* [PATCH net v8 1/4] net: netpoll: fix incorrect refcount handling causing incorrect cleanup
2025-11-04 17:37 [PATCH net v8 0/4] net: netpoll: fix memory leak and add comprehensive selftests Breno Leitao
@ 2025-11-04 17:37 ` Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 2/4] selftest: netcons: refactor target creation Breno Leitao
` (2 subsequent siblings)
3 siblings, 0 replies; 6+ messages in thread
From: Breno Leitao @ 2025-11-04 17:37 UTC (permalink / raw)
To: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Shuah Khan, Simon Horman, david decotigny
Cc: linux-kernel, netdev, linux-kselftest, asantostc, efault, calvin,
kernel-team, calvin, jv, Breno Leitao, stable
commit efa95b01da18 ("netpoll: fix use after free") incorrectly
ignored the refcount and prematurely set dev->npinfo to NULL during
netpoll cleanup, leading to improper behavior and memory leaks.
Scenario causing lack of proper cleanup:
1) A netpoll is associated with a NIC (e.g., eth0) and netdev->npinfo is
allocated, and refcnt = 1
- Keep in mind that npinfo is shared among all netpoll instances. In
this case, there is just one.
2) Another netpoll is also associated with the same NIC and
npinfo->refcnt += 1.
- Now dev->npinfo->refcnt = 2;
- There is just one npinfo associated to the netdev.
3) When the first netpolls goes to clean up:
- The first cleanup succeeds and clears np->dev->npinfo, ignoring
refcnt.
- It basically calls `RCU_INIT_POINTER(np->dev->npinfo, NULL);`
- Set dev->npinfo = NULL, without proper cleanup
- No ->ndo_netpoll_cleanup() is either called
4) Now the second target tries to clean up
- The second cleanup fails because np->dev->npinfo is already NULL.
* In this case, ops->ndo_netpoll_cleanup() was never called, and
the skb pool is not cleaned as well (for the second netpoll
instance)
- This leaks npinfo and skbpool skbs, which is clearly reported by
kmemleak.
Revert commit efa95b01da18 ("netpoll: fix use after free") and adds
clarifying comments emphasizing that npinfo cleanup should only happen
once the refcount reaches zero, ensuring stable and correct netpoll
behavior.
Cc: <stable@vger.kernel.org> # 3.17.x
Cc: Jay Vosburgh <jv@jvosburgh.net>
Fixes: efa95b01da18 ("netpoll: fix use after free")
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Simon Horman <horms@kernel.org>
---
net/core/netpoll.c | 7 +++++--
1 file changed, 5 insertions(+), 2 deletions(-)
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index 60a05d3b7c249..f4a0023428265 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -814,6 +814,10 @@ static void __netpoll_cleanup(struct netpoll *np)
if (!npinfo)
return;
+ /* At this point, there is a single npinfo instance per netdevice, and
+ * its refcnt tracks how many netpoll structures are linked to it. We
+ * only perform npinfo cleanup when the refcnt decrements to zero.
+ */
if (refcount_dec_and_test(&npinfo->refcnt)) {
const struct net_device_ops *ops;
@@ -823,8 +827,7 @@ static void __netpoll_cleanup(struct netpoll *np)
RCU_INIT_POINTER(np->dev->npinfo, NULL);
call_rcu(&npinfo->rcu, rcu_cleanup_netpoll_info);
- } else
- RCU_INIT_POINTER(np->dev->npinfo, NULL);
+ }
skb_pool_flush(np);
}
--
2.47.3
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH net v8 2/4] selftest: netcons: refactor target creation
2025-11-04 17:37 [PATCH net v8 0/4] net: netpoll: fix memory leak and add comprehensive selftests Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 1/4] net: netpoll: fix incorrect refcount handling causing incorrect cleanup Breno Leitao
@ 2025-11-04 17:37 ` Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 3/4] selftest: netcons: create a torture test Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 4/4] selftest: netcons: add test for netconsole over bonded interfaces Breno Leitao
3 siblings, 0 replies; 6+ messages in thread
From: Breno Leitao @ 2025-11-04 17:37 UTC (permalink / raw)
To: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Shuah Khan, Simon Horman, david decotigny
Cc: linux-kernel, netdev, linux-kselftest, asantostc, efault, calvin,
kernel-team, calvin, jv, Breno Leitao
Extract the netconsole target creation from create_dynamic_target(), by
moving it from create_dynamic_target() into a new helper function. This
enables other tests to use the creation of netconsole targets with
arbitrary parameters and no sleep.
The new helper will be utilized by forthcoming torture-type selftests
that require dynamic target management.
