[PATCH v4 net 0/2] ipvlan: addrs_lock made per port

[PATCH v4 net 0/2] ipvlan: addrs_lock made per port

[Dmitry Skorodumov]

First patch fixes a rather minor issues that sometimes
ipvlan-addrs are modified without lock (because
for IPv6 addr can be sometimes added without RTNL)

diff from v3:
Main patch is not changed
Patch 2 (selftest) changed:
  - Remove unneeded modprobe
  - Number of threads is 8, if KSFT_MACHINE_SLOW==yes.
    It is needed, since on debug-build test may take more than 15 minutes.
  - Now veth is created in own namespace
  - Added comment about why test adds/removes random ip

diff from v2:
- Added a small self-test
- added early return in ipvlan_find_addr()
- the iterations over ipvlans in ipvlan_addr_busy()
must be protected by RCU
- Added simple self-test. I haven't invented anything
more sophisticated that this.

Dmitry Skorodumov (2):
  ipvlan: Make the addrs_lock be per port
  selftests: net: simple selftest for ipvtap

 drivers/net/ipvlan/ipvlan.h                |   2 +-
 drivers/net/ipvlan/ipvlan_core.c           |  16 +-
 drivers/net/ipvlan/ipvlan_main.c           |  49 +++---
 tools/testing/selftests/net/Makefile       |   1 +
 tools/testing/selftests/net/config         |   2 +
 tools/testing/selftests/net/ipvtap_test.sh | 167 +++++++++++++++++++++
 6 files changed, 207 insertions(+), 30 deletions(-)
 create mode 100755 tools/testing/selftests/net/ipvtap_test.sh

-- 
2.43.0

[PATCH v4 net 1/2] ipvlan: Make the addrs_lock be per port

[Dmitry Skorodumov]

Make the addrs_lock be per port, not per ipvlan dev.

Initial code seems to be written in the assumption,
that any address change must occur under RTNL.
But it is not so for the case of IPv6. So

1) Introduce per-port addrs_lock.

2) It was needed to fix places where it was forgotten
to take lock (ipvlan_open/ipvlan_close)

This appears to be a very minor problem though.
Since it's highly unlikely that ipvlan_add_addr() will
be called on 2 CPU simultaneously. But nevertheless,
this could cause:

1) False-negative of ipvlan_addr_busy(): one interface
iterated through all port->ipvlans + ipvlan->addrs
under some ipvlan spinlock, and another added IP
under its own lock. Though this is only possible
for IPv6, since looks like only ipvlan_addr6_event() can be
called without rtnl_lock.

2) Race since ipvlan_ht_addr_add(port) is called under
different ipvlan->addrs_lock locks

This should not affect performance, since add/remove IP
is a rare situation and spinlock is not taken on fast
paths.

Fixes: 8230819494b3 ("ipvlan: use per device spinlock to protect addrs list updates")
Signed-off-by: Dmitry Skorodumov <skorodumov.dmitry@huawei.com>
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
---
v4:
  - No changes since v3

 drivers/net/ipvlan/ipvlan.h      |  2 +-
 drivers/net/ipvlan/ipvlan_core.c | 16 +++++------
 drivers/net/ipvlan/ipvlan_main.c | 49 +++++++++++++++++++-------------
 3 files changed, 37 insertions(+), 30 deletions(-)

diff --git a/drivers/net/ipvlan/ipvlan.h b/drivers/net/ipvlan/ipvlan.h
index 50de3ee204db..80f84fc87008 100644
--- a/drivers/net/ipvlan/ipvlan.h
+++ b/drivers/net/ipvlan/ipvlan.h
@@ -69,7 +69,6 @@ struct ipvl_dev {
 	DECLARE_BITMAP(mac_filters, IPVLAN_MAC_FILTER_SIZE);
 	netdev_features_t	sfeatures;
 	u32			msg_enable;
-	spinlock_t		addrs_lock;
 };
 
