[PATCH net-next v2 0/3] net: replace wq users and add WQ_PERCPU to alloc_workqueue() users

[PATCH net-next v2 0/3] net: replace wq users and add WQ_PERCPU to alloc_workqueue() users

[Marco Crivellari]

Hi!

Below is a summary of a discussion about the Workqueue API and cpu isolation
considerations. Details and more information are available here:

        "workqueue: Always use wq_select_unbound_cpu() for WORK_CPU_UNBOUND."
        https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/

=== Current situation: problems ===

Let's consider a nohz_full system with isolated CPUs: wq_unbound_cpumask is
set to the housekeeping CPUs, for !WQ_UNBOUND the local CPU is selected.

This leads to different scenarios if a work item is scheduled on an isolated
CPU where "delay" value is 0 or greater then 0:
        schedule_delayed_work(, 0);

This will be handled by __queue_work() that will queue the work item on the
current local (isolated) CPU, while:

        schedule_delayed_work(, 1);

Will move the timer on an housekeeping CPU, and schedule the work there.

Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.

This lack of consistentcy cannot be addressed without refactoring the API.

=== Plan and future plans ===

This patchset is the first stone on a refactoring needed in order to
address the points aforementioned; it will have a positive impact also
on the cpu isolation, in the long term, moving away percpu workqueue in
favor to an unbound model.

These are the main steps:
1)  API refactoring (that this patch is introducing)
    -   Make more clear and uniform the system wq names, both per-cpu and
        unbound. This to avoid any possible confusion on what should be
        used.

    -   Introduction of WQ_PERCPU: this flag is the complement of WQ_UNBOUND,
        introduced in this patchset and used on all the callers that are not
        currently using WQ_UNBOUND.

        WQ_UNBOUND will be removed in a future release cycle.

        Most users don't need to be per-cpu, because they don't have
        locality requirements, because of that, a next future step will be
        make "unbound" the default behavior.

2)  Check who really needs to be per-cpu
    -   Remove the WQ_PERCPU flag when is not strictly required.

3)  Add a new API (prefer local cpu)
    -   There are users that don't require a local execution, like mentioned
        above; despite that, local execution yeld to performance gain.

        This new API will prefer the local execution, without requiring it.

=== Introduced Changes by this series ===

1) [P 1-2] Replace use of system_wq and system_unbound_wq

        system_wq is a per-CPU workqueue, but his name is not clear.
        system_unbound_wq is to be used when locality is not required.

        Because of that, system_wq has been renamed in system_percpu_wq, and
        system_unbound_wq has been renamed in system_dfl_wq.

2) [P 3] add WQ_PERCPU to remaining alloc_workqueue() users

        Every alloc_workqueue() caller should use one among WQ_PERCPU or
        WQ_UNBOUND.

        WQ_UNBOUND will be removed in a next release cycle.

Thanks!


---
Changes in v2:
- wireless removed from this series
- work rebased on latest net-next changes


Marco Crivellari (3):
  net: replace use of system_unbound_wq with system_dfl_wq
  net: replace use of system_wq with system_percpu_wq
  net: WQ_PERCPU added to alloc_workqueue users

 drivers/net/can/spi/hi311x.c                         |  3 ++-
 drivers/net/can/spi/mcp251x.c                        |  3 ++-
 drivers/net/ethernet/cavium/liquidio/lio_core.c      |  2 +-
 drivers/net/ethernet/cavium/liquidio/lio_main.c      |  8 +++++---
 drivers/net/ethernet/cavium/liquidio/lio_vf_main.c   |  3 ++-
 .../net/ethernet/cavium/liquidio/request_manager.c   |  2 +-
 .../net/ethernet/cavium/liquidio/response_manager.c  |  3 ++-
 drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c     |  2 +-
 .../net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c  |  3 ++-
 drivers/net/ethernet/intel/fm10k/fm10k_main.c        |  2 +-
 drivers/net/ethernet/intel/i40e/i40e_main.c          |  2 +-
 drivers/net/ethernet/marvell/octeontx2/af/cgx.c      |  2 +-
 .../net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c   |  2 +-
 drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c  |  2 +-
 drivers/net/ethernet/marvell/octeontx2/af/rvu_rep.c  |  2 +-
 .../net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c |  3 ++-
 .../net/ethernet/marvell/prestera/prestera_main.c    |  2 +-
 drivers/net/ethernet/marvell/prestera/prestera_pci.c |  2 +-
 drivers/net/ethernet/mellanox/mlxsw/core.c           |  4 ++--
 drivers/net/ethernet/netronome/nfp/nfp_main.c        |  2 +-
 drivers/net/ethernet/qlogic/qed/qed_main.c           |  3 ++-
 drivers/net/ethernet/sfc/efx_channels.c              |  2 +-
 drivers/net/ethernet/sfc/siena/efx_channels.c        |  2 +-
 drivers/net/ethernet/wiznet/w5100.c                  |  2 +-
 drivers/net/fjes/fjes_main.c                         |  5 +++--
 drivers/net/macvlan.c                                |  2 +-
 drivers/net/netdevsim/dev.c                          |  6 +++---
 drivers/net/phy/sfp.c                                | 12 ++++++------
 drivers/net/wireguard/device.c                       |  6 ++++--
 drivers/net/wwan/t7xx/t7xx_hif_dpmaif_rx.c           |  3 ++-
 drivers/net/wwan/wwan_hwsim.c                        |  2 +-
 net/bridge/br_cfm.c                                  |  6 +++---
 net/bridge/br_mrp.c                                  |  8 ++++----
 net/ceph/messenger.c                                 |  3 ++-
 net/ceph/mon_client.c                                |  2 +-
 net/core/link_watch.c                                |  4 ++--
 net/core/skmsg.c                                     |  2 +-
 net/core/sock_diag.c                                 |  2 +-
 net/devlink/core.c                                   |  2 +-
 net/ipv4/inet_fragment.c                             |  2 +-
 net/netfilter/nf_conntrack_ecache.c                  |  2 +-
 net/openvswitch/dp_notify.c                          |  2 +-
 net/rds/ib_rdma.c                                    |  3 ++-
 net/rfkill/input.c                                   |  2 +-
 net/rxrpc/rxperf.c                                   |  2 +-
 net/smc/af_smc.c                                     |  6 +++---
 net/smc/smc_core.c                                   |  4 ++--
 net/tls/tls_device.c                                 |  2 +-
 net/unix/garbage.c                                   |  2 +-
 net/vmw_vsock/af_vsock.c                             |  2 +-
 net/vmw_vsock/virtio_transport.c                     |  2 +-
 net/vmw_vsock/vsock_loopback.c                       |  2 +-
 52 files changed, 88 insertions(+), 73 deletions(-)

-- 
2.51.0

[PATCH net-next v2 1/3] net: replace use of system_unbound_wq with system_dfl_wq

[Marco Crivellari]

Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.

This lack of consistentcy cannot be addressed without refactoring the API.

system_unbound_wq should be the default workqueue so as not to enforce
locality constraints for random work whenever it's not required.

Adding system_dfl_wq to encourage its use when unbound work should be used.

The old system_unbound_wq will be kept for a few release cycles.

