Merge branches 'pm-em' and 'pm-runtime'

rafaeljw · rafaeljw · commit 871302441256 · 2025-03-24T15:06:06.000+01:00
Merge Energy Model handling code updates and updates of the runtime PM
core code for 6.15-rc1:

 - Clean up the Energy Model handling code somewhat (Rafael Wysocki).

 - Use kfree_rcu() to simplify the handling of runtime Energy Model
   updates (Li RongQing).

 - Add an entry for the Energy Model framework to MAINTAINERS as
   properly maintained (Lukasz Luba).

 - Address RCU-related sparse warnings in the Energy Model code (Rafael
   Wysocki).

 - Remove ENERGY_MODEL dependency on SMP and allow it to be selected
   when DEVFREQ is set without CPUFREQ so it can be used on a wider
   range of systems (Jeson Gao).

 - Unify error handling during runtime suspend and runtime resume in the
   core to help drivers to implement more consistent runtime PM error
   handling (Rafael Wysocki).

 - Drop a redundant check from pm_runtime_force_resume() and rearrange
   documentation related to __pm_runtime_disable() (Rafael Wysocki).

* pm-em:
  PM: EM: Rework the depends on for CONFIG_ENERGY_MODEL
  PM: EM: Address RCU-related sparse warnings
  PM: EM: Consify two parameters of em_dev_register_perf_domain()
  MAINTAINERS: Add Energy Model framework as properly maintained
  PM: EM: use kfree_rcu() to simplify the code
  PM: EM: Slightly reduce em_check_capacity_update() overhead
  PM: EM: Drop unused parameter from em_adjust_new_capacity()

* pm-runtime:
  PM: runtime: Unify error handling during suspend and resume
  PM: runtime: Drop status check from pm_runtime_force_resume()
  PM: Rearrange documentation related to __pm_runtime_disable()
diff --git a/MAINTAINERS b/MAINTAINERS
@@ -8528,6 +8528,15 @@ M:	Maxim Levitsky <maximlevitsky@gmail.com>
 S:	Maintained
 F:	drivers/media/rc/ene_ir.*
 
+ENERGY MODEL
+M:	Lukasz Luba <lukasz.luba@arm.com>
+M:	"Rafael J. Wysocki" <rafael@kernel.org>
+L:	linux-pm@vger.kernel.org
+S:	Maintained
+F:	kernel/power/energy_model.c
+F:	include/linux/energy_model.h
+F:	Documentation/power/energy-model.rst
+
 EPAPR HYPERVISOR BYTE CHANNEL DEVICE DRIVER
 M:	Laurentiu Tudor <laurentiu.tudor@nxp.com>
 L:	linuxppc-dev@lists.ozlabs.org
diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
@@ -1404,6 +1404,10 @@ static int device_suspend_late(struct device *dev, pm_message_t state, bool asyn
 	TRACE_DEVICE(dev);
 	TRACE_SUSPEND(0);
 
+	/*
+	 * Disable runtime PM for the device without checking if there is a
+	 * pending resume request for it.
+	 */
 	__pm_runtime_disable(dev, false);
 
 	dpm_wait_for_subordinate(dev, async);
diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
@@ -448,8 +448,19 @@ static int rpm_callback(int (*cb)(struct device *), struct device *dev)
 		retval = __rpm_callback(cb, dev);
 	}
 
-	dev->power.runtime_error = retval;
-	return retval != -EACCES ? retval : -EIO;
+	/*
+	 * Since -EACCES means that runtime PM is disabled for the given device,
+	 * it should not be returned by runtime PM callbacks.  If it is returned
+	 * nevertheless, assume it to be a transient error and convert it to
+	 * -EAGAIN.
+	 */
+	if (retval == -EACCES)
+		retval = -EAGAIN;
+
+	if (retval != -EAGAIN && retval != -EBUSY)
+		dev->power.runtime_error = retval;
+
+	return retval;
 }
 
 /**
@@ -725,21 +736,18 @@ static int rpm_suspend(struct device *dev, int rpmflags)
 	dev->power.deferred_resume = false;
 	wake_up_all(&dev->power.wait_queue);
 
-	if (retval == -EAGAIN || retval == -EBUSY) {
-		dev->power.runtime_error = 0;
+	/*
+	 * On transient errors, if the callback routine failed an autosuspend,
+	 * and if the last_busy time has been updated so that there is a new
+	 * autosuspend expiration time, automatically reschedule another
+	 * autosuspend.
+	 */
+	if (!dev->power.runtime_error && (rpmflags & RPM_AUTO) &&
+	    pm_runtime_autosuspend_expiration(dev) != 0)
+		goto repeat;
+
+	pm_runtime_cancel_pending(dev);
 