 struct ipvl_addr {
@@ -90,6 +89,7 @@ struct ipvl_port {
 	struct net_device	*dev;
 	possible_net_t		pnet;
 	struct hlist_head	hlhead[IPVLAN_HASH_SIZE];
+	spinlock_t		addrs_lock; /* guards hash-table and addrs */
 	struct list_head	ipvlans;
 	u16			mode;
 	u16			flags;
diff --git a/drivers/net/ipvlan/ipvlan_core.c b/drivers/net/ipvlan/ipvlan_core.c
index 2efa3ba148aa..bdb3a46b327c 100644
--- a/drivers/net/ipvlan/ipvlan_core.c
+++ b/drivers/net/ipvlan/ipvlan_core.c
@@ -107,17 +107,15 @@ void ipvlan_ht_addr_del(struct ipvl_addr *addr)
 struct ipvl_addr *ipvlan_find_addr(const struct ipvl_dev *ipvlan,
 				   const void *iaddr, bool is_v6)
 {
-	struct ipvl_addr *addr, *ret = NULL;
+	struct ipvl_addr *addr;
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(addr, &ipvlan->addrs, anode) {
-		if (addr_equal(is_v6, addr, iaddr)) {
-			ret = addr;
-			break;
-		}
+	assert_spin_locked(&ipvlan->port->addrs_lock);
+
+	list_for_each_entry(addr, &ipvlan->addrs, anode) {
+		if (addr_equal(is_v6, addr, iaddr))
+			return addr;
 	}
-	rcu_read_unlock();
-	return ret;
+	return NULL;
 }
 
 bool ipvlan_addr_busy(struct ipvl_port *port, void *iaddr, bool is_v6)
diff --git a/drivers/net/ipvlan/ipvlan_main.c b/drivers/net/ipvlan/ipvlan_main.c
index 660f3db11766..baccdad695fd 100644
--- a/drivers/net/ipvlan/ipvlan_main.c
+++ b/drivers/net/ipvlan/ipvlan_main.c
@@ -75,6 +75,7 @@ static int ipvlan_port_create(struct net_device *dev)
 	for (idx = 0; idx < IPVLAN_HASH_SIZE; idx++)
 		INIT_HLIST_HEAD(&port->hlhead[idx]);
 
+	spin_lock_init(&port->addrs_lock);
 	skb_queue_head_init(&port->backlog);
 	INIT_WORK(&port->wq, ipvlan_process_multicast);
 	ida_init(&port->ida);
@@ -181,6 +182,7 @@ static void ipvlan_uninit(struct net_device *dev)
 static int ipvlan_open(struct net_device *dev)
 {
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
+	struct ipvl_port *port = ipvlan->port;
 	struct ipvl_addr *addr;
 
 	if (ipvlan->port->mode == IPVLAN_MODE_L3 ||
@@ -189,10 +191,10 @@ static int ipvlan_open(struct net_device *dev)
 	else
 		dev->flags &= ~IFF_NOARP;
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(addr, &ipvlan->addrs, anode)
+	spin_lock_bh(&port->addrs_lock);
+	list_for_each_entry(addr, &ipvlan->addrs, anode)
 		ipvlan_ht_addr_add(ipvlan, addr);
-	rcu_read_unlock();
+	spin_unlock_bh(&port->addrs_lock);
 
 	return 0;
 }
@@ -206,10 +208,10 @@ static int ipvlan_stop(struct net_device *dev)
 	dev_uc_unsync(phy_dev, dev);
 	dev_mc_unsync(phy_dev, dev);
 
-	rcu_read_lock();
-	list_for_each_entry_rcu(addr, &ipvlan->addrs, anode)
+	spin_lock_bh(&ipvlan->port->addrs_lock);
+	list_for_each_entry(addr, &ipvlan->addrs, anode)
 		ipvlan_ht_addr_del(addr);
-	rcu_read_unlock();
+	spin_unlock_bh(&ipvlan->port->addrs_lock);
 
 	return 0;
 }
@@ -579,7 +581,6 @@ int ipvlan_link_new(struct net_device *dev, struct rtnl_newlink_params *params,
 	if (!tb[IFLA_MTU])
 		ipvlan_adjust_mtu(ipvlan, phy_dev);
 	INIT_LIST_HEAD(&ipvlan->addrs);
-	spin_lock_init(&ipvlan->addrs_lock);
 