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Simon Horman <horms@kernel.org>
---
.../selftests/drivers/net/lib/sh/lib_netcons.sh | 30 ++++++++++++++--------
1 file changed, 19 insertions(+), 11 deletions(-)
diff --git a/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh b/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh
index 8e1085e896472..9b5ef8074440c 100644
--- a/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh
+++ b/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh
@@ -113,31 +113,39 @@ function set_network() {
configure_ip
}
-function create_dynamic_target() {
- local FORMAT=${1:-"extended"}
+function _create_dynamic_target() {
+ local FORMAT="${1:?FORMAT parameter required}"
+ local NCPATH="${2:?NCPATH parameter required}"
DSTMAC=$(ip netns exec "${NAMESPACE}" \
ip link show "${DSTIF}" | awk '/ether/ {print $2}')
# Create a dynamic target
- mkdir "${NETCONS_PATH}"
+ mkdir "${NCPATH}"
- echo "${DSTIP}" > "${NETCONS_PATH}"/remote_ip
- echo "${SRCIP}" > "${NETCONS_PATH}"/local_ip
- echo "${DSTMAC}" > "${NETCONS_PATH}"/remote_mac
- echo "${SRCIF}" > "${NETCONS_PATH}"/dev_name
+ echo "${DSTIP}" > "${NCPATH}"/remote_ip
+ echo "${SRCIP}" > "${NCPATH}"/local_ip
+ echo "${DSTMAC}" > "${NCPATH}"/remote_mac
+ echo "${SRCIF}" > "${NCPATH}"/dev_name
if [ "${FORMAT}" == "basic" ]
then
# Basic target does not support release
- echo 0 > "${NETCONS_PATH}"/release
- echo 0 > "${NETCONS_PATH}"/extended
+ echo 0 > "${NCPATH}"/release
+ echo 0 > "${NCPATH}"/extended
elif [ "${FORMAT}" == "extended" ]
then
- echo 1 > "${NETCONS_PATH}"/extended
+ echo 1 > "${NCPATH}"/extended
fi
- echo 1 > "${NETCONS_PATH}"/enabled
+ echo 1 > "${NCPATH}"/enabled
+
+}
+
+function create_dynamic_target() {
+ local FORMAT=${1:-"extended"}
+ local NCPATH=${2:-"$NETCONS_PATH"}
+ _create_dynamic_target "${FORMAT}" "${NCPATH}"
# This will make sure that the kernel was able to
# load the netconsole driver configuration. The console message
--
2.47.3
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH net v8 3/4] selftest: netcons: create a torture test
2025-11-04 17:37 [PATCH net v8 0/4] net: netpoll: fix memory leak and add comprehensive selftests Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 1/4] net: netpoll: fix incorrect refcount handling causing incorrect cleanup Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 2/4] selftest: netcons: refactor target creation Breno Leitao
@ 2025-11-04 17:37 ` Breno Leitao
2025-11-04 17:37 ` [PATCH net v8 4/4] selftest: netcons: add test for netconsole over bonded interfaces Breno Leitao
3 siblings, 0 replies; 6+ messages in thread
From: Breno Leitao @ 2025-11-04 17:37 UTC (permalink / raw)
To: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Shuah Khan, Simon Horman, david decotigny
Cc: linux-kernel, netdev, linux-kselftest, asantostc, efault, calvin,
kernel-team, calvin, jv, Breno Leitao
Create a netconsole test that puts a lot of pressure on the netconsole
list manipulation. Do it by creating dynamic targets and deleting
targets while messages are being sent. Also put interface down while the
messages are being sent, as creating parallel targets.
The code launches three background jobs on distinct schedules:
* Toggle netcons target every 30 iterations
* create and delete random_target every 50 iterations
* toggle iface every 70 iterations
This creates multiple concurrency sources that interact with netconsole
states. This is good practice to simulate stress, and exercise netpoll
and netconsole locks.
This test already found an issue as reported in [1]
Link: https://lore.kernel.org/all/20250901-netpoll_memleak-v1-1-34a181977dfc@debian.org/ [1]
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Andre Carvalho <asantostc@gmail.com>
Reviewed-by: Simon Horman <horms@kernel.org>
---
tools/testing/selftests/drivers/net/Makefile | 1 +
.../selftests/drivers/net/netcons_torture.sh | 130 +++++++++++++++++++++
2 files changed, 131 insertions(+)
diff --git a/tools/testing/selftests/drivers/net/Makefile b/tools/testing/selftests/drivers/net/Makefile
index 6e41635bd55a4..71ee69e524d77 100644
--- a/tools/testing/selftests/drivers/net/Makefile
+++ b/tools/testing/selftests/drivers/net/Makefile
@@ -18,6 +18,7 @@ TEST_PROGS := \
netcons_fragmented_msg.sh \
netcons_overflow.sh \
netcons_sysdata.sh \
+ netcons_torture.sh \
netpoll_basic.py \
ping.py \
psp.py \
diff --git a/tools/testing/selftests/drivers/net/netcons_torture.sh b/tools/testing/selftests/drivers/net/netcons_torture.sh
new file mode 100755
index 0000000000000..2ce9ee3719d1a
--- /dev/null
+++ b/tools/testing/selftests/drivers/net/netcons_torture.sh
@@ -0,0 +1,130 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: GPL-2.0
+
+# Repeatedly send kernel messages, toggles netconsole targets on and off,
+# creates and deletes targets in parallel, and toggles the source interface to
+# simulate stress conditions.
+#
+# This test aims to verify the robustness of netconsole under dynamic
+# configurations and concurrent operations.
+#
+# The major goal is to run this test with LOCKDEP, Kmemleak and KASAN to make
+# sure no issues is reported.
+#
+# Author: Breno Leitao <leitao@debian.org>
+
+set -euo pipefail
+
+SCRIPTDIR=$(dirname "$(readlink -e "${BASH_SOURCE[0]}")")
+
+source "${SCRIPTDIR}"/lib/sh/lib_netcons.sh
+
+# Number of times the main loop run
+ITERATIONS=${1:-150}
+
+# Only test extended format
+FORMAT="extended"
+# And ipv6 only
+IP_VERSION="ipv6"
+
+# Create, enable and delete some targets.
