests"

Signed-off-by: Adrian Moreno <amorenoz@...hat.com>
---
 .../selftests/net/openvswitch/openvswitch.sh  | 49 ++++++++++++++++---
 .../selftests/net/openvswitch/ovs-dpctl.py    |  1 +
 2 files changed, 42 insertions(+), 8 deletions(-)

diff --git a/tools/testing/selftests/net/openvswitch/openvswitch.sh b/tools/testing/selftests/net/openvswitch/openvswitch.sh
index bc71dbc18b21..83407b42073a 100755
--- a/tools/testing/selftests/net/openvswitch/openvswitch.sh
+++ b/tools/testing/selftests/net/openvswitch/openvswitch.sh
@@ -11,6 +11,7 @@ ksft_skip=4
 PAUSE_ON_FAIL=no
 VERBOSE=0
 TRACING=0
+WAIT_TIMEOUT=5
 
 tests="
 	arp_ping				eth-arp: Basic arp ping between two NS
@@ -29,6 +30,32 @@ info() {
 	[ $VERBOSE = 0 ] || echo $*
 }
 
+ovs_wait() {
+	info "waiting $WAIT_TIMEOUT s for: $@"
+
+	"$@"
+	if [[ $? -eq 0 ]]; then
+		info "wait succeeded inmediately"
+		return 0
+	fi
+
+	# A quick re-check helps speed up small races in fast systems.
+	# However, fractional sleeps might not necessarily work.
+	local start=0
+	sleep 0.1 || { sleep 1; start=1; }
+
+	for (( i=start; i<WAIT_TIMEOUT; i++ )); do
+		"$@"
+		if [[ $? -eq 0 ]]; then
+			info "wait succeeded after $i seconds"
+			return 0
+		fi
+		sleep 1
+	done
+	info "wait failed after $i seconds"
+	return 1
+}
+
 ovs_base=`pwd`
 sbxs=
 sbx_add () {
@@ -278,20 +305,21 @@ test_psample() {
 
 	# Record psample data.
 	ovs_spawn_daemon "test_psample" python3 $ovs_base/ovs-dpctl.py psample-events
+	ovs_wait grep -q "listening for psample events" ${ovs_dir}/stdout
 
 	# Send a single ping.
-	sleep 1
 	ovs_sbx "test_psample" ip netns exec client ping -I c1 172.31.110.20 -c 1 || return 1
-	sleep 1
 
 	# We should have received one userspace action upcall and 2 psample packets.
-	grep -E "userspace action command" $ovs_dir/s0.out >/dev/null 2>&1 || return 1
+	ovs_wait grep -q "userspace action command" $ovs_dir/s0.out
+	[[ $? -eq 0 ]] || return 1
 
 	# client -> server samples should only contain the first 14 bytes of the packet.
-	grep -E "rate:4294967295,group:1,cookie:c0ffee data:[0-9a-f]{28}$" \
-			 $ovs_dir/stdout >/dev/null 2>&1 || return 1
-	grep -E "rate:4294967295,group:2,cookie:eeff0c" \
-			 $ovs_dir/stdout >/dev/null 2>&1 || return 1
+	ovs_wait grep -qE "rate:4294967295,group:1,cookie:c0ffee data:[0-9a-f]{28}$" $ovs_dir/stdout
+	[[ $? -eq 0 ]] || return 1
+
+	ovs_wait grep -q "rate:4294967295,group:2,cookie:eeff0c" $ovs_dir/stdout
+	[[ $? -eq 0 ]] || return 1
 
 	return 0
 }
@@ -711,7 +739,8 @@ test_upcall_interfaces() {
 	ovs_add_netns_and_veths "test_upcall_interfaces" ui0 upc left0 l0 \
 	    172.31.110.1/24 -u || return 1
 
-	sleep 1
+	ovs_wait grep -q "listening on upcall packet handler" ${ovs_dir}/left0.out
+
 	info "sending arping"
 	ip netns exec upc arping -I l0 172.31.110.20 -c 1 \
 	    >$ovs_dir/arping.stdout 2>$ovs_dir/arping.stderr
@@ -811,6 +840,10 @@ shift $(($OPTIND-1))
 IFS="	
 "
 
+if test "X$KSFT_MACHINE_SLOW" == "Xyes"; then
+	WAIT_TIMEOUT=10
+fi
+
 for arg do
 	# Check first that all requested tests are available before running any
 	command -v > /dev/null "test_${arg}" || { echo "=== Test ${arg} not found"; usage; }
diff --git a/tools/testing/selftests/net/openvswitch/ovs-dpctl.py b/tools/testing/selftests/net/openvswitch/ovs-dpctl.py
index 1e15b0818074..8a0396bfaf99 100644
--- a/tools/testing/selftests/net/openvswitch/ovs-dpctl.py
+++ b/tools/testing/selftests/net/openvswitch/ovs-dpctl.py
@@ -2520,6 +2520,7 @@ class PsampleEvent(EventSocket):
     marshal_class = psample_msg
 
     def read_samples(self):
+        print("listening for psample events", flush=True)
         while True:
             try:
                 for msg in self.get():
-- 
2.45.2