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/net/macvlan.c       | 2 +-
 drivers/net/netdevsim/dev.c | 6 +++---
 net/core/link_watch.c       | 4 ++--
 net/unix/garbage.c          | 2 +-
 4 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/net/macvlan.c b/drivers/net/macvlan.c
index 4df991e494bd..7966545512cf 100644
--- a/drivers/net/macvlan.c
+++ b/drivers/net/macvlan.c
@@ -369,7 +369,7 @@ static void macvlan_broadcast_enqueue(struct macvlan_port *port,
 	}
 	spin_unlock(&port->bc_queue.lock);
 
-	queue_work(system_unbound_wq, &port->bc_work);
+	queue_work(system_dfl_wq, &port->bc_work);
 
 	if (err)
 		goto free_nskb;
diff --git a/drivers/net/netdevsim/dev.c b/drivers/net/netdevsim/dev.c
index 2672d071b325..95f66c1f59db 100644
--- a/drivers/net/netdevsim/dev.c
+++ b/drivers/net/netdevsim/dev.c
@@ -851,7 +851,7 @@ static void nsim_dev_trap_report_work(struct work_struct *work)
 	nsim_dev = nsim_trap_data->nsim_dev;
 
 	if (!devl_trylock(priv_to_devlink(nsim_dev))) {
-		queue_delayed_work(system_unbound_wq,
+		queue_delayed_work(system_dfl_wq,
 				   &nsim_dev->trap_data->trap_report_dw, 1);
 		return;
 	}
@@ -867,7 +867,7 @@ static void nsim_dev_trap_report_work(struct work_struct *work)
 		cond_resched();
 	}
 	devl_unlock(priv_to_devlink(nsim_dev));
-	queue_delayed_work(system_unbound_wq,
+	queue_delayed_work(system_dfl_wq,
 			   &nsim_dev->trap_data->trap_report_dw,
 			   msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
 }
@@ -924,7 +924,7 @@ static int nsim_dev_traps_init(struct devlink *devlink)
 
 	INIT_DELAYED_WORK(&nsim_dev->trap_data->trap_report_dw,
 			  nsim_dev_trap_report_work);
-	queue_delayed_work(system_unbound_wq,
+	queue_delayed_work(system_dfl_wq,
 			   &nsim_dev->trap_data->trap_report_dw,
 			   msecs_to_jiffies(NSIM_TRAP_REPORT_INTERVAL_MS));
 
diff --git a/net/core/link_watch.c b/net/core/link_watch.c
index 864f3bbc3a4c..212cde35affa 100644
--- a/net/core/link_watch.c
+++ b/net/core/link_watch.c
@@ -157,9 +157,9 @@ static void linkwatch_schedule_work(int urgent)
 	 * override the existing timer.
 	 */
 	if (test_bit(LW_URGENT, &linkwatch_flags))
-		mod_delayed_work(system_unbound_wq, &linkwatch_work, 0);
+		mod_delayed_work(system_dfl_wq, &linkwatch_work, 0);
 	else
-		queue_delayed_work(system_unbound_wq, &linkwatch_work, delay);
+		queue_delayed_work(system_dfl_wq, &linkwatch_work, delay);
 }
 
 
diff --git a/net/unix/garbage.c b/net/unix/garbage.c
index 01e2b9452c75..684ab03137b6 100644
--- a/net/unix/garbage.c
+++ b/net/unix/garbage.c
@@ -592,7 +592,7 @@ static DECLARE_WORK(unix_gc_work, __unix_gc);
 void unix_gc(void)
 {
 	WRITE_ONCE(gc_in_progress, true);
-	queue_work(system_unbound_wq, &unix_gc_work);
+	queue_work(system_dfl_wq, &unix_gc_work);
 }
 
 #define UNIX_INFLIGHT_TRIGGER_GC 16000
-- 
2.51.0

[PATCH net-next v2 2/3] net: replace use of system_wq with system_percpu_wq

[Marco Crivellari]

Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.

This lack of consistentcy cannot be addressed without refactoring the API.

system_unbound_wq should be the default workqueue so as not to enforce
locality constraints for random work whenever it's not required.

Adding system_dfl_wq to encourage its use when unbound work should be used.

The old system_unbound_wq will be kept for a few release cycles.

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/net/ethernet/sfc/efx_channels.c       |  2 +-
 drivers/net/ethernet/sfc/siena/efx_channels.c |  2 +-
 drivers/net/phy/sfp.c                         | 12 ++++++------
 net/bridge/br_cfm.c                           |  6 +++---
 net/bridge/br_mrp.c                           |  8 ++++----
 net/ceph/mon_client.c                         |  2 +-
 net/core/skmsg.c                              |  2 +-
 net/devlink/core.c                            |  2 +-
 net/ipv4/inet_fragment.c                      |  2 +-
 net/netfilter/nf_conntrack_ecache.c           |  2 +-
 net/openvswitch/dp_notify.c                   |  2 +-
 net/rfkill/input.c                            |  2 +-
 net/smc/smc_core.c                            |  2 +-
 net/vmw_vsock/af_vsock.c                      |  2 +-
 14 files changed, 24 insertions(+), 24 deletions(-)

diff --git a/drivers/net/ethernet/sfc/efx_channels.c b/drivers/net/ethernet/sfc/efx_channels.c
index 0f66324ed351..ed3a96ebc7f3 100644
--- a/drivers/net/ethernet/sfc/efx_channels.c
+++ b/drivers/net/ethernet/sfc/efx_channels.c
@@ -1281,7 +1281,7 @@ static int efx_poll(struct napi_struct *napi, int budget)
 		time = jiffies - channel->rfs_last_expiry;
 		/* Would our quota be >= 20? */
 		if (channel->rfs_filter_count * time >= 600 * HZ)
-			mod_delayed_work(system_wq, &channel->filter_work, 0);
+			mod_delayed_work(system_percpu_wq, &channel->filter_work, 0);
 #endif
 
 		/* There is no race here; although napi_disable() will
diff --git a/drivers/net/ethernet/sfc/siena/efx_channels.c b/drivers/net/ethernet/sfc/siena/efx_channels.c
index 703419866d18..fc075ab6b7b5 100644
--- a/drivers/net/ethernet/sfc/siena/efx_channels.c
+++ b/drivers/net/ethernet/sfc/siena/efx_channels.c
@@ -1300,7 +1300,7 @@ static int efx_poll(struct napi_struct *napi, int budget)
 		time = jiffies - channel->rfs_last_expiry;
 		/* Would our quota be >= 20? */
 		if (channel->rfs_filter_count * time >= 600 * HZ)
-			mod_delayed_work(system_wq, &channel->filter_work, 0);
+			mod_delayed_work(system_percpu_wq, &channel->filter_work, 0);
 #endif
 
 		/* There is no race here; although napi_disable() will
diff --git a/drivers/net/phy/sfp.c b/drivers/net/phy/sfp.c
index 4cd1d6c51dc2..eaccd92253a1 100644
--- a/drivers/net/phy/sfp.c
+++ b/drivers/net/phy/sfp.c
@@ -912,7 +912,7 @@ static void sfp_soft_start_poll(struct sfp *sfp)
 
 	if (sfp->state_soft_mask & (SFP_F_LOS | SFP_F_TX_FAULT) &&
 	    !sfp->need_poll)
-		mod_delayed_work(system_wq, &sfp->poll, poll_jiffies);
+		mod_delayed_work(system_percpu_wq, &sfp->poll, poll_jiffies);
 	mutex_unlock(&sfp->st_mutex);
 }
 
@@ -1683,7 +1683,7 @@ static void sfp_hwmon_probe(struct work_struct *work)
 	err = sfp_read(sfp, true, 0, &sfp->diag, sizeof(sfp->diag));
 	if (err < 0) {
 		if (sfp->hwmon_tries--) {
-			mod_delayed_work(system_wq, &sfp->hwmon_probe,
+			mod_delayed_work(system_percpu_wq, &sfp->hwmon_probe,
 					 T_PROBE_RETRY_SLOW);
 		} else {
 			dev_warn(sfp->dev, "hwmon probe failed: %pe\n",
@@ -1710,7 +1710,7 @@ static void sfp_hwmon_probe(struct work_struct *work)
 static int sfp_hwmon_insert(struct sfp *sfp)
 {
 	if (sfp->have_a2 && sfp->id.ext.diagmon & SFP_DIAGMON_DDM) {
-		mod_delayed_work(system_wq, &sfp->hwmon_probe, 1);
+		mod_delayed_work(system_percpu_wq, &sfp->hwmon_probe, 1);
 		sfp->hwmon_tries = R_PROBE_RETRY_SLOW;
 	}
 