 	/* TODO Probably put random address here to be presented to the
 	 * world but keep using the physical-dev address for the outgoing
@@ -657,13 +658,13 @@ void ipvlan_link_delete(struct net_device *dev, struct list_head *head)
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
 	struct ipvl_addr *addr, *next;
 
-	spin_lock_bh(&ipvlan->addrs_lock);
+	spin_lock_bh(&ipvlan->port->addrs_lock);
 	list_for_each_entry_safe(addr, next, &ipvlan->addrs, anode) {
 		ipvlan_ht_addr_del(addr);
 		list_del_rcu(&addr->anode);
 		kfree_rcu(addr, rcu);
 	}
-	spin_unlock_bh(&ipvlan->addrs_lock);
+	spin_unlock_bh(&ipvlan->port->addrs_lock);
 
 	ida_free(&ipvlan->port->ida, dev->dev_id);
 	list_del_rcu(&ipvlan->pnode);
@@ -817,6 +818,8 @@ static int ipvlan_add_addr(struct ipvl_dev *ipvlan, void *iaddr, bool is_v6)
 {
 	struct ipvl_addr *addr;
 
+	assert_spin_locked(&ipvlan->port->addrs_lock);
+
 	addr = kzalloc(sizeof(struct ipvl_addr), GFP_ATOMIC);
 	if (!addr)
 		return -ENOMEM;
@@ -847,16 +850,16 @@ static void ipvlan_del_addr(struct ipvl_dev *ipvlan, void *iaddr, bool is_v6)
 {
 	struct ipvl_addr *addr;
 
-	spin_lock_bh(&ipvlan->addrs_lock);
+	spin_lock_bh(&ipvlan->port->addrs_lock);
 	addr = ipvlan_find_addr(ipvlan, iaddr, is_v6);
 	if (!addr) {
-		spin_unlock_bh(&ipvlan->addrs_lock);
+		spin_unlock_bh(&ipvlan->port->addrs_lock);
 		return;
 	}
 
 	ipvlan_ht_addr_del(addr);
 	list_del_rcu(&addr->anode);
-	spin_unlock_bh(&ipvlan->addrs_lock);
+	spin_unlock_bh(&ipvlan->port->addrs_lock);
 	kfree_rcu(addr, rcu);
 }
 
@@ -878,14 +881,14 @@ static int ipvlan_add_addr6(struct ipvl_dev *ipvlan, struct in6_addr *ip6_addr)
 {
 	int ret = -EINVAL;
 
-	spin_lock_bh(&ipvlan->addrs_lock);
+	spin_lock_bh(&ipvlan->port->addrs_lock);
 	if (ipvlan_addr_busy(ipvlan->port, ip6_addr, true))
 		netif_err(ipvlan, ifup, ipvlan->dev,
 			  "Failed to add IPv6=%pI6c addr for %s intf\n",
 			  ip6_addr, ipvlan->dev->name);
 	else
 		ret = ipvlan_add_addr(ipvlan, ip6_addr, true);
-	spin_unlock_bh(&ipvlan->addrs_lock);
+	spin_unlock_bh(&ipvlan->port->addrs_lock);
 	return ret;
 }
 
@@ -924,21 +927,24 @@ static int ipvlan_addr6_validator_event(struct notifier_block *unused,
 	struct in6_validator_info *i6vi = (struct in6_validator_info *)ptr;
 	struct net_device *dev = (struct net_device *)i6vi->i6vi_dev->dev;
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
+	int ret = NOTIFY_OK;
 
 	if (!ipvlan_is_valid_dev(dev))
 		return NOTIFY_DONE;
 
 	switch (event) {
 	case NETDEV_UP:
+		spin_lock_bh(&ipvlan->port->addrs_lock);
 		if (ipvlan_addr_busy(ipvlan->port, &i6vi->i6vi_addr, true)) {
 			NL_SET_ERR_MSG(i6vi->extack,
 				       "Address already assigned to an ipvlan device");
-			return notifier_from_errno(-EADDRINUSE);
+			ret = notifier_from_errno(-EADDRINUSE);
 		}
+		spin_unlock_bh(&ipvlan->port->addrs_lock);
 		break;
 	}
 