+create_and_delete_random_target() {
+ COUNT=2
+ RND_PREFIX=$(mktemp -u netcons_rnd_XXXX_)
+
+ if [ -d "${NETCONS_CONFIGFS}/${RND_PREFIX}${COUNT}" ] || \
+ [ -d "${NETCONS_CONFIGFS}/${RND_PREFIX}0" ]; then
+ echo "Function didn't finish yet, skipping it." >&2
+ return
+ fi
+
+ # enable COUNT targets
+ for i in $(seq ${COUNT})
+ do
+ RND_TARGET="${RND_PREFIX}"${i}
+ RND_TARGET_PATH="${NETCONS_CONFIGFS}"/"${RND_TARGET}"
+
+ # Basic population so the target can come up
+ _create_dynamic_target "${FORMAT}" "${RND_TARGET_PATH}"
+ done
+
+ echo "netconsole selftest: ${COUNT} additional targets were created" > /dev/kmsg
+ # disable them all
+ for i in $(seq ${COUNT})
+ do
+ RND_TARGET="${RND_PREFIX}"${i}
+ RND_TARGET_PATH="${NETCONS_CONFIGFS}"/"${RND_TARGET}"
+ if [[ $(cat "${RND_TARGET_PATH}/enabled") -eq 1 ]]
+ then
+ echo 0 > "${RND_TARGET_PATH}"/enabled
+ fi
+ rmdir "${RND_TARGET_PATH}"
+ done
+}
+
+# Disable and enable the target mid-air, while messages
+# are being transmitted.
+toggle_netcons_target() {
+ for i in $(seq 2)
+ do
+ if [ ! -d "${NETCONS_PATH}" ]
+ then
+ break
+ fi
+ echo 0 > "${NETCONS_PATH}"/enabled 2> /dev/null || true
+ # Try to enable a bit harder, given it might fail to enable
+ # Write to `enabled` might fail depending on the lock, which is
+ # highly contentious here
+ for _ in $(seq 5)
+ do
+ echo 1 > "${NETCONS_PATH}"/enabled 2> /dev/null || true
+ done
+ done
+}
+
+toggle_iface(){
+ ip link set "${SRCIF}" down
+ ip link set "${SRCIF}" up
+}
+
+# Start here
+
+modprobe netdevsim 2> /dev/null || true
+modprobe netconsole 2> /dev/null || true
+
+# Check for basic system dependency and exit if not found
+check_for_dependencies
+# Set current loglevel to KERN_INFO(6), and default to KERN_NOTICE(5)
+echo "6 5" > /proc/sys/kernel/printk
+# Remove the namespace, interfaces and netconsole target on exit
+trap cleanup EXIT
+# Create one namespace and two interfaces
+set_network "${IP_VERSION}"
+# Create a dynamic target for netconsole
+create_dynamic_target "${FORMAT}"
+
+for i in $(seq "$ITERATIONS")
+do
+ for _ in $(seq 10)
+ do
+ echo "${MSG}: ${TARGET} ${i}" > /dev/kmsg
+ done
+ wait
+
+ if (( i % 30 == 0 )); then
+ toggle_netcons_target &
+ fi
+
+ if (( i % 50 == 0 )); then
+ # create some targets, enable them, send msg and disable
+ # all in a parallel thread
+ create_and_delete_random_target &
+ fi
+
+ if (( i % 70 == 0 )); then
+ toggle_iface &
+ fi
+done
+wait
+
+exit "${EXIT_STATUS}"
--
2.47.3
^ permalink raw reply related [flat|nested] 6+ messages in thread* [PATCH net v8 4/4] selftest: netcons: add test for netconsole over bonded interfaces
2025-11-04 17:37 [PATCH net v8 0/4] net: netpoll: fix memory leak and add comprehensive selftests Breno Leitao
` (2 preceding siblings ...)
2025-11-04 17:37 ` [PATCH net v8 3/4] selftest: netcons: create a torture test Breno Leitao
@ 2025-11-04 17:37 ` Breno Leitao
2025-11-05 16:04 ` Breno Leitao
3 siblings, 1 reply; 6+ messages in thread
From: Breno Leitao @ 2025-11-04 17:37 UTC (permalink / raw)
To: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Shuah Khan, Simon Horman, david decotigny
Cc: linux-kernel, netdev, linux-kselftest, asantostc, efault, calvin,
kernel-team, calvin, jv, Breno Leitao
This patch adds a selftest that verifies netconsole functionality
over bonded network interfaces using netdevsim. It sets up two bonded
interfaces acting as transmit (TX) and receive (RX) ends, placed in
separate network namespaces. The test sends kernel log messages and
verifies that they are properly received on the bonded RX interfaces
with both IPv4 and IPv6, and using basic and extended netconsole
formats.
This patchset aims to test a long-standing netpoll subsystem where
netpoll has multiple users. (in this case netconsole and bonding). A
similar selftest has been discussed in [1] and [2].
This test also tries to enable bonding and netpoll in different order,
just to guarantee that all the possibilities are exercised.