@@ -2564,7 +2564,7 @@ static void sfp_sm_module(struct sfp *sfp, unsigned int event)
 		/* Force a poll to re-read the hardware signal state after
 		 * sfp_sm_mod_probe() changed state_hw_mask.
 		 */
-		mod_delayed_work(system_wq, &sfp->poll, 1);
+		mod_delayed_work(system_percpu_wq, &sfp->poll, 1);
 
 		err = sfp_hwmon_insert(sfp);
 		if (err)
@@ -3009,7 +3009,7 @@ static void sfp_poll(struct work_struct *work)
 	// it's unimportant if we race while reading this.
 	if (sfp->state_soft_mask & (SFP_F_LOS | SFP_F_TX_FAULT) ||
 	    sfp->need_poll)
-		mod_delayed_work(system_wq, &sfp->poll, poll_jiffies);
+		mod_delayed_work(system_percpu_wq, &sfp->poll, poll_jiffies);
 }
 
 static struct sfp *sfp_alloc(struct device *dev)
@@ -3179,7 +3179,7 @@ static int sfp_probe(struct platform_device *pdev)
 	}
 
 	if (sfp->need_poll)
-		mod_delayed_work(system_wq, &sfp->poll, poll_jiffies);
+		mod_delayed_work(system_percpu_wq, &sfp->poll, poll_jiffies);
 
 	/* We could have an issue in cases no Tx disable pin is available or
 	 * wired as modules using a laser as their light source will continue to
diff --git a/net/bridge/br_cfm.c b/net/bridge/br_cfm.c
index a3c755d0a09d..c2c1c7d44c61 100644
--- a/net/bridge/br_cfm.c
+++ b/net/bridge/br_cfm.c
@@ -134,7 +134,7 @@ static void ccm_rx_timer_start(struct br_cfm_peer_mep *peer_mep)
 	 * of the configured CC 'expected_interval'
 	 * in order to detect CCM defect after 3.25 interval.
 	 */
-	queue_delayed_work(system_wq, &peer_mep->ccm_rx_dwork,
+	queue_delayed_work(system_percpu_wq, &peer_mep->ccm_rx_dwork,
 			   usecs_to_jiffies(interval_us / 4));
 }
 
@@ -285,7 +285,7 @@ static void ccm_tx_work_expired(struct work_struct *work)
 		ccm_frame_tx(skb);
 
 	interval_us = interval_to_us(mep->cc_config.exp_interval);
-	queue_delayed_work(system_wq, &mep->ccm_tx_dwork,
+	queue_delayed_work(system_percpu_wq, &mep->ccm_tx_dwork,
 			   usecs_to_jiffies(interval_us));
 }
 
@@ -809,7 +809,7 @@ int br_cfm_cc_ccm_tx(struct net_bridge *br, const u32 instance,
 	 * to send first frame immediately
 	 */
 	mep->ccm_tx_end = jiffies + usecs_to_jiffies(tx_info->period * 1000000);
-	queue_delayed_work(system_wq, &mep->ccm_tx_dwork, 0);
+	queue_delayed_work(system_percpu_wq, &mep->ccm_tx_dwork, 0);
 
 save:
 	mep->cc_ccm_tx_info = *tx_info;
diff --git a/net/bridge/br_mrp.c b/net/bridge/br_mrp.c
index fd2de35ffb3c..3c36fa24bc05 100644
--- a/net/bridge/br_mrp.c
+++ b/net/bridge/br_mrp.c
@@ -341,7 +341,7 @@ static void br_mrp_test_work_expired(struct work_struct *work)
 out:
 	rcu_read_unlock();
 
-	queue_delayed_work(system_wq, &mrp->test_work,
+	queue_delayed_work(system_percpu_wq, &mrp->test_work,
 			   usecs_to_jiffies(mrp->test_interval));
 }
 
@@ -418,7 +418,7 @@ static void br_mrp_in_test_work_expired(struct work_struct *work)
 out:
 	rcu_read_unlock();
 
-	queue_delayed_work(system_wq, &mrp->in_test_work,
+	queue_delayed_work(system_percpu_wq, &mrp->in_test_work,
 			   usecs_to_jiffies(mrp->in_test_interval));
 }
 
@@ -725,7 +725,7 @@ int br_mrp_start_test(struct net_bridge *br,
 	mrp->test_max_miss = test->max_miss;
 	mrp->test_monitor = test->monitor;
 	mrp->test_count_miss = 0;
-	queue_delayed_work(system_wq, &mrp->test_work,
+	queue_delayed_work(system_percpu_wq, &mrp->test_work,
 			   usecs_to_jiffies(test->interval));
 
 	return 0;
@@ -865,7 +865,7 @@ int br_mrp_start_in_test(struct net_bridge *br,
 	mrp->in_test_end = jiffies + usecs_to_jiffies(in_test->period);
 	mrp->in_test_max_miss = in_test->max_miss;
 	mrp->in_test_count_miss = 0;
-	queue_delayed_work(system_wq, &mrp->in_test_work,
+	queue_delayed_work(system_percpu_wq, &mrp->in_test_work,
 			   usecs_to_jiffies(in_test->interval));
 
 	return 0;
diff --git a/net/ceph/mon_client.c b/net/ceph/mon_client.c
index ab66b599ac47..c227ececa925 100644
--- a/net/ceph/mon_client.c
+++ b/net/ceph/mon_client.c
@@ -314,7 +314,7 @@ static void __schedule_delayed(struct ceph_mon_client *monc)
 		delay = CEPH_MONC_PING_INTERVAL;
 
 	dout("__schedule_delayed after %lu\n", delay);
-	mod_delayed_work(system_wq, &monc->delayed_work,
+	mod_delayed_work(system_percpu_wq, &monc->delayed_work,
 			 round_jiffies_relative(delay));
 }
 
diff --git a/net/core/skmsg.c b/net/core/skmsg.c
index 83c78379932e..2ac7731e1e0a 100644
--- a/net/core/skmsg.c
+++ b/net/core/skmsg.c
@@ -876,7 +876,7 @@ void sk_psock_drop(struct sock *sk, struct sk_psock *psock)
 	sk_psock_stop(psock);
 
 	INIT_RCU_WORK(&psock->rwork, sk_psock_destroy);
-	queue_rcu_work(system_wq, &psock->rwork);
+	queue_rcu_work(system_percpu_wq, &psock->rwork);
 }
 EXPORT_SYMBOL_GPL(sk_psock_drop);
 
diff --git a/net/devlink/core.c b/net/devlink/core.c
index 7203c39532fc..58093f49c090 100644
--- a/net/devlink/core.c
+++ b/net/devlink/core.c
@@ -320,7 +320,7 @@ static void devlink_release(struct work_struct *work)
 void devlink_put(struct devlink *devlink)
 {
 	if (refcount_dec_and_test(&devlink->refcount))
-		queue_rcu_work(system_wq, &devlink->rwork);
+		queue_rcu_work(system_percpu_wq, &devlink->rwork);
 }
 
 struct devlink *devlinks_xa_find_get(struct net *net, unsigned long *indexp)
diff --git a/net/ipv4/inet_fragment.c b/net/ipv4/inet_fragment.c
index 470ab17ceb51..025895eb6ec5 100644
--- a/net/ipv4/inet_fragment.c
+++ b/net/ipv4/inet_fragment.c
@@ -183,7 +183,7 @@ static void fqdir_work_fn(struct work_struct *work)
 	rhashtable_free_and_destroy(&fqdir->rhashtable, inet_frags_free_cb, NULL);
 