-		/*
-		 * If the callback routine failed an autosuspend, and
-		 * if the last_busy time has been updated so that there
-		 * is a new autosuspend expiration time, automatically
-		 * reschedule another autosuspend.
-		 */
-		if ((rpmflags & RPM_AUTO) &&
-		    pm_runtime_autosuspend_expiration(dev) != 0)
-			goto repeat;
-	} else {
-		pm_runtime_cancel_pending(dev);
-	}
 	goto out;
 }
 
@@ -1460,20 +1468,6 @@ int pm_runtime_barrier(struct device *dev)
 }
 EXPORT_SYMBOL_GPL(pm_runtime_barrier);
 
-/**
- * __pm_runtime_disable - Disable runtime PM of a device.
- * @dev: Device to handle.
- * @check_resume: If set, check if there's a resume request for the device.
- *
- * Increment power.disable_depth for the device and if it was zero previously,
- * cancel all pending runtime PM requests for the device and wait for all
- * operations in progress to complete.  The device can be either active or
- * suspended after its runtime PM has been disabled.
- *
- * If @check_resume is set and there's a resume request pending when
- * __pm_runtime_disable() is called and power.disable_depth is zero, the
- * function will wake up the device before disabling its runtime PM.
- */
 void __pm_runtime_disable(struct device *dev, bool check_resume)
 {
 	spin_lock_irq(&dev->power.lock);
@@ -1959,7 +1953,7 @@ int pm_runtime_force_resume(struct device *dev)
 	int (*callback)(struct device *);
 	int ret = 0;
 
-	if (!pm_runtime_status_suspended(dev) || !dev->power.needs_force_resume)
+	if (!dev->power.needs_force_resume)
 		goto out;
 
 	/*
diff --git a/drivers/powercap/Kconfig b/drivers/powercap/Kconfig
@@ -82,7 +82,7 @@ config DTPM
 
 config DTPM_CPU
 	bool "Add CPU power capping based on the energy model"
-	depends on DTPM && ENERGY_MODEL
+	depends on DTPM && ENERGY_MODEL && SMP
 	help
 	  This enables support for CPU power limitation based on
 	  energy model.
diff --git a/include/linux/energy_model.h b/include/linux/energy_model.h
@@ -167,13 +167,13 @@ struct em_data_callback {
 struct em_perf_domain *em_cpu_get(int cpu);
 struct em_perf_domain *em_pd_get(struct device *dev);
 int em_dev_update_perf_domain(struct device *dev,
-			      struct em_perf_table __rcu *new_table);
+			      struct em_perf_table *new_table);
 int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states,
-				struct em_data_callback *cb, cpumask_t *span,
-				bool microwatts);
+				const struct em_data_callback *cb,
+				const cpumask_t *cpus, bool microwatts);
 void em_dev_unregister_perf_domain(struct device *dev);
-struct em_perf_table __rcu *em_table_alloc(struct em_perf_domain *pd);
-void em_table_free(struct em_perf_table __rcu *table);
+struct em_perf_table *em_table_alloc(struct em_perf_domain *pd);
+void em_table_free(struct em_perf_table *table);
 int em_dev_compute_costs(struct device *dev, struct em_perf_state *table,
 			 int nr_states);
 int em_dev_update_chip_binning(struct device *dev);
@@ -346,8 +346,8 @@ struct em_data_callback {};
 
 static inline
 int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states,
-				struct em_data_callback *cb, cpumask_t *span,
-				bool microwatts)
+				const struct em_data_callback *cb,
+				const cpumask_t *cpus, bool microwatts)
 {
 	return -EINVAL;
 }
@@ -373,14 +373,14 @@ static inline int em_pd_nr_perf_states(struct em_perf_domain *pd)
 	return 0;
 }
 static inline
-struct em_perf_table __rcu *em_table_alloc(struct em_perf_domain *pd)
+struct em_perf_table *em_table_alloc(struct em_perf_domain *pd)
 {
 	return NULL;
 }
-static inline void em_table_free(struct em_perf_table __rcu *table) {}
+static inline void em_table_free(struct em_perf_table *table) {}
 static inline
 int em_dev_update_perf_domain(struct device *dev,
-			      struct em_perf_table __rcu *new_table)
+			      struct em_perf_table *new_table)
 {
 	return -EINVAL;
 }
diff --git a/include/linux/pm_runtime.h b/include/linux/pm_runtime.h
@@ -556,11 +556,18 @@ static inline int pm_runtime_set_suspended(struct device *dev)
  * pm_runtime_disable - Disable runtime PM for a device.
  * @dev: Target device.
  *
- * Prevent the runtime PM framework from working with @dev (by incrementing its
- * "blocking" counter).
- *
- * For each invocation of this function for @dev there must be a matching
- * pm_runtime_enable() call in order for runtime PM to be enabled for it.
+ * Prevent the runtime PM framework from working with @dev by incrementing its
+ * "disable" counter.
+ *
+ * If the counter is zero when this function runs and there is a pending runtime
+ * resume request for @dev, it will be resumed.  If the counter is still zero at
+ * that point, all of the pending runtime PM requests for @dev will be canceled
+ * and all runtime PM operations in progress involving it will be waited for to
+ * complete.
+ *
+ * For each invocation of this function for @dev, there must be a matching
+ * pm_runtime_enable() call, so that runtime PM is eventually enabled for it
+ * again.
  */
 static inline void pm_runtime_disable(struct device *dev)
 {
diff --git a/kernel/power/Kconfig b/kernel/power/Kconfig
@@ -380,8 +380,7 @@ config CPU_PM
 