Link: https://lore.kernel.org/all/20250905-netconsole_torture-v3-0-875c7febd316@debian.org/ [1]
Link: https://lore.kernel.org/lkml/96b940137a50e5c387687bb4f57de8b0435a653f.1404857349.git.decot@googlers.com/ [2]
Signed-off-by: Breno Leitao <leitao@debian.org>
---
.../testing/selftests/drivers/net/bonding/Makefile | 2 +
tools/testing/selftests/drivers/net/bonding/config | 4 +
.../drivers/net/bonding/netcons_over_bonding.sh | 361 +++++++++++++++++++++
.../selftests/drivers/net/lib/sh/lib_netcons.sh | 56 +++-
4 files changed, 416 insertions(+), 7 deletions(-)
diff --git a/tools/testing/selftests/drivers/net/bonding/Makefile b/tools/testing/selftests/drivers/net/bonding/Makefile
index 402d4ee84f2e8..6c5c60adb5e85 100644
--- a/tools/testing/selftests/drivers/net/bonding/Makefile
+++ b/tools/testing/selftests/drivers/net/bonding/Makefile
@@ -14,6 +14,7 @@ TEST_PROGS := \
dev_addr_lists.sh \
mode-1-recovery-updelay.sh \
mode-2-recovery-updelay.sh \
+ netcons_over_bonding.sh \
# end of TEST_PROGS
TEST_FILES := \
@@ -24,6 +25,7 @@ TEST_FILES := \
TEST_INCLUDES := \
../../../net/lib.sh \
+ ../lib/sh/lib_netcons.sh \
../../../net/forwarding/lib.sh \
# end of TEST_INCLUDES
diff --git a/tools/testing/selftests/drivers/net/bonding/config b/tools/testing/selftests/drivers/net/bonding/config
index 6bb290abd48bf..57f5a5d5cea67 100644
--- a/tools/testing/selftests/drivers/net/bonding/config
+++ b/tools/testing/selftests/drivers/net/bonding/config
@@ -1,5 +1,6 @@
CONFIG_BONDING=y
CONFIG_BRIDGE=y
+CONFIG_CONFIGFS_FS=y
CONFIG_DUMMY=y
CONFIG_INET_ESP=y
CONFIG_INET_ESP_OFFLOAD=y
@@ -11,6 +12,9 @@ CONFIG_NET_CLS_FLOWER=y
CONFIG_NET_CLS_MATCHALL=m
CONFIG_NETDEVSIM=m
CONFIG_NET_SCH_INGRESS=y
+CONFIG_NETCONSOLE=m
+CONFIG_NETCONSOLE_DYNAMIC=y
+CONFIG_NETCONSOLE_EXTENDED_LOG=y
CONFIG_NLMON=y
CONFIG_VETH=y
CONFIG_VLAN_8021Q=m
diff --git a/tools/testing/selftests/drivers/net/bonding/netcons_over_bonding.sh b/tools/testing/selftests/drivers/net/bonding/netcons_over_bonding.sh
new file mode 100755
index 0000000000000..477cc9379500a
--- /dev/null
+++ b/tools/testing/selftests/drivers/net/bonding/netcons_over_bonding.sh
@@ -0,0 +1,361 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: GPL-2.0
+#
+# This selftest exercises trying to have multiple netpoll users at the same
+# time.
+#
+# This selftest has multiple smalls test inside, and the goal is to
+# get interfaces with bonding and netconsole in different orders in order
+# to catch any possible issue.
+#
+# The main test composes of four interfaces being created using netdevsim; two
+# of them are bonded to serve as the netconsole's transmit interface. The
+# remaining two interfaces are similarly bonded and assigned to a separate
+# network namespace, which acts as the receive interface, where socat monitors
+# for incoming messages.
+#
+# A netconsole message is then sent to ensure it is properly received across
+# this configuration.
+#
+# Later, run a few other tests, to make sure that bonding and netconsole
+# cannot coexist.
+#
+# The test's objective is to exercise netpoll usage when managed simultaneously
+# by multiple subsystems (netconsole and bonding).
+#
+# Author: Breno Leitao <leitao@debian.org>
+
+set -euo pipefail
+
+SCRIPTDIR=$(dirname "$(readlink -e "${BASH_SOURCE[0]}")")
+
+source "${SCRIPTDIR}"/../lib/sh/lib_netcons.sh
+
+modprobe netdevsim 2> /dev/null || true
+modprobe netconsole 2> /dev/null || true
+modprobe bonding 2> /dev/null || true
+modprobe veth 2> /dev/null || true
+
+# The content of kmsg will be save to the following file
+OUTPUT_FILE="/tmp/${TARGET}"
+
+# Check for basic system dependency and exit if not found
+check_for_dependencies
+# Set current loglevel to KERN_INFO(6), and default to KERN_NOTICE(5)
+echo "6 5" > /proc/sys/kernel/printk
+# Remove the namespace, interfaces and netconsole target on exit
+trap cleanup_bond EXIT
+
+FORMAT="extended"
+IP_VERSION="ipv4"
+VETH0="veth"$(( RANDOM % 256))
+VETH1="veth"$((256 + RANDOM % 256))
+TXNS=""
+RXNS=""
+
+# Create "bond_tx_XX" and "bond_rx_XX" interfaces, and set DSTIF and SRCIF with
+# the bonding interfaces
+function setup_bonding_ifaces() {
+ local RAND=$(( RANDOM % 100 ))
+ BOND_TX_MAIN_IF="bond_tx_$RAND"
+ BOND_RX_MAIN_IF="bond_rx_$RAND"
+
+ # Setup TX
+ if ! ip -n "${TXNS}" link add "${BOND_TX_MAIN_IF}" type bond mode balance-rr
+ then
+ echo "Failed to create bond TX interface. Is CONFIG_BONDING set?" >&2
+ # only clean nsim ifaces and namespace. Nothing else has been
+ # initialized
+ cleanup_bond_nsim
+ trap - EXIT
+ exit "${ksft_skip}"
+ fi
+
+ # create_netdevsim() got the interface up, but it needs to be down
+ # before being enslaved.