 	if (llist_add(&fqdir->free_list, &fqdir_free_list))
-		queue_delayed_work(system_wq, &fqdir_free_work, HZ);
+		queue_delayed_work(system_percpu_wq, &fqdir_free_work, HZ);
 }
 
 int fqdir_init(struct fqdir **fqdirp, struct inet_frags *f, struct net *net)
diff --git a/net/netfilter/nf_conntrack_ecache.c b/net/netfilter/nf_conntrack_ecache.c
index af68c64acaab..81baf2082604 100644
--- a/net/netfilter/nf_conntrack_ecache.c
+++ b/net/netfilter/nf_conntrack_ecache.c
@@ -301,7 +301,7 @@ void nf_conntrack_ecache_work(struct net *net, enum nf_ct_ecache_state state)
 		net->ct.ecache_dwork_pending = true;
 	} else if (state == NFCT_ECACHE_DESTROY_SENT) {
 		if (!hlist_nulls_empty(&cnet->ecache.dying_list))
-			mod_delayed_work(system_wq, &cnet->ecache.dwork, 0);
+			mod_delayed_work(system_percpu_wq, &cnet->ecache.dwork, 0);
 		else
 			net->ct.ecache_dwork_pending = false;
 	}
diff --git a/net/openvswitch/dp_notify.c b/net/openvswitch/dp_notify.c
index 7af0cde8b293..a2af90ee99af 100644
--- a/net/openvswitch/dp_notify.c
+++ b/net/openvswitch/dp_notify.c
@@ -75,7 +75,7 @@ static int dp_device_event(struct notifier_block *unused, unsigned long event,
 
 		/* schedule vport destroy, dev_put and genl notification */
 		ovs_net = net_generic(dev_net(dev), ovs_net_id);
-		queue_work(system_wq, &ovs_net->dp_notify_work);
+		queue_work(system_percpu_wq, &ovs_net->dp_notify_work);
 	}
 
 	return NOTIFY_DONE;
diff --git a/net/rfkill/input.c b/net/rfkill/input.c
index 598d0a61bda7..53d286b10843 100644
--- a/net/rfkill/input.c
+++ b/net/rfkill/input.c
@@ -159,7 +159,7 @@ static void rfkill_schedule_global_op(enum rfkill_sched_op op)
 	rfkill_op_pending = true;
 	if (op == RFKILL_GLOBAL_OP_EPO && !rfkill_is_epo_lock_active()) {
 		/* bypass the limiter for EPO */
-		mod_delayed_work(system_wq, &rfkill_op_work, 0);
+		mod_delayed_work(system_percpu_wq, &rfkill_op_work, 0);
 		rfkill_last_scheduled = jiffies;
 	} else
 		rfkill_schedule_ratelimited();
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index 2a559a98541c..e216d237865b 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -85,7 +85,7 @@ static void smc_lgr_schedule_free_work(struct smc_link_group *lgr)
 	 * otherwise there is a risk of out-of-sync link groups.
 	 */
 	if (!lgr->freeing) {
-		mod_delayed_work(system_wq, &lgr->free_work,
+		mod_delayed_work(system_percpu_wq, &lgr->free_work,
 				 (!lgr->is_smcd && lgr->role == SMC_CLNT) ?
 						SMC_LGR_FREE_DELAY_CLNT :
 						SMC_LGR_FREE_DELAY_SERV);
diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index 0538948d5fd9..4c2db6cca557 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c
@@ -1649,7 +1649,7 @@ static int vsock_connect(struct socket *sock, struct sockaddr *addr,
 			 * reschedule it, then ungrab the socket refcount to
 			 * keep it balanced.
 			 */
-			if (mod_delayed_work(system_wq, &vsk->connect_work,
+			if (mod_delayed_work(system_percpu_wq, &vsk->connect_work,
 					     timeout))
 				sock_put(sk);
 
-- 
2.51.0

[PATCH net-next v2 3/3] net: WQ_PERCPU added to alloc_workqueue users

[Marco Crivellari]

Currently if a user enqueue a work item using schedule_delayed_work() the
used wq is "system_wq" (per-cpu wq) while queue_delayed_work() use
WORK_CPU_UNBOUND (used when a cpu is not specified). The same applies to
schedule_work() that is using system_wq and queue_work(), that makes use
again of WORK_CPU_UNBOUND.
This lack of consistentcy cannot be addressed without refactoring the API.

alloc_workqueue() treats all queues as per-CPU by default, while unbound
workqueues must opt-in via WQ_UNBOUND.

This default is suboptimal: most workloads benefit from unbound queues,
allowing the scheduler to place worker threads where they’re needed and
reducing noise when CPUs are isolated.

This change adds a new WQ_PERCPU flag at the network subsystem, to explicitly
request the use of the per-CPU behavior. Both flags coexist for one release
cycle to allow callers to transition their calls.

Once migration is complete, WQ_UNBOUND can be removed and unbound will
become the implicit default.

With the introduction of the WQ_PERCPU flag (equivalent to !WQ_UNBOUND),
any alloc_workqueue() caller that doesn’t explicitly specify WQ_UNBOUND
must now use WQ_PERCPU.

All existing users have been updated accordingly.

Suggested-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Marco Crivellari <marco.crivellari@suse.com>
---
 drivers/net/can/spi/hi311x.c                             | 3 ++-
 drivers/net/can/spi/mcp251x.c                            | 3 ++-
 drivers/net/ethernet/cavium/liquidio/lio_core.c          | 2 +-
 drivers/net/ethernet/cavium/liquidio/lio_main.c          | 8 +++++---
 drivers/net/ethernet/cavium/liquidio/lio_vf_main.c       | 3 ++-
 drivers/net/ethernet/cavium/liquidio/request_manager.c   | 2 +-
 drivers/net/ethernet/cavium/liquidio/response_manager.c  | 3 ++-
 drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c         | 2 +-
 drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c  | 3 ++-
 drivers/net/ethernet/intel/fm10k/fm10k_main.c            | 2 +-
 drivers/net/ethernet/intel/i40e/i40e_main.c              | 2 +-
 drivers/net/ethernet/marvell/octeontx2/af/cgx.c          | 2 +-
 drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c   | 2 +-
 drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c      | 2 +-
 drivers/net/ethernet/marvell/octeontx2/af/rvu_rep.c      | 2 +-
 drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c | 3 ++-
 drivers/net/ethernet/marvell/prestera/prestera_main.c    | 2 +-
 drivers/net/ethernet/marvell/prestera/prestera_pci.c     | 2 +-
 drivers/net/ethernet/mellanox/mlxsw/core.c               | 4 ++--
 drivers/net/ethernet/netronome/nfp/nfp_main.c            | 2 +-
 drivers/net/ethernet/qlogic/qed/qed_main.c               | 3 ++-
 drivers/net/ethernet/wiznet/w5100.c                      | 2 +-
 drivers/net/fjes/fjes_main.c                             | 5 +++--
 drivers/net/wireguard/device.c                           | 6 ++++--
 drivers/net/wwan/t7xx/t7xx_hif_dpmaif_rx.c               | 3 ++-
 drivers/net/wwan/wwan_hwsim.c                            | 2 +-
 net/ceph/messenger.c                                     | 3 ++-
 net/core/sock_diag.c                                     | 2 +-
 net/rds/ib_rdma.c                                        | 3 ++-
 net/rxrpc/rxperf.c                                       | 2 +-
 net/smc/af_smc.c                                         | 6 +++---
 net/smc/smc_core.c                                       | 2 +-
 net/tls/tls_device.c                                     | 2 +-
 net/vmw_vsock/virtio_transport.c                         | 2 +-
 net/vmw_vsock/vsock_loopback.c                           | 2 +-
 35 files changed, 57 insertions(+), 42 deletions(-)

diff --git a/drivers/net/can/spi/hi311x.c b/drivers/net/can/spi/hi311x.c
index 09ae218315d7..96f23311b4ee 100644
--- a/drivers/net/can/spi/hi311x.c
+++ b/drivers/net/can/spi/hi311x.c
@@ -770,7 +770,8 @@ static int hi3110_open(struct net_device *net)
 		goto out_close;
 	}
 
