Merge tag 'pm-4.21-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "These add sysadmin documentation for cpuidle, extend the cpuidle
  subsystem somewhat, improve the handling of performance states in the
  generic power domains (genpd) and operating performance points (OPP)
  frameworks, add a new cpufreq driver for Qualcomm SoCs, update some
  other cpufreq drivers, switch over the runtime PM framework to using
  high-res timers for device autosuspend, fix a problem with
  suspend-to-idle on ACPI-based platforms, add system-wide suspend and
  resume handling to the devfreq framework, do some janitorial cleanups
  all over and update some utilities.

  Specifics:

   - Add sysadmin documentation for cpuidle (Rafael Wysocki).

   - Make it possible to specify a cpuidle governor from kernel command
     line, add new cpuidle state sysfs attributes for governor
     evaluation, and improve the "polling" idle state handling (Rafael
     Wysocki).

   - Fix the handling of the "required-opps" DT property in the
     operating performance points (OPP) framework, improve the
     integration of it with the generic power domains (genpd) framework,
     improve the handling of performance states in them and clean up the
     idle states vs performance states separation in genpd (Viresh
     Kumar, Ulf Hansson).

   - Add a cpufreq driver called "qcom-hw" for Qualcomm SoCs using a
     hardware engine to control CPU frequency transitions along with DT
     bindings for it (Taniya Das).

   - Fix an intel_pstate driver issue related to CPU offline and update
     the documentation of it (Srinivas Pandruvada).

   - Clean up the imx6q cpufreq driver (Anson Huang).

   - Add SPDX license IDs to cpufreq schedutil governor files (Daniel
     Lezcano).

   - Switch over the runtime PM framework to using high-res timers for
     device autosuspend to allow the control of it to be more precise
     (Vincent Guittot).

   - Disable non-wakeup ACPI GPEs during suspend-to-idle so that they
     don't prevent the system from reaching the target low-power state
     and simplify the suspend-to-idle handling on ACPI platforms without
     full Low-Power S0 Idle (LPS0) support (Rafael Wysocki).

   - Add system-wide suspend and resume support to the devfreq framework
     (Lukasz Luba).

   - Clean up the SmartReflex adaptive voltage scaling (AVS) driver and
     add an SPDX license ID to it (Nishanth Menon, Uwe Kleine-König,
     Thomas Meyer).

   - Get rid of code duplication by using the DEFINE_SHOW_ATTRIBUTE
     macro in some places, fix some DT node refcount leaks, and do some
     other janitorial cleanups (Yangtao Li).

   - Update the cpupower, intel_pstate_tracer and turbosat utilities
     (Abhishek Goel, Doug Smythies, Len Brown)"

* tag 'pm-4.21-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (54 commits)
  PM / Domains: remove define_genpd_open_function() and define_genpd_debugfs_fops()
  PM-runtime: Switch autosuspend over to using hrtimers
  cpufreq: qcom-hw: Add support for QCOM cpufreq HW driver
  dt-bindings: cpufreq: Introduce QCOM cpufreq firmware bindings
  ACPI: PM: Loop in full LPS0 mode only
  ACPI: EC / PM: Disable non-wakeup GPEs for suspend-to-idle
  tools/power/x86/intel_pstate_tracer: Fix non root execution for post processing a trace file
  tools/power turbostat: consolidate duplicate model numbers
  tools/power turbostat: fix goldmont C-state limit decoding
  PM / Domains: Propagate performance state updates
  PM / Domains: Factorize dev_pm_genpd_set_performance_state()
  PM / Domains: Save OPP table pointer in genpd
  OPP: Don't return 0 on error from of_get_required_opp_performance_state()
  OPP: Add dev_pm_opp_xlate_performance_state() helper
  OPP: Improve _find_table_of_opp_np()
  PM / Domains: Make genpd performance states orthogonal to the idlestates
  PM / sleep: convert to DEFINE_SHOW_ATTRIBUTE
  cpuidle: Add 'above' and 'below' idle state metrics
  PM / AVS: SmartReflex: Switch to SPDX Licence ID
  PM / AVS: SmartReflex: NULL check before some freeing functions is not needed
  ...

