[PATCH 0/2] wifi: replace system_wq and system_unbound_wq uses

[PATCH 0/2] wifi: replace system_wq and system_unbound_wq uses

[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.

Thanks!

Marco Crivellari (2):
  wifi: iwlwifi: replace use of system_unbound_wq with system_dfl_wq
  wifi: replace use of system_wq with system_percpu_wq

 drivers/net/wireless/intel/ipw2x00/ipw2100.c   | 6 +++---
 drivers/net/wireless/intel/ipw2x00/ipw2200.c   | 2 +-
 drivers/net/wireless/intel/iwlwifi/fw/dbg.c    | 4 ++--
 drivers/net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
 drivers/net/wireless/intel/iwlwifi/mvm/tdls.c  | 6 +++---
 5 files changed, 10 insertions(+), 10 deletions(-)

-- 
2.51.0

[PATCH 1/2] wifi: iwlwifi: 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/wireless/intel/iwlwifi/fw/dbg.c    | 4 ++--
 drivers/net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/wireless/intel/iwlwifi/fw/dbg.c b/drivers/net/wireless/intel/iwlwifi/fw/dbg.c
index 2ce55859641c..e5f31b82339a 100644
--- a/drivers/net/wireless/intel/iwlwifi/fw/dbg.c
+++ b/drivers/net/wireless/intel/iwlwifi/fw/dbg.c
@@ -2933,7 +2933,7 @@ int iwl_fw_dbg_collect_desc(struct iwl_fw_runtime *fwrt,
 	IWL_WARN(fwrt, "Collecting data: trigger %d fired.\n",
 		 le32_to_cpu(desc->trig_desc.type));
 
-	queue_delayed_work(system_unbound_wq, &wk_data->wk,
+	queue_delayed_work(system_dfl_wq, &wk_data->wk,
 			   usecs_to_jiffies(delay));
 
 	return 0;
@@ -3237,7 +3237,7 @@ int iwl_fw_dbg_ini_collect(struct iwl_fw_runtime *fwrt,
 	if (sync)
 		iwl_fw_dbg_collect_sync(fwrt, idx);
 	else
-		queue_delayed_work(system_unbound_wq,
+		queue_delayed_work(system_dfl_wq,
 				   &fwrt->dump.wks[idx].wk,
 				   usecs_to_jiffies(delay));
 
diff --git a/drivers/net/wireless/intel/iwlwifi/iwl-trans.h b/drivers/net/wireless/intel/iwlwifi/iwl-trans.h
index a552669db6e2..2f24b639c133 100644
--- a/drivers/net/wireless/intel/iwlwifi/iwl-trans.h
+++ b/drivers/net/wireless/intel/iwlwifi/iwl-trans.h
@@ -1092,7 +1092,7 @@ static inline void iwl_trans_schedule_reset(struct iwl_trans *trans,
 	 */
 	trans->restart.during_reset = test_bit(STATUS_IN_SW_RESET,
 					       &trans->status);
-	queue_delayed_work(system_unbound_wq, &trans->restart.wk, 0);
+	queue_delayed_work(system_dfl_wq, &trans->restart.wk, 0);
 }
 
 static inline void iwl_trans_fw_error(struct iwl_trans *trans,
-- 
2.51.0

[PATCH 2/2] wifi: 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_wq is a per-cpu workqueue, but its name is not clear. Because of
that has been renamed in system_percpu_wq.

The old system_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/wireless/intel/ipw2x00/ipw2100.c  | 6 +++---
 drivers/net/wireless/intel/ipw2x00/ipw2200.c  | 2 +-
 drivers/net/wireless/intel/iwlwifi/mvm/tdls.c | 6 +++---
 3 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/drivers/net/wireless/intel/ipw2x00/ipw2100.c b/drivers/net/wireless/intel/ipw2x00/ipw2100.c
index 215814861cbd..c7c5bc0f1650 100644
--- a/drivers/net/wireless/intel/ipw2x00/ipw2100.c
+++ b/drivers/net/wireless/intel/ipw2x00/ipw2100.c
@@ -2143,7 +2143,7 @@ static void isr_indicate_rf_kill(struct ipw2100_priv *priv, u32 status)
 
 	/* Make sure the RF Kill check timer is running */
 	priv->stop_rf_kill = 0;
-	mod_delayed_work(system_wq, &priv->rf_kill, round_jiffies_relative(HZ));
+	mod_delayed_work(system_percpu_wq, &priv->rf_kill, round_jiffies_relative(HZ));
 }
 
 static void ipw2100_scan_event(struct work_struct *work)
@@ -2170,7 +2170,7 @@ static void isr_scan_complete(struct ipw2100_priv *priv, u32 status)
 				      round_jiffies_relative(msecs_to_jiffies(4000)));
 	} else {
 		priv->user_requested_scan = 0;
-		mod_delayed_work(system_wq, &priv->scan_event, 0);
+		mod_delayed_work(system_percpu_wq, &priv->scan_event, 0);
 	}
 }
 