+ ip -n "${TXNS}" \
+ link set "${BOND_TX1_SLAVE_IF}" down
+ ip -n "${TXNS}" \
+ link set "${BOND_TX2_SLAVE_IF}" down
+ ip -n "${TXNS}" \
+ link set "${BOND_TX1_SLAVE_IF}" master "${BOND_TX_MAIN_IF}"
+ ip -n "${TXNS}" \
+ link set "${BOND_TX2_SLAVE_IF}" master "${BOND_TX_MAIN_IF}"
+ ip -n "${TXNS}" \
+ link set "${BOND_TX_MAIN_IF}" up
+
+ # Setup RX
+ ip -n "${RXNS}" \
+ link add "${BOND_RX_MAIN_IF}" type bond mode balance-rr
+ ip -n "${RXNS}" \
+ link set "${BOND_RX1_SLAVE_IF}" down
+ ip -n "${RXNS}" \
+ link set "${BOND_RX2_SLAVE_IF}" down
+ ip -n "${RXNS}" \
+ link set "${BOND_RX1_SLAVE_IF}" master "${BOND_RX_MAIN_IF}"
+ ip -n "${RXNS}" \
+ link set "${BOND_RX2_SLAVE_IF}" master "${BOND_RX_MAIN_IF}"
+ ip -n "${RXNS}" \
+ link set "${BOND_RX_MAIN_IF}" up
+
+ export DSTIF="${BOND_RX_MAIN_IF}"
+ export SRCIF="${BOND_TX_MAIN_IF}"
+}
+
+# Create 4 netdevsim interfaces. Two of them will be bound to TX bonding iface
+# and the other two will be bond to the RX interface (on the other namespace)
+function create_ifaces_bond() {
+ BOND_TX1_SLAVE_IF=$(create_netdevsim "${NSIM_BOND_TX_1}" "${TXNS}")
+ BOND_TX2_SLAVE_IF=$(create_netdevsim "${NSIM_BOND_TX_2}" "${TXNS}")
+ BOND_RX1_SLAVE_IF=$(create_netdevsim "${NSIM_BOND_RX_1}" "${RXNS}")
+ BOND_RX2_SLAVE_IF=$(create_netdevsim "${NSIM_BOND_RX_2}" "${RXNS}")
+}
+
+# netdevsim link BOND_TX to BOND_RX interfaces
+function link_ifaces_bond() {
+ local BOND_TX1_SLAVE_IFIDX
+ local BOND_TX2_SLAVE_IFIDX
+ local BOND_RX1_SLAVE_IFIDX
+ local BOND_RX2_SLAVE_IFIDX
+ local TXNS_FD
+ local RXNS_FD
+
+ BOND_TX1_SLAVE_IFIDX=$(ip netns exec "${TXNS}" \
+ cat /sys/class/net/"$BOND_TX1_SLAVE_IF"/ifindex)
+ BOND_TX2_SLAVE_IFIDX=$(ip netns exec "${TXNS}" \
+ cat /sys/class/net/"$BOND_TX2_SLAVE_IF"/ifindex)
+ BOND_RX1_SLAVE_IFIDX=$(ip netns exec "${RXNS}" \
+ cat /sys/class/net/"$BOND_RX1_SLAVE_IF"/ifindex)
+ BOND_RX2_SLAVE_IFIDX=$(ip netns exec "${RXNS}" \
+ cat /sys/class/net/"$BOND_RX2_SLAVE_IF"/ifindex)
+
+ exec {TXNS_FD}</var/run/netns/"${TXNS}"
+ exec {RXNS_FD}</var/run/netns/"${RXNS}"
+
+ # Linking TX ifaces to the RX ones (on the other namespace)
+ echo "${TXNS_FD}:$BOND_TX1_SLAVE_IFIDX $RXNS_FD:$BOND_RX1_SLAVE_IFIDX" \
+ > "$NSIM_DEV_SYS_LINK"
+ echo "${TXNS_FD}:$BOND_TX2_SLAVE_IFIDX $RXNS_FD:$BOND_RX2_SLAVE_IFIDX" \
+ > "$NSIM_DEV_SYS_LINK"
+
+ exec {TXNS_FD}<&-
+ exec {RXNS_FD}<&-
+}
+
+function create_all_ifaces() {
+ # setup_ns function is coming from lib.sh
+ setup_ns TXNS RXNS
+ export NAMESPACE="${RXNS}"
+
+ # Create two interfaces for RX and two for TX
+ create_ifaces_bond
+ # Link netlink ifaces
+ link_ifaces_bond
+}
+
+# configure DSTIF and SRCIF IPs
+function configure_ifaces_ips() {
+ local IP_VERSION=${1:-"ipv4"}
+ select_ipv4_or_ipv6 "${IP_VERSION}"
+
+ ip -n "${RXNS}" addr add "${DSTIP}"/24 dev "${DSTIF}"
+ ip -n "${RXNS}" link set "${DSTIF}" up
+
+ ip -n "${TXNS}" addr add "${SRCIP}"/24 dev "${SRCIF}"
+ ip -n "${TXNS}" link set "${SRCIF}" up
+}
+
+function test_enable_netpoll_on_enslaved_iface() {
+ echo 0 > "${NETCONS_PATH}"/enabled
+
+ # At this stage, BOND_TX1_SLAVE_IF is enslaved to BOND_TX_MAIN_IF, and
+ # linked to BOND_RX1_SLAVE_IF inside the namespace.