-	priv->wq = alloc_workqueue("hi3110_wq", WQ_FREEZABLE | WQ_MEM_RECLAIM,
+	priv->wq = alloc_workqueue("hi3110_wq",
+				   WQ_FREEZABLE | WQ_MEM_RECLAIM | WQ_PERCPU,
 				   0);
 	if (!priv->wq) {
 		ret = -ENOMEM;
diff --git a/drivers/net/can/spi/mcp251x.c b/drivers/net/can/spi/mcp251x.c
index 313e1d241f01..b797e08499d7 100644
--- a/drivers/net/can/spi/mcp251x.c
+++ b/drivers/net/can/spi/mcp251x.c
@@ -1378,7 +1378,8 @@ static int mcp251x_can_probe(struct spi_device *spi)
 	if (ret)
 		goto out_clk;
 
-	priv->wq = alloc_workqueue("mcp251x_wq", WQ_FREEZABLE | WQ_MEM_RECLAIM,
+	priv->wq = alloc_workqueue("mcp251x_wq",
+				   WQ_FREEZABLE | WQ_MEM_RECLAIM | WQ_PERCPU,
 				   0);
 	if (!priv->wq) {
 		ret = -ENOMEM;
diff --git a/drivers/net/ethernet/cavium/liquidio/lio_core.c b/drivers/net/ethernet/cavium/liquidio/lio_core.c
index 674c54831875..215dac201b4a 100644
--- a/drivers/net/ethernet/cavium/liquidio/lio_core.c
+++ b/drivers/net/ethernet/cavium/liquidio/lio_core.c
@@ -472,7 +472,7 @@ int setup_rx_oom_poll_fn(struct net_device *netdev)
 		q_no = lio->linfo.rxpciq[q].s.q_no;
 		wq = &lio->rxq_status_wq[q_no];
 		wq->wq = alloc_workqueue("rxq-oom-status",
-					 WQ_MEM_RECLAIM, 0);
+					 WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 		if (!wq->wq) {
 			dev_err(&oct->pci_dev->dev, "unable to create cavium rxq oom status wq\n");
 			return -ENOMEM;
diff --git a/drivers/net/ethernet/cavium/liquidio/lio_main.c b/drivers/net/ethernet/cavium/liquidio/lio_main.c
index 1d79f6eaa41f..8e2fcec26ea1 100644
--- a/drivers/net/ethernet/cavium/liquidio/lio_main.c
+++ b/drivers/net/ethernet/cavium/liquidio/lio_main.c
@@ -526,7 +526,8 @@ static inline int setup_link_status_change_wq(struct net_device *netdev)
 	struct octeon_device *oct = lio->oct_dev;
 
 	lio->link_status_wq.wq = alloc_workqueue("link-status",
-						 WQ_MEM_RECLAIM, 0);
+						 WQ_MEM_RECLAIM | WQ_PERCPU,
+						 0);
 	if (!lio->link_status_wq.wq) {
 		dev_err(&oct->pci_dev->dev, "unable to create cavium link status wq\n");
 		return -1;
@@ -659,7 +660,8 @@ static inline int setup_sync_octeon_time_wq(struct net_device *netdev)
 	struct octeon_device *oct = lio->oct_dev;
 
 	lio->sync_octeon_time_wq.wq =
-		alloc_workqueue("update-octeon-time", WQ_MEM_RECLAIM, 0);
+		alloc_workqueue("update-octeon-time",
+				WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (!lio->sync_octeon_time_wq.wq) {
 		dev_err(&oct->pci_dev->dev, "Unable to create wq to update octeon time\n");
 		return -1;
@@ -1734,7 +1736,7 @@ static inline int setup_tx_poll_fn(struct net_device *netdev)
 	struct octeon_device *oct = lio->oct_dev;
 
 	lio->txq_status_wq.wq = alloc_workqueue("txq-status",
-						WQ_MEM_RECLAIM, 0);
+						WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (!lio->txq_status_wq.wq) {
 		dev_err(&oct->pci_dev->dev, "unable to create cavium txq status wq\n");
 		return -1;
diff --git a/drivers/net/ethernet/cavium/liquidio/lio_vf_main.c b/drivers/net/ethernet/cavium/liquidio/lio_vf_main.c
index 62c2eadc33e3..3230dff5ba05 100644
--- a/drivers/net/ethernet/cavium/liquidio/lio_vf_main.c
+++ b/drivers/net/ethernet/cavium/liquidio/lio_vf_main.c
@@ -304,7 +304,8 @@ static int setup_link_status_change_wq(struct net_device *netdev)
 	struct octeon_device *oct = lio->oct_dev;
 
 	lio->link_status_wq.wq = alloc_workqueue("link-status",
-						 WQ_MEM_RECLAIM, 0);
+						 WQ_MEM_RECLAIM | WQ_PERCPU,
+						 0);
 	if (!lio->link_status_wq.wq) {
 		dev_err(&oct->pci_dev->dev, "unable to create cavium link status wq\n");
 		return -1;
diff --git a/drivers/net/ethernet/cavium/liquidio/request_manager.c b/drivers/net/ethernet/cavium/liquidio/request_manager.c
index de8a6ce86ad7..8b8e9953c4ee 100644
--- a/drivers/net/ethernet/cavium/liquidio/request_manager.c
+++ b/drivers/net/ethernet/cavium/liquidio/request_manager.c
@@ -132,7 +132,7 @@ int octeon_init_instr_queue(struct octeon_device *oct,
 	oct->fn_list.setup_iq_regs(oct, iq_no);
 
 	oct->check_db_wq[iq_no].wq = alloc_workqueue("check_iq_db",
-						     WQ_MEM_RECLAIM,
+						     WQ_MEM_RECLAIM | WQ_PERCPU,
 						     0);
 	if (!oct->check_db_wq[iq_no].wq) {
 		vfree(iq->request_list);
diff --git a/drivers/net/ethernet/cavium/liquidio/response_manager.c b/drivers/net/ethernet/cavium/liquidio/response_manager.c
index 861050966e18..de1a8335b545 100644
--- a/drivers/net/ethernet/cavium/liquidio/response_manager.c
+++ b/drivers/net/ethernet/cavium/liquidio/response_manager.c
@@ -39,7 +39,8 @@ int octeon_setup_response_list(struct octeon_device *oct)
 	}
 	spin_lock_init(&oct->cmd_resp_wqlock);
 
-	oct->dma_comp_wq.wq = alloc_workqueue("dma-comp", WQ_MEM_RECLAIM, 0);
+	oct->dma_comp_wq.wq = alloc_workqueue("dma-comp",
+					      WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (!oct->dma_comp_wq.wq) {
 		dev_err(&oct->pci_dev->dev, "failed to create wq thread\n");
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
index 0f4efd505332..c96d1d6ba8fe 100644
--- a/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
+++ b/drivers/net/ethernet/freescale/dpaa2/dpaa2-eth.c
@@ -4884,7 +4884,7 @@ static int dpaa2_eth_probe(struct fsl_mc_device *dpni_dev)
 	priv->tx_tstamp_type = HWTSTAMP_TX_OFF;
 	priv->rx_tstamp = false;
 
-	priv->dpaa2_ptp_wq = alloc_workqueue("dpaa2_ptp_wq", 0, 0);
+	priv->dpaa2_ptp_wq = alloc_workqueue("dpaa2_ptp_wq", WQ_PERCPU, 0);
 	if (!priv->dpaa2_ptp_wq) {
 		err = -ENOMEM;
 		goto err_wq_alloc;
diff --git a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
index f5457ae0b64f..9d34d28ff168 100644
--- a/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
+++ b/drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c
@@ -12912,7 +12912,8 @@ static int __init hclge_init(void)
 {
 	pr_debug("%s is initializing\n", HCLGE_NAME);
 