2 files changed, 233 insertions, 173 deletions

diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c
index 7f38a92b444a..500de1dee967 100644
--- a/drivers/base/power/domain.c
+++ b/drivers/base/power/domain.c
@@ -239,6 +239,127 @@ static void genpd_update_accounting(struct generic_pm_domain *genpd)
 static inline void genpd_update_accounting(struct generic_pm_domain *genpd) {}
 #endif
 
+static int _genpd_reeval_performance_state(struct generic_pm_domain *genpd,
+					   unsigned int state)
+{
+	struct generic_pm_domain_data *pd_data;
+	struct pm_domain_data *pdd;
+	struct gpd_link *link;
+
+	/* New requested state is same as Max requested state */
+	if (state == genpd->performance_state)
+		return state;
+
+	/* New requested state is higher than Max requested state */
+	if (state > genpd->performance_state)
+		return state;
+
+	/* Traverse all devices within the domain */
+	list_for_each_entry(pdd, &genpd->dev_list, list_node) {
+		pd_data = to_gpd_data(pdd);
+
+		if (pd_data->performance_state > state)
+			state = pd_data->performance_state;
+	}
+
+	/*
+	 * Traverse all sub-domains within the domain. This can be
+	 * done without any additional locking as the link->performance_state
+	 * field is protected by the master genpd->lock, which is already taken.
+	 *
+	 * Also note that link->performance_state (subdomain's performance state
+	 * requirement to master domain) is different from
+	 * link->slave->performance_state (current performance state requirement
+	 * of the devices/sub-domains of the subdomain) and so can have a
+	 * different value.
+	 *
+	 * Note that we also take vote from powered-off sub-domains into account
+	 * as the same is done for devices right now.
+	 */
+	list_for_each_entry(link, &genpd->master_links, master_node) {
+		if (link->performance_state > state)
+			state = link->performance_state;
+	}
+
+	return state;
+}
+
+static int _genpd_set_performance_state(struct generic_pm_domain *genpd,
+					unsigned int state, int depth)
+{
+	struct generic_pm_domain *master;
+	struct gpd_link *link;
+	int master_state, ret;
+
+	if (state == genpd->performance_state)
+		return 0;
+
+	/* Propagate to masters of genpd */
+	list_for_each_entry(link, &genpd->slave_links, slave_node) {
+		master = link->master;
+
+		if (!master->set_performance_state)
+			continue;
+
+		/* Find master's performance state */
+		ret = dev_pm_opp_xlate_performance_state(genpd->opp_table,
+							 master->opp_table,
+							 state);
+		if (unlikely(ret < 0))
+			goto err;
+
+		master_state = ret;
+
+		genpd_lock_nested(master, depth + 1);
+
+		link->prev_performance_state = link->performance_state;
+		link->performance_state = master_state;
+		master_state = _genpd_reeval_performance_state(master,
+						master_state);
+		ret = _genpd_set_performance_state(master, master_state, depth + 1);
+		if (ret)
+			link->performance_state = link->prev_performance_state;
+
+		genpd_unlock(master);
+
+		if (ret)
+			goto err;
+	}
+
+	ret = genpd->set_performance_state(genpd, state);
+	if (ret)
+		goto err;
+
+	genpd->performance_state = state;
+	return 0;
+
+err:
+	/* Encountered an error, lets rollback */
+	list_for_each_entry_continue_reverse(link, &genpd->slave_links,
+					     slave_node) {
+		master = link->master;
+
+		if (!master->set_performance_state)
+			continue;
+
+		genpd_lock_nested(master, depth + 1);
+
+		master_state = link->prev_performance_state;
+		link->performance_state = master_state;
+
+		master_state = _genpd_reeval_performance_state(master,
+						master_state);
+		if (_genpd_set_performance_state(master, master_state, depth + 1)) {
+			pr_err("%s: Failed to roll back to %d performance state\n",
+			       master->name, master_state);
+		}
+
+		genpd_unlock(master);
+	}
+
+	return ret;
+}
+
 /**
  * dev_pm_genpd_set_performance_state- Set performance state of device's power
  * domain.
@@ -257,10 +378,9 @@ static inline void genpd_update_accounting(struct generic_pm_domain *genpd) {}
 int dev_pm_genpd_set_performance_state(struct device *dev, unsigned int state)
 {
 	struct generic_pm_domain *genpd;
-	struct generic_pm_domain_data *gpd_data, *pd_data;
-	struct pm_domain_data *pdd;
+	struct generic_pm_domain_data *gpd_data;
 	unsigned int prev;
-	int ret = 0;
+	int ret;
 