+ echo "${BOND_TX1_SLAVE_IF}" > "${NETCONS_PATH}"/dev_name
+
+ # This should fail with the following message in dmesg:
+ # netpoll: netconsole: ethX is a slave device, aborting
+ set +e
+ enable_netcons_ns 2> /dev/null
+ set -e
+
+ if [[ $(cat "${NETCONS_PATH}"/enabled) -eq 1 ]]
+ then
+ echo "test failed: Bonding and netpoll cannot co-exists." >&2
+ exit "${ksft_fail}"
+ fi
+}
+
+function test_delete_bond_and_reenable_target() {
+ ip -n "${TXNS}" \
+ link delete "${BOND_TX_MAIN_IF}" type bond
+
+ # BOND_TX1_SLAVE_IF is not attached to a bond interface anymore
+ # netpoll can be plugged in there
+ echo "${BOND_TX1_SLAVE_IF}" > "${NETCONS_PATH}"/dev_name
+
+ # this should work, since the interface is not enslaved
+ enable_netcons_ns
+
+ if [[ $(cat "${NETCONS_PATH}"/enabled) -eq 0 ]]
+ then
+ echo "test failed: Unable to start netpoll on an unbond iface." >&2
+ exit "${ksft_fail}"
+ fi
+}
+
+# Send a netconsole message to the netconsole target
+function test_send_netcons_msg_through_bond_iface() {
+ # Listen for netconsole port inside the namespace and
+ # destination interface
+ listen_port_and_save_to "${OUTPUT_FILE}" "${IP_VERSION}" &
+ # Wait for socat to start and listen to the port.
+ wait_for_port "${RXNS}" "${PORT}" "${IP_VERSION}"
+ # Send the message
+ echo "${MSG}: ${TARGET}" > /dev/kmsg
+ # Wait until socat saves the file to disk
+ busywait "${BUSYWAIT_TIMEOUT}" test -s "${OUTPUT_FILE}"
+ # Make sure the message was received in the dst part
+ # and exit
+ validate_result "${OUTPUT_FILE}" "${FORMAT}"
+ # kill socat in case it is still running
+ pkill_socat
+}
+
+# BOND_TX1_SLAVE_IF has netconsole enabled on it, bind it to BOND_TX_MAIN_IF.
+# Given BOND_TX_MAIN_IF was deleted, recreate it first
+function test_enslave_netcons_enabled_iface {
+ # netconsole got disabled while the interface was down
+ if [[ $(cat "${NETCONS_PATH}"/enabled) -eq 0 ]]
+ then
+ echo "test failed: netconsole expected to be enabled against BOND_TX1_SLAVE_IF" >&2
+ exit "${ksft_fail}"
+ fi
+
+ # recreate the bonding iface. it got deleted by previous
+ # test (test_delete_bond_and_reenable_target)
+ ip -n "${TXNS}" \
+ link add "${BOND_TX_MAIN_IF}" type bond mode balance-rr
+
+ # sub-interface need to be down before attaching to bonding
+ # This will also disable netconsole.
+ ip -n "${TXNS}" \
+ link set "${BOND_TX1_SLAVE_IF}" down
+ ip -n "${TXNS}" \
+ link set "${BOND_TX1_SLAVE_IF}" master "${BOND_TX_MAIN_IF}"
+ ip -n "${TXNS}" \
+ link set "${BOND_TX_MAIN_IF}" up
+
+ # netconsole got disabled while the interface was down
+ if [[ $(cat "${NETCONS_PATH}"/enabled) -eq 1 ]]
+ then
+ echo "test failed: Device is part of a bond iface, cannot have netcons enabled" >&2
+ exit "${ksft_fail}"
+ fi
+}
+
+# Get netconsole enabled on a bonding interface and attach a second
+# sub-interface.
+function test_enslave_iface_to_bond {
+ # BOND_TX_MAIN_IF has only BOND_TX1_SLAVE_IF right now
+ echo "${BOND_TX_MAIN_IF}" > "${NETCONS_PATH}"/dev_name
+ enable_netcons_ns
+
+ # netcons is attached to bond0 and BOND_TX1_SLAVE_IF is
+ # part of BOND_TX_MAIN_IF. Attach BOND_TX2_SLAVE_IF to BOND_TX_MAIN_IF.
+ ip -n "${TXNS}" \
+ link set "${BOND_TX2_SLAVE_IF}" master "${BOND_TX_MAIN_IF}"
+ if [[ $(cat "${NETCONS_PATH}"/enabled) -eq 0 ]]
+ then
+ echo "test failed: Netconsole should be enabled on bonding interface. Failed" >&2
+ exit "${ksft_fail}"
+ fi
+}
+
+function test_enslave_iff_disabled_netpoll_iface {
+ local ret
+
+ # Create two interfaces. veth interfaces it known to have
+ # IFF_DISABLE_NETPOLL set
+ if ! ip link add "${VETH0}" type veth peer name "${VETH1}"
+ then
+ echo "Failed to create veth TX interface. Is CONFIG_VETH set?" >&2
+ exit "${ksft_skip}"
+ fi
+ set +e
+ # This will print RTNETLINK answers: Device or resource busy
+ ip link set "${VETH0}" master "${BOND_TX_MAIN_IF}" 2> /dev/null
+ ret=$?