-	hclge_wq = alloc_workqueue("%s", WQ_UNBOUND, 0, HCLGE_NAME);
+	hclge_wq = alloc_workqueue("%s", WQ_UNBOUND, 0,
+				   HCLGE_NAME);
 	if (!hclge_wq) {
 		pr_err("%s: failed to create workqueue\n", HCLGE_NAME);
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/intel/fm10k/fm10k_main.c b/drivers/net/ethernet/intel/fm10k/fm10k_main.c
index 142f07ca8bc0..b8c15b837fda 100644
--- a/drivers/net/ethernet/intel/fm10k/fm10k_main.c
+++ b/drivers/net/ethernet/intel/fm10k/fm10k_main.c
@@ -37,7 +37,7 @@ static int __init fm10k_init_module(void)
 	pr_info("%s\n", fm10k_copyright);
 
 	/* create driver workqueue */
-	fm10k_workqueue = alloc_workqueue("%s", WQ_MEM_RECLAIM, 0,
+	fm10k_workqueue = alloc_workqueue("%s", WQ_MEM_RECLAIM | WQ_PERCPU, 0,
 					  fm10k_driver_name);
 	if (!fm10k_workqueue)
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/intel/i40e/i40e_main.c b/drivers/net/ethernet/intel/i40e/i40e_main.c
index b14019d44b58..02fccdbbc288 100644
--- a/drivers/net/ethernet/intel/i40e/i40e_main.c
+++ b/drivers/net/ethernet/intel/i40e/i40e_main.c
@@ -16617,7 +16617,7 @@ static int __init i40e_init_module(void)
 	 * since we need to be able to guarantee forward progress even under
 	 * memory pressure.
 	 */
-	i40e_wq = alloc_workqueue("%s", 0, 0, i40e_driver_name);
+	i40e_wq = alloc_workqueue("%s", WQ_PERCPU, 0, i40e_driver_name);
 	if (!i40e_wq) {
 		pr_err("%s: Failed to create workqueue\n", i40e_driver_name);
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/cgx.c b/drivers/net/ethernet/marvell/octeontx2/af/cgx.c
index 0c46ba8a5adc..ca1343f43379 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/cgx.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/cgx.c
@@ -2005,7 +2005,7 @@ static int cgx_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 
 	/* init wq for processing linkup requests */
 	INIT_WORK(&cgx->cgx_cmd_work, cgx_lmac_linkup_work);
-	cgx->cgx_cmd_workq = alloc_workqueue("cgx_cmd_workq", 0, 0);
+	cgx->cgx_cmd_workq = alloc_workqueue("cgx_cmd_workq", WQ_PERCPU, 0);
 	if (!cgx->cgx_cmd_workq) {
 		dev_err(dev, "alloc workqueue failed for cgx cmd");
 		err = -ENOMEM;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c b/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
index d7030dfa5dad..a80c8e7c94f2 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/mcs_rvu_if.c
@@ -913,7 +913,7 @@ int rvu_mcs_init(struct rvu *rvu)
 	/* Initialize the wq for handling mcs interrupts */
 	INIT_LIST_HEAD(&rvu->mcs_intrq_head);
 	INIT_WORK(&rvu->mcs_intr_work, mcs_intr_handler_task);
-	rvu->mcs_intr_wq = alloc_workqueue("mcs_intr_wq", 0, 0);
+	rvu->mcs_intr_wq = alloc_workqueue("mcs_intr_wq", WQ_PERCPU, 0);
 	if (!rvu->mcs_intr_wq) {
 		dev_err(rvu->dev, "mcs alloc workqueue failed\n");
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
index 3303c475414a..3abd750a4bd7 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c
@@ -315,7 +315,7 @@ static int cgx_lmac_event_handler_init(struct rvu *rvu)
 	spin_lock_init(&rvu->cgx_evq_lock);
 	INIT_LIST_HEAD(&rvu->cgx_evq_head);
 	INIT_WORK(&rvu->cgx_evh_work, cgx_evhandler_task);
-	rvu->cgx_evh_wq = alloc_workqueue("rvu_evh_wq", 0, 0);
+	rvu->cgx_evh_wq = alloc_workqueue("rvu_evh_wq", WQ_PERCPU, 0);
 	if (!rvu->cgx_evh_wq) {
 		dev_err(rvu->dev, "alloc workqueue failed");
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/marvell/octeontx2/af/rvu_rep.c b/drivers/net/ethernet/marvell/octeontx2/af/rvu_rep.c
index 03099bc570bd..4415d0ce9aef 100644
--- a/drivers/net/ethernet/marvell/octeontx2/af/rvu_rep.c
+++ b/drivers/net/ethernet/marvell/octeontx2/af/rvu_rep.c
@@ -376,7 +376,7 @@ int rvu_rep_install_mcam_rules(struct rvu *rvu)
 	spin_lock_init(&rvu->rep_evtq_lock);
 	INIT_LIST_HEAD(&rvu->rep_evtq_head);
 	INIT_WORK(&rvu->rep_evt_work, rvu_rep_wq_handler);
-	rvu->rep_evt_wq = alloc_workqueue("rep_evt_wq", 0, 0);
+	rvu->rep_evt_wq = alloc_workqueue("rep_evt_wq", WQ_PERCPU, 0);
 	if (!rvu->rep_evt_wq) {
 		dev_err(rvu->dev, "REP workqueue allocation failed\n");
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c
index c691f0722154..77543d472345 100644
--- a/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c
+++ b/drivers/net/ethernet/marvell/octeontx2/nic/cn10k_ipsec.c
@@ -798,7 +798,8 @@ int cn10k_ipsec_init(struct net_device *netdev)
 	pf->ipsec.sa_size = sa_size;
 
 	INIT_WORK(&pf->ipsec.sa_work, cn10k_ipsec_sa_wq_handler);
-	pf->ipsec.sa_workq = alloc_workqueue("cn10k_ipsec_sa_workq", 0, 0);
+	pf->ipsec.sa_workq = alloc_workqueue("cn10k_ipsec_sa_workq",
+					     WQ_PERCPU, 0);
 	if (!pf->ipsec.sa_workq) {
 		netdev_err(pf->netdev, "SA alloc workqueue failed\n");
 		return -ENOMEM;
diff --git a/drivers/net/ethernet/marvell/prestera/prestera_main.c b/drivers/net/ethernet/marvell/prestera/prestera_main.c
index 71ffb55d1fc4..65e7ef033bde 100644
--- a/drivers/net/ethernet/marvell/prestera/prestera_main.c
+++ b/drivers/net/ethernet/marvell/prestera/prestera_main.c
@@ -1500,7 +1500,7 @@ EXPORT_SYMBOL(prestera_device_unregister);
 
 static int __init prestera_module_init(void)
 {
-	prestera_wq = alloc_workqueue("prestera", 0, 0);
+	prestera_wq = alloc_workqueue("prestera", WQ_PERCPU, 0);
 	if (!prestera_wq)
 		return -ENOMEM;
 
diff --git a/drivers/net/ethernet/marvell/prestera/prestera_pci.c b/drivers/net/ethernet/marvell/prestera/prestera_pci.c
index c45d108b2f6d..3e13322470da 100644
--- a/drivers/net/ethernet/marvell/prestera/prestera_pci.c
+++ b/drivers/net/ethernet/marvell/prestera/prestera_pci.c
@@ -898,7 +898,7 @@ static int prestera_pci_probe(struct pci_dev *pdev,
 
 	dev_info(fw->dev.dev, "Prestera FW is ready\n");
 