 	genpd = dev_to_genpd(dev);
 	if (IS_ERR(genpd))
@@ -281,47 +401,11 @@ int dev_pm_genpd_set_performance_state(struct device *dev, unsigned int state)
 	prev = gpd_data->performance_state;
 	gpd_data->performance_state = state;
 
-	/* New requested state is same as Max requested state */
-	if (state == genpd->performance_state)
-		goto unlock;
-
-	/* New requested state is higher than Max requested state */
-	if (state > genpd->performance_state)
-		goto update_state;
-
-	/* Traverse all devices within the domain */
-	list_for_each_entry(pdd, &genpd->dev_list, list_node) {
-		pd_data = to_gpd_data(pdd);
-
-		if (pd_data->performance_state > state)
-			state = pd_data->performance_state;
-	}
-
-	if (state == genpd->performance_state)
-		goto unlock;
-
-	/*
-	 * We aren't propagating performance state changes of a subdomain to its
-	 * masters as we don't have hardware that needs it. Over that, the
-	 * performance states of subdomain and its masters may not have
-	 * one-to-one mapping and would require additional information. We can
-	 * get back to this once we have hardware that needs it. For that
-	 * reason, we don't have to consider performance state of the subdomains
-	 * of genpd here.
-	 */
-
-update_state:
-	if (genpd_status_on(genpd)) {
-		ret = genpd->set_performance_state(genpd, state);
-		if (ret) {
-			gpd_data->performance_state = prev;
-			goto unlock;
-		}
-	}
-
-	genpd->performance_state = state;
+	state = _genpd_reeval_performance_state(genpd, state);
+	ret = _genpd_set_performance_state(genpd, state, 0);
+	if (ret)
+		gpd_data->performance_state = prev;
 
-unlock:
 	genpd_unlock(genpd);
 
 	return ret;
@@ -347,15 +431,6 @@ static int _genpd_power_on(struct generic_pm_domain *genpd, bool timed)
 		return ret;
 
 	elapsed_ns = ktime_to_ns(ktime_sub(ktime_get(), time_start));
-
-	if (unlikely(genpd->set_performance_state)) {
-		ret = genpd->set_performance_state(genpd, genpd->performance_state);
-		if (ret) {
-			pr_warn("%s: Failed to set performance state %d (%d)\n",
-				genpd->name, genpd->performance_state, ret);
-		}
-	}
-
 	if (elapsed_ns <= genpd->states[state_idx].power_on_latency_ns)
 		return ret;
 
@@ -1907,12 +1982,21 @@ int of_genpd_add_provider_simple(struct device_node *np,
 				ret);
 			goto unlock;
 		}
+
+		/*
+		 * Save table for faster processing while setting performance
+		 * state.
+		 */
+		genpd->opp_table = dev_pm_opp_get_opp_table(&genpd->dev);
+		WARN_ON(!genpd->opp_table);
 	}
 
 	ret = genpd_add_provider(np, genpd_xlate_simple, genpd);
 	if (ret) {
-		if (genpd->set_performance_state)
+		if (genpd->set_performance_state) {
+			dev_pm_opp_put_opp_table(genpd->opp_table);
 			dev_pm_opp_of_remove_table(&genpd->dev);
+		}
 
 		goto unlock;
 	}
@@ -1965,6 +2049,13 @@ int of_genpd_add_provider_onecell(struct device_node *np,
 					i, ret);
 				goto error;
 			}
+
+			/*
+			 * Save table for faster processing while setting
+			 * performance state.
+			 */
+			genpd->opp_table = dev_pm_opp_get_opp_table_indexed(&genpd->dev, i);
+			WARN_ON(!genpd->opp_table);
 		}
 
 		genpd->provider = &np->fwnode;
@@ -1989,8 +2080,10 @@ error:
 		genpd->provider = NULL;
 		genpd->has_provider = false;
 