-	return NOTIFY_OK;
+	return ret;
 }
 #endif
 
@@ -946,14 +952,14 @@ static int ipvlan_add_addr4(struct ipvl_dev *ipvlan, struct in_addr *ip4_addr)
 {
 	int ret = -EINVAL;
 
-	spin_lock_bh(&ipvlan->addrs_lock);
+	spin_lock_bh(&ipvlan->port->addrs_lock);
 	if (ipvlan_addr_busy(ipvlan->port, ip4_addr, false))
 		netif_err(ipvlan, ifup, ipvlan->dev,
 			  "Failed to add IPv4=%pI4 on %s intf.\n",
 			  ip4_addr, ipvlan->dev->name);
 	else
 		ret = ipvlan_add_addr(ipvlan, ip4_addr, false);
-	spin_unlock_bh(&ipvlan->addrs_lock);
+	spin_unlock_bh(&ipvlan->port->addrs_lock);
 	return ret;
 }
 
@@ -995,21 +1001,24 @@ static int ipvlan_addr4_validator_event(struct notifier_block *unused,
 	struct in_validator_info *ivi = (struct in_validator_info *)ptr;
 	struct net_device *dev = (struct net_device *)ivi->ivi_dev->dev;
 	struct ipvl_dev *ipvlan = netdev_priv(dev);
+	int ret = NOTIFY_OK;
 
 	if (!ipvlan_is_valid_dev(dev))
 		return NOTIFY_DONE;
 
 	switch (event) {
 	case NETDEV_UP:
+		spin_lock_bh(&ipvlan->port->addrs_lock);
 		if (ipvlan_addr_busy(ipvlan->port, &ivi->ivi_addr, false)) {
 			NL_SET_ERR_MSG(ivi->extack,
 				       "Address already assigned to an ipvlan device");
-			return notifier_from_errno(-EADDRINUSE);
+			ret = notifier_from_errno(-EADDRINUSE);
 		}
+		spin_unlock_bh(&ipvlan->port->addrs_lock);
 		break;
 	}
 
-	return NOTIFY_OK;
+	return ret;
 }
 
 static struct notifier_block ipvlan_addr4_notifier_block __read_mostly = {
-- 
2.43.0

[PATCH v4 net 2/2] selftests: net: simple selftest for ipvtap

[Dmitry Skorodumov]

This is a simple ipvtap test to test handling
IP-address add/remove on ipvlan interface.

It creates a veth-interface and then creates several
network-namespace with ipvlan0 interface in it linked to veth.

Then it starts to add/remove addresses on ipvlan0 interfaces
in several threads.

At finish, it checks that there is no duplicated addresses.

Signed-off-by: Dmitry Skorodumov <skorodumov.dmitry@huawei.com>
---
v4:
  - Removed unneeded modprobe
  - Number of threads is 8, if KSFT_MACHINE_SLOW==yes.
    It is needed, since on debug-build test may take more than 15 minutes.
  - Now veth is created in own namespace
  - Added comment about why test adds/removes random ip

 tools/testing/selftests/net/Makefile       |   1 +
 tools/testing/selftests/net/config         |   2 +
 tools/testing/selftests/net/ipvtap_test.sh | 166 +++++++++++++++++++++
 3 files changed, 169 insertions(+)
 create mode 100755 tools/testing/selftests/net/ipvtap_test.sh