@@ -4252,7 +4252,7 @@ static int ipw_radio_kill_sw(struct ipw2100_priv *priv, int disable_radio)
 					  "disabled by HW switch\n");
 			/* Make sure the RF_KILL check timer is running */
 			priv->stop_rf_kill = 0;
-			mod_delayed_work(system_wq, &priv->rf_kill,
+			mod_delayed_work(system_percpu_wq, &priv->rf_kill,
 					 round_jiffies_relative(HZ));
 		} else
 			schedule_reset(priv);
diff --git a/drivers/net/wireless/intel/ipw2x00/ipw2200.c b/drivers/net/wireless/intel/ipw2x00/ipw2200.c
index 24a5624ef207..09035a77e775 100644
--- a/drivers/net/wireless/intel/ipw2x00/ipw2200.c
+++ b/drivers/net/wireless/intel/ipw2x00/ipw2200.c
@@ -4415,7 +4415,7 @@ static void handle_scan_event(struct ipw_priv *priv)
 				      round_jiffies_relative(msecs_to_jiffies(4000)));
 	} else {
 		priv->user_requested_scan = 0;
-		mod_delayed_work(system_wq, &priv->scan_event, 0);
+		mod_delayed_work(system_percpu_wq, &priv->scan_event, 0);
 	}
 }
 
diff --git a/drivers/net/wireless/intel/iwlwifi/mvm/tdls.c b/drivers/net/wireless/intel/iwlwifi/mvm/tdls.c
index 36379b738de1..0df31639fa5e 100644
--- a/drivers/net/wireless/intel/iwlwifi/mvm/tdls.c
+++ b/drivers/net/wireless/intel/iwlwifi/mvm/tdls.c
@@ -234,7 +234,7 @@ void iwl_mvm_rx_tdls_notif(struct iwl_mvm *mvm, struct iwl_rx_cmd_buffer *rxb)
 	 * Also convert TU to msec.
 	 */
 	delay = TU_TO_MS(vif->bss_conf.dtim_period * vif->bss_conf.beacon_int);
-	mod_delayed_work(system_wq, &mvm->tdls_cs.dwork,
+	mod_delayed_work(system_percpu_wq, &mvm->tdls_cs.dwork,
 			 msecs_to_jiffies(delay));
 
 	iwl_mvm_tdls_update_cs_state(mvm, IWL_MVM_TDLS_SW_ACTIVE);
@@ -548,7 +548,7 @@ iwl_mvm_tdls_channel_switch(struct ieee80211_hw *hw,
 	 */
 	delay = 2 * TU_TO_MS(vif->bss_conf.dtim_period *
 			     vif->bss_conf.beacon_int);
-	mod_delayed_work(system_wq, &mvm->tdls_cs.dwork,
+	mod_delayed_work(system_percpu_wq, &mvm->tdls_cs.dwork,
 			 msecs_to_jiffies(delay));
 	return 0;
 }
@@ -659,6 +659,6 @@ iwl_mvm_tdls_recv_channel_switch(struct ieee80211_hw *hw,
 	/* register a timeout in case we don't succeed in switching */
 	delay = vif->bss_conf.dtim_period * vif->bss_conf.beacon_int *
 		1024 / 1000;
-	mod_delayed_work(system_wq, &mvm->tdls_cs.dwork,
+	mod_delayed_work(system_percpu_wq, &mvm->tdls_cs.dwork,
 			 msecs_to_jiffies(delay));
 }
-- 
2.51.0

Re: [PATCH 0/2] wifi: replace system_wq and system_unbound_wq uses

[Marco Crivellari]

On Thu, Sep 18, 2025 at 5:26 PM Marco Crivellari
<marco.crivellari@suse.com> wrote:
>[...]
> Marco Crivellari (2):
>   wifi: iwlwifi: replace use of system_unbound_wq with system_dfl_wq
>   wifi: replace use of system_wq with system_percpu_wq
>
>  drivers/net/wireless/intel/ipw2x00/ipw2100.c   | 6 +++---
>  drivers/net/wireless/intel/ipw2x00/ipw2200.c   | 2 +-
>  drivers/net/wireless/intel/iwlwifi/fw/dbg.c    | 4 ++--
>  drivers/net/wireless/intel/iwlwifi/iwl-trans.h | 2 +-
>  drivers/net/wireless/intel/iwlwifi/mvm/tdls.c  | 6 +++---
>  5 files changed, 10 insertions(+), 10 deletions(-)
>
> --
> 2.51.0
>

Gentle ping.

Thank you!

-- 

Marco Crivellari

L3 Support Engineer, Technology & Product