-	fw->wq = alloc_workqueue("prestera_fw_wq", WQ_HIGHPRI, 1);
+	fw->wq = alloc_workqueue("prestera_fw_wq", WQ_HIGHPRI | WQ_PERCPU, 1);
 	if (!fw->wq) {
 		err = -ENOMEM;
 		goto err_wq_alloc;
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.c b/drivers/net/ethernet/mellanox/mlxsw/core.c
index 980f3223f124..83c7cf3bbea3 100644
--- a/drivers/net/ethernet/mellanox/mlxsw/core.c
+++ b/drivers/net/ethernet/mellanox/mlxsw/core.c
@@ -886,7 +886,7 @@ static int mlxsw_emad_init(struct mlxsw_core *mlxsw_core)
 	if (!(mlxsw_core->bus->features & MLXSW_BUS_F_TXRX))
 		return 0;
 
-	emad_wq = alloc_workqueue("mlxsw_core_emad", 0, 0);
+	emad_wq = alloc_workqueue("mlxsw_core_emad", WQ_PERCPU, 0);
 	if (!emad_wq)
 		return -ENOMEM;
 	mlxsw_core->emad_wq = emad_wq;
@@ -3381,7 +3381,7 @@ static int __init mlxsw_core_module_init(void)
 	if (err)
 		return err;
 
-	mlxsw_wq = alloc_workqueue(mlxsw_core_driver_name, 0, 0);
+	mlxsw_wq = alloc_workqueue(mlxsw_core_driver_name, WQ_PERCPU, 0);
 	if (!mlxsw_wq) {
 		err = -ENOMEM;
 		goto err_alloc_workqueue;
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_main.c b/drivers/net/ethernet/netronome/nfp/nfp_main.c
index 71301dbd8fb5..48390b2fd44d 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_main.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_main.c
@@ -797,7 +797,7 @@ static int nfp_pci_probe(struct pci_dev *pdev,
 	pf->pdev = pdev;
 	pf->dev_info = dev_info;
 
-	pf->wq = alloc_workqueue("nfp-%s", 0, 2, pci_name(pdev));
+	pf->wq = alloc_workqueue("nfp-%s", WQ_PERCPU, 2, pci_name(pdev));
 	if (!pf->wq) {
 		err = -ENOMEM;
 		goto err_pci_priv_unset;
diff --git a/drivers/net/ethernet/qlogic/qed/qed_main.c b/drivers/net/ethernet/qlogic/qed/qed_main.c
index 886061d7351a..d4685ad4b169 100644
--- a/drivers/net/ethernet/qlogic/qed/qed_main.c
+++ b/drivers/net/ethernet/qlogic/qed/qed_main.c
@@ -1214,7 +1214,8 @@ static int qed_slowpath_wq_start(struct qed_dev *cdev)
 		hwfn = &cdev->hwfns[i];
 
 		hwfn->slowpath_wq = alloc_workqueue("slowpath-%02x:%02x.%02x",
-					 0, 0, cdev->pdev->bus->number,
+					 WQ_PERCPU, 0,
+					 cdev->pdev->bus->number,
 					 PCI_SLOT(cdev->pdev->devfn),
 					 hwfn->abs_pf_id);
 
diff --git a/drivers/net/ethernet/wiznet/w5100.c b/drivers/net/ethernet/wiznet/w5100.c
index b77f096eaf99..c5424d882135 100644
--- a/drivers/net/ethernet/wiznet/w5100.c
+++ b/drivers/net/ethernet/wiznet/w5100.c
@@ -1142,7 +1142,7 @@ int w5100_probe(struct device *dev, const struct w5100_ops *ops,
 	if (err < 0)
 		goto err_register;
 
-	priv->xfer_wq = alloc_workqueue("%s", WQ_MEM_RECLAIM, 0,
+	priv->xfer_wq = alloc_workqueue("%s", WQ_MEM_RECLAIM | WQ_PERCPU, 0,
 					netdev_name(ndev));
 	if (!priv->xfer_wq) {
 		err = -ENOMEM;
diff --git a/drivers/net/fjes/fjes_main.c b/drivers/net/fjes/fjes_main.c
index 4a4ed2ccf72f..b63965d9a1ba 100644
--- a/drivers/net/fjes/fjes_main.c
+++ b/drivers/net/fjes/fjes_main.c
@@ -1364,14 +1364,15 @@ static int fjes_probe(struct platform_device *plat_dev)
 	adapter->force_reset = false;
 	adapter->open_guard = false;
 
-	adapter->txrx_wq = alloc_workqueue(DRV_NAME "/txrx", WQ_MEM_RECLAIM, 0);
+	adapter->txrx_wq = alloc_workqueue(DRV_NAME "/txrx",
+					   WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (unlikely(!adapter->txrx_wq)) {
 		err = -ENOMEM;
 		goto err_free_netdev;
 	}
 
 	adapter->control_wq = alloc_workqueue(DRV_NAME "/control",
-					      WQ_MEM_RECLAIM, 0);
+					      WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (unlikely(!adapter->control_wq)) {
 		err = -ENOMEM;
 		goto err_free_txrx_wq;
diff --git a/drivers/net/wireguard/device.c b/drivers/net/wireguard/device.c
index 813bd10d3dc7..46a71ec36af8 100644
--- a/drivers/net/wireguard/device.c
+++ b/drivers/net/wireguard/device.c
@@ -333,7 +333,8 @@ static int wg_newlink(struct net_device *dev,
 		goto err_free_peer_hashtable;
 
 	wg->handshake_receive_wq = alloc_workqueue("wg-kex-%s",
-			WQ_CPU_INTENSIVE | WQ_FREEZABLE, 0, dev->name);
+			WQ_CPU_INTENSIVE | WQ_FREEZABLE | WQ_PERCPU, 0,
+			dev->name);
 	if (!wg->handshake_receive_wq)
 		goto err_free_index_hashtable;
 
@@ -343,7 +344,8 @@ static int wg_newlink(struct net_device *dev,
 		goto err_destroy_handshake_receive;
 
 	wg->packet_crypt_wq = alloc_workqueue("wg-crypt-%s",
-			WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM, 0, dev->name);
+			WQ_CPU_INTENSIVE | WQ_MEM_RECLAIM | WQ_PERCPU, 0,
+			dev->name);
 	if (!wg->packet_crypt_wq)
 		goto err_destroy_handshake_send;
 
diff --git a/drivers/net/wwan/t7xx/t7xx_hif_dpmaif_rx.c b/drivers/net/wwan/t7xx/t7xx_hif_dpmaif_rx.c
index 6a7a26085fc7..2310493203d3 100644
--- a/drivers/net/wwan/t7xx/t7xx_hif_dpmaif_rx.c
+++ b/drivers/net/wwan/t7xx/t7xx_hif_dpmaif_rx.c
@@ -1085,7 +1085,8 @@ static void t7xx_dpmaif_bat_release_work(struct work_struct *work)
 int t7xx_dpmaif_bat_rel_wq_alloc(struct dpmaif_ctrl *dpmaif_ctrl)
 {
 	dpmaif_ctrl->bat_release_wq = alloc_workqueue("dpmaif_bat_release_work_queue",
-						      WQ_MEM_RECLAIM, 1);
+						      WQ_MEM_RECLAIM | WQ_PERCPU,
+						      1);
 	if (!dpmaif_ctrl->bat_release_wq)
 		return -ENOMEM;
 
diff --git a/drivers/net/wwan/wwan_hwsim.c b/drivers/net/wwan/wwan_hwsim.c
index b02befd1b6fb..733688cd4607 100644
--- a/drivers/net/wwan/wwan_hwsim.c
+++ b/drivers/net/wwan/wwan_hwsim.c
@@ -509,7 +509,7 @@ static int __init wwan_hwsim_init(void)
 	if (wwan_hwsim_devsnum < 0 || wwan_hwsim_devsnum > 128)
 		return -EINVAL;
 