 config ENERGY_MODEL
 	bool "Energy Model for devices with DVFS (CPUs, GPUs, etc)"
-	depends on SMP
-	depends on CPU_FREQ
+	depends on CPU_FREQ || PM_DEVFREQ
 	help
 	  Several subsystems (thermal and/or the task scheduler for example)
 	  can leverage information about the energy consumed by devices to
diff --git a/kernel/power/energy_model.c b/kernel/power/energy_model.c
@@ -161,22 +161,10 @@ static void em_debug_create_pd(struct device *dev) {}
 static void em_debug_remove_pd(struct device *dev) {}
 #endif
 
-static void em_destroy_table_rcu(struct rcu_head *rp)
-{
-	struct em_perf_table __rcu *table;
-
-	table = container_of(rp, struct em_perf_table, rcu);
-	kfree(table);
-}
-
 static void em_release_table_kref(struct kref *kref)
 {
-	struct em_perf_table __rcu *table;
-
 	/* It was the last owner of this table so we can free */
-	table = container_of(kref, struct em_perf_table, kref);
-
-	call_rcu(&table->rcu, em_destroy_table_rcu);
+	kfree_rcu(container_of(kref, struct em_perf_table, kref), rcu);
 }
 
 /**
@@ -185,7 +173,7 @@ static void em_release_table_kref(struct kref *kref)
  *
  * No return values.
  */
-void em_table_free(struct em_perf_table __rcu *table)
+void em_table_free(struct em_perf_table *table)
 {
 	kref_put(&table->kref, em_release_table_kref);
 }
@@ -198,9 +186,9 @@ void em_table_free(struct em_perf_table __rcu *table)
  * has a user.
  * Returns allocated table or NULL.
  */
-struct em_perf_table __rcu *em_table_alloc(struct em_perf_domain *pd)
+struct em_perf_table *em_table_alloc(struct em_perf_domain *pd)
 {
-	struct em_perf_table __rcu *table;
+	struct em_perf_table *table;
 	int table_size;
 
 	table_size = sizeof(struct em_perf_state) * pd->nr_perf_states;
@@ -239,7 +227,7 @@ static void em_init_performance(struct device *dev, struct em_perf_domain *pd,
 }
 
 static int em_compute_costs(struct device *dev, struct em_perf_state *table,
-			    struct em_data_callback *cb, int nr_states,
+			    const struct em_data_callback *cb, int nr_states,
 			    unsigned long flags)
 {
 	unsigned long prev_cost = ULONG_MAX;
@@ -308,9 +296,9 @@ int em_dev_compute_costs(struct device *dev, struct em_perf_state *table,
  * Return 0 on success or an error code on failure.
  */
 int em_dev_update_perf_domain(struct device *dev,
-			      struct em_perf_table __rcu *new_table)
+			      struct em_perf_table *new_table)
 {
-	struct em_perf_table __rcu *old_table;
+	struct em_perf_table *old_table;
 	struct em_perf_domain *pd;
 
 	if (!dev)
@@ -327,7 +315,8 @@ int em_dev_update_perf_domain(struct device *dev,
 
 	kref_get(&new_table->kref);
 