+ set -e
+ if [[ $ret -eq 0 ]]
+ then
+ echo "test failed: veth interface could not be enslaved"
+ exit "${ksft_fail}"
+ fi
+}
+
+# Given that netconsole picks the current net namespace, we need to enable it
+# from inside the TXNS namespace
+function enable_netcons_ns() {
+ ip netns exec "${TXNS}" sh -c \
+ "mount -t configfs configfs /sys/kernel/config && echo 1 > $NETCONS_PATH/enabled"
+}
+
+####################
+# Tests start here #
+####################
+
+# Create regular interfaces using netdevsim and link them
+create_all_ifaces
+
+# Setup the bonding interfaces
+# BOND_RX_MAIN_IF has BOND_RX{1,2}_SLAVE_IF
+# BOND_TX_MAIN_IF has BOND_TX{1,2}_SLAVE_IF
+setup_bonding_ifaces
+
+# Configure the ips as BOND_RX1_SLAVE_IF and BOND_TX1_SLAVE_IF
+configure_ifaces_ips "${IP_VERSION}"
+
+_create_dynamic_target "${FORMAT}" "${NETCONS_PATH}"
+enable_netcons_ns
+set_user_data
+
+# Test #1 : Create an bonding interface and attach netpoll into
+# the bonding interface. Netconsole/netpoll should work on
+# the bonding interface.
+test_send_netcons_msg_through_bond_iface
+echo "test #1: netpoll on bonding interface worked. Test passed" >&2
+
+# Test #2: Attach netpoll to an enslaved interface
+# Try to attach netpoll to an enslaved sub-interface (while still being part of
+# a bonding interface), which shouldn't be allowed
+test_enable_netpoll_on_enslaved_iface
+echo "test #2: netpoll correctly rejected enslaved interface (expected behavior). Test passed." >&2
+
+# Test #3: Unplug the sub-interface from bond and enable netconsole
+# Detach the interface from a bonding interface and attach netpoll again
+test_delete_bond_and_reenable_target
+echo "test #3: Able to attach to an unbound interface. Test passed." >&2
+
+# Test #4: Enslave a sub-interface that had netconsole enabled
+# Try to enslave an interface that has netconsole/netpoll enabled.
+# Previous test has netconsole enabled in BOND_TX1_SLAVE_IF, try to enslave it
+test_enslave_netcons_enabled_iface
+echo "test #4: Enslaving an interface with netpoll attached. Test passed." >&2
+
+# Test #5: Enslave a sub-interface to a bonding interface
+# Enslave an interface to a bond interface that has netpoll attached
+# At this stage, BOND_TX_MAIN_IF is created and BOND_TX1_SLAVE_IF is part of
+# it. Netconsole is currently disabled
+test_enslave_iface_to_bond
+echo "test #5: Enslaving an interface to bond+netpoll. Test passed." >&2
+
+# Test #6: Enslave a IFF_DISABLE_NETPOLL sub-interface to a bonding interface
+# At this stage, BOND_TX_MAIN_IF has both sub interface and netconsole is
+# enabled. This test will try to enslave an a veth (IFF_DISABLE_NETPOLL) interface
+# and it should fail, with netpoll: veth0 doesn't support polling
+test_enslave_iff_disabled_netpoll_iface
+echo "test #6: Enslaving IFF_DISABLE_NETPOLL ifaces to bond iface is not supported. Test passed." >&2
+
+cleanup_bond
+trap - EXIT
+exit "${EXIT_STATUS}"
diff --git a/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh b/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh
index 9b5ef8074440c..09553ecd50e39 100644
--- a/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh
+++ b/tools/testing/selftests/drivers/net/lib/sh/lib_netcons.sh
@@ -11,9 +11,11 @@ set -euo pipefail
LIBDIR=$(dirname "$(readlink -e "${BASH_SOURCE[0]}")")
SRCIF="" # to be populated later
+SRCIP="" # to be populated later
SRCIP4="192.0.2.1"
SRCIP6="fc00::1"
DSTIF="" # to be populated later
+DSTIP="" # to be populated later
DSTIP4="192.0.2.2"
DSTIP6="fc00::2"
@@ -28,17 +30,23 @@ NETCONS_PATH="${NETCONS_CONFIGFS}"/"${TARGET}"
# NAMESPACE will be populated by setup_ns with a random value
NAMESPACE=""
-# IDs for netdevsim
+# IDs for netdevsim. We either use NSIM_DEV_{1,2}_ID for standard test
+# or NSIM_BOND_{T,R}X_{1,2} for the bonding tests. Not both at the
+# same time.