diff --git a/tools/testing/selftests/net/Makefile b/tools/testing/selftests/net/Makefile
index b66ba04f19d9..45c4ea381bc3 100644
--- a/tools/testing/selftests/net/Makefile
+++ b/tools/testing/selftests/net/Makefile
@@ -48,6 +48,7 @@ TEST_PROGS := \
 	ipv6_flowlabel.sh \
 	ipv6_force_forwarding.sh \
 	ipv6_route_update_soft_lockup.sh \
+	ipvtap_test.sh \
 	l2_tos_ttl_inherit.sh \
 	l2tp.sh \
 	link_netns.py \
diff --git a/tools/testing/selftests/net/config b/tools/testing/selftests/net/config
index 1e1f253118f5..5702ab8fa5ad 100644
--- a/tools/testing/selftests/net/config
+++ b/tools/testing/selftests/net/config
@@ -48,6 +48,7 @@ CONFIG_IPV6_SEG6_LWTUNNEL=y
 CONFIG_IPV6_SIT=y
 CONFIG_IPV6_VTI=y
 CONFIG_IPVLAN=m
+CONFIG_IPVTAP=m
 CONFIG_KALLSYMS=y
 CONFIG_L2TP=m
 CONFIG_L2TP_ETH=m
@@ -122,6 +123,7 @@ CONFIG_TEST_BPF=m
 CONFIG_TLS=m
 CONFIG_TRACEPOINTS=y
 CONFIG_TUN=y
+CONFIG_TAP=m
 CONFIG_USER_NS=y
 CONFIG_VETH=y
 CONFIG_VLAN_8021Q=y
diff --git a/tools/testing/selftests/net/ipvtap_test.sh b/tools/testing/selftests/net/ipvtap_test.sh
new file mode 100755
index 000000000000..b4e18fc7ada0
--- /dev/null
+++ b/tools/testing/selftests/net/ipvtap_test.sh
@@ -0,0 +1,167 @@
+#!/bin/bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# Simple tests for ipvtap
+
+
+#
+# The testing environment looks this way:
+#
+# |------HNS-------|     |------PHY-------|
+# |      veth<----------------->veth      |
+# |------|--|------|     |----------------|
+#        |  |
+#        |  |            |-----TST0-------|
+#        |  |------------|----ipvlan      |
+#        |               |----------------|
+#        |
+#        |               |-----TST1-------|
+#        |---------------|----ipvlan      |
+#                        |----------------|
+#
+
+ALL_TESTS="
+	test_ip_set
+"
+
+source lib.sh
+
+DEBUG=0
+
+VETH_HOST=vethtst.h
+VETH_PHY=vethtst.p
+
+NS_COUNT=32
+IP_ITERATIONS=1024
+[ "$KSFT_MACHINE_SLOW" = "yes" ] && NS_COUNT=8
+
+ns_run() {
+	ns=$1
+	shift
+	if [[ "$ns" == "global" ]]; then
+		"$@" >/dev/null
+	else
+		ip netns exec "$ns" "$@" >/dev/null
+	fi
+}
+
+test_ip_setup_env() {
+	setup_ns NS_PHY
+	setup_ns HST_NS
+
+	# setup simulated other-host (phy) and host itself
+	ns_run "$HST_NS" ip link add $VETH_HOST type veth peer name $VETH_PHY \
+		netns "$NS_PHY" >/dev/null
+	ns_run "$HST_NS" ip link set $VETH_HOST up
+	ns_run "$NS_PHY" ip link set $VETH_PHY up
+
+	for ((i=0; i<NS_COUNT; i++)); do
+		setup_ns ipvlan_ns_$i
+		ns="ipvlan_ns_$i"
+		if [ "$DEBUG" = "1" ]; then
+			echo "created NS ${!ns}"
+		fi
+		if ! ns_run "$HST_NS" ip link add netns ${!ns} ipvlan0 \
+		    link $VETH_HOST \
+		    type ipvtap mode l2 bridge; then
+			exit_error "FAIL: Failed to configure ipvlan link."
+		fi
+	done
+}
+
+test_ip_cleanup_env() {
+	ns_run "$HST_NS" ip link del $VETH_HOST
+	cleanup_all_ns
+}
+
+exit_error() {
+	echo "$1"
+	exit $ksft_fail
+}
+
+rnd() {
+	echo $(( RANDOM % 32 + 16 ))
+}
+
+test_ip_set_thread() {
+	# Here we are trying to create some IP conflicts between namespaces.
+	# If just add/remove IP, nothing interesting will happen.
+	# But if add random IP and then remove random IP,
+	# eventually conflicts start to apear.
+	ip link set ipvlan0 up
+	for ((i=0; i<IP_ITERATIONS; i++)); do
+		v=$(rnd)
+		ip a a "172.25.0.$v/24" dev ipvlan0 2>/dev/null
+		ip a a "fc00::$v/64" dev ipvlan0 2>/dev/null
+		v=$(rnd)
+		ip a d "172.25.0.$v/24" dev ipvlan0 2>/dev/null
+		ip a d "fc00::$v/64" dev ipvlan0 2>/dev/null
+	done
+}
+
+test_ip_set() {
+	RET=0
+
+	trap test_ip_cleanup_env EXIT
+
+	test_ip_setup_env
+
+	declare -A ns_pids
+	for ((i=0; i<NS_COUNT; i++)); do
+		ns="ipvlan_ns_$i"
+		ns_run ${!ns} bash -c "$0 test_ip_set_thread"&
+		ns_pids[$i]=$!
+	done
+
+	for ((i=0; i<NS_COUNT; i++)); do
+		wait "${ns_pids[$i]}"
+	done
+
+	declare -A all_ips
+	for ((i=0; i<NS_COUNT; i++)); do
+		ns="ipvlan_ns_$i"
+		ip_output=$(ip netns exec ${!ns} ip a l dev ipvlan0 | grep inet)
+		while IFS= read -r nsip_out; do
+			if [[ -z $nsip_out ]]; then
+				continue;
+			fi
+			nsip=$(awk '{print $2}' <<< "$nsip_out")
+			if [[ -v all_ips[$nsip] ]]; then
+				RET=$ksft_fail
+				log_test "conflict for $nsip"
+				return "$RET"
+			else
+				all_ips[$nsip]=$i
+			fi
+		done <<< "$ip_output"
+	done
+
+	if [ "$DEBUG" = "1" ]; then
+		for key in "${!all_ips[@]}"; do
+			echo "$key: ${all_ips[$key]}"
+		done
+	fi
+
+	trap - EXIT
+	test_ip_cleanup_env
+
+	log_test "test multithreaded ip set"
+}
+
+if [[ "$1" == "-d" ]]; then
+	DEBUG=1
+	shift
+fi
+
+if [[ "$1" == "-t" ]]; then
+	shift
+	TESTS="$*"
+fi
+
+if [[ "$1" == "test_ip_set_thread" ]]; then
+	test_ip_set_thread
+else
+	require_command ip
+
+	tests_run
+fi
-- 
2.43.0