-	wwan_wq = alloc_workqueue("wwan_wq", 0, 0);
+	wwan_wq = alloc_workqueue("wwan_wq", WQ_PERCPU, 0);
 	if (!wwan_wq)
 		return -ENOMEM;
 
diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index d1b5705dc0c6..183c1e0b405a 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -252,7 +252,8 @@ int __init ceph_msgr_init(void)
 	 * The number of active work items is limited by the number of
 	 * connections, so leave @max_active at default.
 	 */
-	ceph_msgr_wq = alloc_workqueue("ceph-msgr", WQ_MEM_RECLAIM, 0);
+	ceph_msgr_wq = alloc_workqueue("ceph-msgr",
+				       WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (ceph_msgr_wq)
 		return 0;
 
diff --git a/net/core/sock_diag.c b/net/core/sock_diag.c
index b23594c767f2..026ce9bd9e5e 100644
--- a/net/core/sock_diag.c
+++ b/net/core/sock_diag.c
@@ -348,7 +348,7 @@ static struct pernet_operations diag_net_ops = {
 
 static int __init sock_diag_init(void)
 {
-	broadcast_wq = alloc_workqueue("sock_diag_events", 0, 0);
+	broadcast_wq = alloc_workqueue("sock_diag_events", WQ_PERCPU, 0);
 	BUG_ON(!broadcast_wq);
 	return register_pernet_subsys(&diag_net_ops);
 }
diff --git a/net/rds/ib_rdma.c b/net/rds/ib_rdma.c
index d1cfceeff133..6585164c7059 100644
--- a/net/rds/ib_rdma.c
+++ b/net/rds/ib_rdma.c
@@ -672,7 +672,8 @@ struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_ibdev,
 
 int rds_ib_mr_init(void)
 {
-	rds_ib_mr_wq = alloc_workqueue("rds_mr_flushd", WQ_MEM_RECLAIM, 0);
+	rds_ib_mr_wq = alloc_workqueue("rds_mr_flushd",
+				       WQ_MEM_RECLAIM | WQ_PERCPU, 0);
 	if (!rds_ib_mr_wq)
 		return -ENOMEM;
 	return 0;
diff --git a/net/rxrpc/rxperf.c b/net/rxrpc/rxperf.c
index 0377301156b0..2ea71e3831f7 100644
--- a/net/rxrpc/rxperf.c
+++ b/net/rxrpc/rxperf.c
@@ -630,7 +630,7 @@ static int __init rxperf_init(void)
 
 	pr_info("Server registering\n");
 
-	rxperf_workqueue = alloc_workqueue("rxperf", 0, 0);
+	rxperf_workqueue = alloc_workqueue("rxperf", WQ_PERCPU, 0);
 	if (!rxperf_workqueue)
 		goto error_workqueue;
 
diff --git a/net/smc/af_smc.c b/net/smc/af_smc.c
index a7187e5873ec..9097e4f24d2b 100644
--- a/net/smc/af_smc.c
+++ b/net/smc/af_smc.c
@@ -3535,15 +3535,15 @@ static int __init smc_init(void)
 
 	rc = -ENOMEM;
 
-	smc_tcp_ls_wq = alloc_workqueue("smc_tcp_ls_wq", 0, 0);
+	smc_tcp_ls_wq = alloc_workqueue("smc_tcp_ls_wq", WQ_PERCPU, 0);
 	if (!smc_tcp_ls_wq)
 		goto out_pnet;
 
-	smc_hs_wq = alloc_workqueue("smc_hs_wq", 0, 0);
+	smc_hs_wq = alloc_workqueue("smc_hs_wq", WQ_PERCPU, 0);
 	if (!smc_hs_wq)
 		goto out_alloc_tcp_ls_wq;
 
-	smc_close_wq = alloc_workqueue("smc_close_wq", 0, 0);
+	smc_close_wq = alloc_workqueue("smc_close_wq", WQ_PERCPU, 0);
 	if (!smc_close_wq)
 		goto out_alloc_hs_wq;
 
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index e216d237865b..a9e80f44307d 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -896,7 +896,7 @@ static int smc_lgr_create(struct smc_sock *smc, struct smc_init_info *ini)
 		rc = SMC_CLC_DECL_MEM;
 		goto ism_put_vlan;
 	}
-	lgr->tx_wq = alloc_workqueue("smc_tx_wq-%*phN", 0, 0,
+	lgr->tx_wq = alloc_workqueue("smc_tx_wq-%*phN", WQ_PERCPU, 0,
 				     SMC_LGR_ID_SIZE, &lgr->id);
 	if (!lgr->tx_wq) {
 		rc = -ENOMEM;
diff --git a/net/tls/tls_device.c b/net/tls/tls_device.c
index a82fdcf19969..a64ae15b1a60 100644
--- a/net/tls/tls_device.c
+++ b/net/tls/tls_device.c
@@ -1412,7 +1412,7 @@ int __init tls_device_init(void)
 	if (!dummy_page)
 		return -ENOMEM;
 
-	destruct_wq = alloc_workqueue("ktls_device_destruct", 0, 0);
+	destruct_wq = alloc_workqueue("ktls_device_destruct", WQ_PERCPU, 0);
 	if (!destruct_wq) {
 		err = -ENOMEM;
 		goto err_free_dummy;
diff --git a/net/vmw_vsock/virtio_transport.c b/net/vmw_vsock/virtio_transport.c
index b6569b0ca2bb..8c867023a2e5 100644
--- a/net/vmw_vsock/virtio_transport.c
+++ b/net/vmw_vsock/virtio_transport.c
@@ -926,7 +926,7 @@ static int __init virtio_vsock_init(void)
 {
 	int ret;
 
-	virtio_vsock_workqueue = alloc_workqueue("virtio_vsock", 0, 0);
+	virtio_vsock_workqueue = alloc_workqueue("virtio_vsock", WQ_PERCPU, 0);
 	if (!virtio_vsock_workqueue)
 		return -ENOMEM;
 
diff --git a/net/vmw_vsock/vsock_loopback.c b/net/vmw_vsock/vsock_loopback.c
index 6e78927a598e..bc2ff918b315 100644
--- a/net/vmw_vsock/vsock_loopback.c
+++ b/net/vmw_vsock/vsock_loopback.c
@@ -139,7 +139,7 @@ static int __init vsock_loopback_init(void)
 	struct vsock_loopback *vsock = &the_vsock_loopback;
 	int ret;
 
-	vsock->workqueue = alloc_workqueue("vsock-loopback", 0, 0);
+	vsock->workqueue = alloc_workqueue("vsock-loopback", WQ_PERCPU, 0);
 	if (!vsock->workqueue)
 		return -ENOMEM;
 
-- 
2.51.0

Re: [PATCH net-next v2 0/3] net: replace wq users and add WQ_PERCPU to alloc_workqueue() users

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (main)
by Jakub Kicinski <kuba@kernel.org>:

On Thu, 18 Sep 2025 16:24:24 +0200 you wrote:
> Hi!
> 
> Below is a summary of a discussion about the Workqueue API and cpu isolation
> considerations. Details and more information are available here:
> 
>         "workqueue: Always use wq_select_unbound_cpu() for WORK_CPU_UNBOUND."
>         https://lore.kernel.org/all/20250221112003.1dSuoGyc@linutronix.de/
> 
> [...]

Here is the summary with links:
  - [net-next,v2,1/3] net: replace use of system_unbound_wq with system_dfl_wq
    https://git.kernel.org/netdev/net-next/c/9870d350e45a
  - [net-next,v2,2/3] net: replace use of system_wq with system_percpu_wq
    https://git.kernel.org/netdev/net-next/c/5fd8bb982e10
  - [net-next,v2,3/3] net: WQ_PERCPU added to alloc_workqueue users
    https://git.kernel.org/netdev/net-next/c/27ce71e1ce81

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html