NSIM_DEV_1_ID=$((256 + RANDOM % 256))
NSIM_DEV_2_ID=$((512 + RANDOM % 256))
+NSIM_BOND_TX_1=$((768 + RANDOM % 256))
+NSIM_BOND_TX_2=$((1024 + RANDOM % 256))
+NSIM_BOND_RX_1=$((1280 + RANDOM % 256))
+NSIM_BOND_RX_2=$((1536 + RANDOM % 256))
NSIM_DEV_SYS_NEW="/sys/bus/netdevsim/new_device"
+NSIM_DEV_SYS_LINK="/sys/bus/netdevsim/link_device"
# Used to create and delete namespaces
source "${LIBDIR}"/../../../../net/lib.sh
# Create netdevsim interfaces
create_ifaces() {
-
echo "$NSIM_DEV_2_ID" > "$NSIM_DEV_SYS_NEW"
echo "$NSIM_DEV_1_ID" > "$NSIM_DEV_SYS_NEW"
udevadm settle 2> /dev/null || true
@@ -137,15 +145,17 @@ function _create_dynamic_target() {
then
echo 1 > "${NCPATH}"/extended
fi
+}
+function create_and_enable_dynamic_target() {
+ _create_dynamic_target "${FORMAT}" "${NCPATH}"
echo 1 > "${NCPATH}"/enabled
-
}
function create_dynamic_target() {
local FORMAT=${1:-"extended"}
local NCPATH=${2:-"$NETCONS_PATH"}
- _create_dynamic_target "${FORMAT}" "${NCPATH}"
+ create_and_enable_dynamic_target "${FORMAT}" "${NCPATH}"
# This will make sure that the kernel was able to
# load the netconsole driver configuration. The console message
@@ -181,7 +191,6 @@ function disable_release_append() {
function do_cleanup() {
local NSIM_DEV_SYS_DEL="/sys/bus/netdevsim/del_device"
-
# Delete netdevsim devices
echo "$NSIM_DEV_2_ID" > "$NSIM_DEV_SYS_DEL"
echo "$NSIM_DEV_1_ID" > "$NSIM_DEV_SYS_DEL"
@@ -193,14 +202,26 @@ function do_cleanup() {
echo "${DEFAULT_PRINTK_VALUES}" > /proc/sys/kernel/printk
}
-function cleanup() {
+function cleanup_netcons() {
# delete netconsole dynamic reconfiguration
- echo 0 > "${NETCONS_PATH}"/enabled
+ # do not fail if the target is already disabled
+ if [[ ! -d "${NETCONS_PATH}" ]]
+ then
+ # in some cases this is called before netcons path is created
+ return
+ fi
+ if [[ $(cat "${NETCONS_PATH}"/enabled) != 0 ]]
+ then
+ echo 0 > "${NETCONS_PATH}"/enabled || true
+ fi
# Remove all the keys that got created during the selftest
find "${NETCONS_PATH}/userdata/" -mindepth 1 -type d -delete
# Remove the configfs entry
rmdir "${NETCONS_PATH}"
+}
+function cleanup() {
+ cleanup_netcons
do_cleanup
}
@@ -377,3 +398,24 @@ function wait_for_port() {
# more frequently on IPv6
sleep 1
}
+
+# Clean up netdevsim ifaces created for bonding test
+function cleanup_bond_nsim() {
+ ip -n "${TXNS}" \
+ link delete "${BOND_TX_MAIN_IF}" type bond || true
+ ip -n "${RXNS}" \
+ link delete "${BOND_RX_MAIN_IF}" type bond || true
+
+ cleanup_netdevsim "$NSIM_BOND_TX_1"
+ cleanup_netdevsim "$NSIM_BOND_TX_2"
+ cleanup_netdevsim "$NSIM_BOND_RX_1"
+ cleanup_netdevsim "$NSIM_BOND_RX_2"
+}
+
+# cleanup tests that use bonding interfaces
+function cleanup_bond() {
+ cleanup_netcons
+ cleanup_bond_nsim
+ cleanup_all_ns
+ ip link delete "${VETH0}" || true
+}
--
2.47.3
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH net v8 4/4] selftest: netcons: add test for netconsole over bonded interfaces
2025-11-04 17:37 ` [PATCH net v8 4/4] selftest: netcons: add test for netconsole over bonded interfaces Breno Leitao
@ 2025-11-05 16:04 ` Breno Leitao
0 siblings, 0 replies; 6+ messages in thread
From: Breno Leitao @ 2025-11-05 16:04 UTC (permalink / raw)
To: Andrew Lunn, David S. Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Shuah Khan, Simon Horman, david decotigny
Cc: linux-kernel, netdev, linux-kselftest, asantostc, efault, calvin,
kernel-team, jv
On Tue, Nov 04, 2025 at 09:37:04AM -0800, Breno Leitao wrote:
> diff --git a/tools/testing/selftests/drivers/net/bonding/config b/tools/testing/selftests/drivers/net/bonding/config
> index 6bb290abd48bf..57f5a5d5cea67 100644
> --- a/tools/testing/selftests/drivers/net/bonding/config
> +++ b/tools/testing/selftests/drivers/net/bonding/config
> @@ -1,5 +1,6 @@
> CONFIG_BONDING=y
> CONFIG_BRIDGE=y
> +CONFIG_CONFIGFS_FS=y
> CONFIG_DUMMY=y
> CONFIG_INET_ESP=y
> CONFIG_INET_ESP_OFFLOAD=y
> @@ -11,6 +12,9 @@ CONFIG_NET_CLS_FLOWER=y
> CONFIG_NET_CLS_MATCHALL=m
> CONFIG_NETDEVSIM=m
> CONFIG_NET_SCH_INGRESS=y
> +CONFIG_NETCONSOLE=m
> +CONFIG_NETCONSOLE_DYNAMIC=y
> +CONFIG_NETCONSOLE_EXTENDED_LOG=y
I've just realized that check_selftest discards the '_' when checking
for the order. This means NETCONSOLE should come before
CONFIG_NET_SCH_INGRESS.
I will wait for additional review, before updating.
^ permalink raw reply [flat|nested] 6+ messages in thread