-	old_table = pd->em_table;
+	old_table = rcu_dereference_protected(pd->em_table,
+					      lockdep_is_held(&em_pd_mutex));
 	rcu_assign_pointer(pd->em_table, new_table);
 
 	em_cpufreq_update_efficiencies(dev, new_table->state);
@@ -341,7 +330,7 @@ EXPORT_SYMBOL_GPL(em_dev_update_perf_domain);
 
 static int em_create_perf_table(struct device *dev, struct em_perf_domain *pd,
 				struct em_perf_state *table,
-				struct em_data_callback *cb,
+				const struct em_data_callback *cb,
 				unsigned long flags)
 {
 	unsigned long power, freq, prev_freq = 0;
@@ -396,10 +385,11 @@ static int em_create_perf_table(struct device *dev, struct em_perf_domain *pd,
 }
 
 static int em_create_pd(struct device *dev, int nr_states,
-			struct em_data_callback *cb, cpumask_t *cpus,
+			const struct em_data_callback *cb,
+			const cpumask_t *cpus,
 			unsigned long flags)
 {
-	struct em_perf_table __rcu *em_table;
+	struct em_perf_table *em_table;
 	struct em_perf_domain *pd;
 	struct device *cpu_dev;
 	int cpu, ret, num_cpus;
@@ -556,9 +546,10 @@ EXPORT_SYMBOL_GPL(em_cpu_get);
  * Return 0 on success
  */
 int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states,
-				struct em_data_callback *cb, cpumask_t *cpus,
-				bool microwatts)
+				const struct em_data_callback *cb,
+				const cpumask_t *cpus, bool microwatts)
 {
+	struct em_perf_table *em_table;
 	unsigned long cap, prev_cap = 0;
 	unsigned long flags = 0;
 	int cpu, ret;
@@ -631,7 +622,9 @@ int em_dev_register_perf_domain(struct device *dev, unsigned int nr_states,
 	dev->em_pd->min_perf_state = 0;
 	dev->em_pd->max_perf_state = nr_states - 1;
 
-	em_cpufreq_update_efficiencies(dev, dev->em_pd->em_table->state);
+	em_table = rcu_dereference_protected(dev->em_pd->em_table,
+					     lockdep_is_held(&em_pd_mutex));
+	em_cpufreq_update_efficiencies(dev, em_table->state);
 
 	em_debug_create_pd(dev);
 	dev_info(dev, "EM: created perf domain\n");
@@ -668,17 +661,18 @@ void em_dev_unregister_perf_domain(struct device *dev)
 	mutex_lock(&em_pd_mutex);
 	em_debug_remove_pd(dev);
 
-	em_table_free(dev->em_pd->em_table);
+	em_table_free(rcu_dereference_protected(dev->em_pd->em_table,
+						lockdep_is_held(&em_pd_mutex)));
 
 	kfree(dev->em_pd);
 	dev->em_pd = NULL;
 	mutex_unlock(&em_pd_mutex);
 }
 EXPORT_SYMBOL_GPL(em_dev_unregister_perf_domain);
 
-static struct em_perf_table __rcu *em_table_dup(struct em_perf_domain *pd)
+static struct em_perf_table *em_table_dup(struct em_perf_domain *pd)
 {
-	struct em_perf_table __rcu *em_table;
+	struct em_perf_table *em_table;
 	struct em_perf_state *ps, *new_ps;
 	int ps_size;
 
@@ -700,7 +694,7 @@ static struct em_perf_table __rcu *em_table_dup(struct em_perf_domain *pd)
 }
 
 static int em_recalc_and_update(struct device *dev, struct em_perf_domain *pd,
-				struct em_perf_table __rcu *em_table)
+				struct em_perf_table *em_table)
 {
 	int ret;
 
@@ -728,10 +722,9 @@ static int em_recalc_and_update(struct device *dev, struct em_perf_domain *pd,
  * are correctly calculated.
  */
 static void em_adjust_new_capacity(struct device *dev,
-				   struct em_perf_domain *pd,
-				   u64 max_cap)
+				   struct em_perf_domain *pd)
 {
-	struct em_perf_table __rcu *em_table;
+	struct em_perf_table *em_table;
 
 	em_table = em_table_dup(pd);
 	if (!em_table) {
@@ -775,7 +768,8 @@ static void em_check_capacity_update(void)
 		}
 		cpufreq_cpu_put(policy);
 
-		pd = em_cpu_get(cpu);
+		dev = get_cpu_device(cpu);
+		pd = em_pd_get(dev);
 		if (!pd || em_is_artificial(pd))
 			continue;
 
@@ -799,8 +793,7 @@ static void em_check_capacity_update(void)
 		pr_debug("updating cpu%d cpu_cap=%lu old capacity=%lu\n",
 			 cpu, cpu_capacity, em_max_perf);
 
-		dev = get_cpu_device(cpu);
-		em_adjust_new_capacity(dev, pd, cpu_capacity);
+		em_adjust_new_capacity(dev, pd);
 	}
 
 	free_cpumask_var(cpu_done_mask);
@@ -822,7 +815,7 @@ static void em_update_workfn(struct work_struct *work)
  */
 int em_dev_update_chip_binning(struct device *dev)
 {
-	struct em_perf_table __rcu *em_table;
+	struct em_perf_table *em_table;
 	struct em_perf_domain *pd;
 	int i, ret;
 