Re: [PATCH v4 net 2/2] selftests: net: simple selftest for ipvtap

[Jakub Kicinski]

On Tue, 30 Dec 2025 17:03:24 +0300 Dmitry Skorodumov wrote:
> @@ -122,6 +123,7 @@ CONFIG_TEST_BPF=m
>  CONFIG_TLS=m
>  CONFIG_TRACEPOINTS=y
>  CONFIG_TUN=y
> +CONFIG_TAP=m

alphabetical order please

>  CONFIG_USER_NS=y
>  CONFIG_VETH=y
>  CONFIG_VLAN_8021Q=y
> diff --git a/tools/testing/selftests/net/ipvtap_test.sh b/tools/testing/selftests/net/ipvtap_test.sh
> new file mode 100755

> +NS_COUNT=32
> +IP_ITERATIONS=1024
> +[ "$KSFT_MACHINE_SLOW" = "yes" ] && NS_COUNT=8

This still fails on our slow VMs with debug kernel enabled (even tho
they set KSFT_MACHINE_SLOW). The test runs for 15+ minutes without
printing anything. That's pretty bad, humans running this tests will
also have no idea what's going on. Maybe limit the runtime of the test
as well as the iterations? To 1min? It takes <50sec on non-debug kernels
in our CI.
-- 
pw-bot: cr