-		if (genpd->set_performance_state)
+		if (genpd->set_performance_state) {
+			dev_pm_opp_put_opp_table(genpd->opp_table);
 			dev_pm_opp_of_remove_table(&genpd->dev);
+		}
 	}
 
 	mutex_unlock(&gpd_list_lock);
@@ -2024,6 +2117,7 @@ void of_genpd_del_provider(struct device_node *np)
 					if (!gpd->set_performance_state)
 						continue;
 
+					dev_pm_opp_put_opp_table(gpd->opp_table);
 					dev_pm_opp_of_remove_table(&gpd->dev);
 				}
 			}
@@ -2338,7 +2432,7 @@ EXPORT_SYMBOL_GPL(genpd_dev_pm_attach);
 struct device *genpd_dev_pm_attach_by_id(struct device *dev,
 					 unsigned int index)
 {
-	struct device *genpd_dev;
+	struct device *virt_dev;
 	int num_domains;
 	int ret;
 
@@ -2352,31 +2446,31 @@ struct device *genpd_dev_pm_attach_by_id(struct device *dev,
 		return NULL;
 
 	/* Allocate and register device on the genpd bus. */
-	genpd_dev = kzalloc(sizeof(*genpd_dev), GFP_KERNEL);
-	if (!genpd_dev)
+	virt_dev = kzalloc(sizeof(*virt_dev), GFP_KERNEL);
+	if (!virt_dev)
 		return ERR_PTR(-ENOMEM);
 
-	dev_set_name(genpd_dev, "genpd:%u:%s", index, dev_name(dev));
-	genpd_dev->bus = &genpd_bus_type;
-	genpd_dev->release = genpd_release_dev;
+	dev_set_name(virt_dev, "genpd:%u:%s", index, dev_name(dev));
+	virt_dev->bus = &genpd_bus_type;
+	virt_dev->release = genpd_release_dev;
 
-	ret = device_register(genpd_dev);
+	ret = device_register(virt_dev);
 	if (ret) {
-		kfree(genpd_dev);
+		kfree(virt_dev);
 		return ERR_PTR(ret);
 	}
 
 	/* Try to attach the device to the PM domain at the specified index. */
-	ret = __genpd_dev_pm_attach(genpd_dev, dev->of_node, index, false);
+	ret = __genpd_dev_pm_attach(virt_dev, dev->of_node, index, false);
 	if (ret < 1) {
-		device_unregister(genpd_dev);
+		device_unregister(virt_dev);
 		return ret ? ERR_PTR(ret) : NULL;
 	}
 
-	pm_runtime_enable(genpd_dev);
-	genpd_queue_power_off_work(dev_to_genpd(genpd_dev));
+	pm_runtime_enable(virt_dev);
+	genpd_queue_power_off_work(dev_to_genpd(virt_dev));
 
-	return genpd_dev;
+	return virt_dev;
 }
 EXPORT_SYMBOL_GPL(genpd_dev_pm_attach_by_id);
 
@@ -2521,52 +2615,36 @@ int of_genpd_parse_idle_states(struct device_node *dn,
 EXPORT_SYMBOL_GPL(of_genpd_parse_idle_states);
 
 /**
- * of_genpd_opp_to_performance_state- Gets performance state of device's
- * power domain corresponding to a DT node's "required-opps" property.
+ * pm_genpd_opp_to_performance_state - Gets performance state of the genpd from its OPP node.
  *
- * @dev: Device for which the performance-state needs to be found.
- * @np: DT node where the "required-opps" property is present. This can be
- *	the device node itself (if it doesn't have an OPP table) or a node
- *	within the OPP table of a device (if device has an OPP table).
+ * @genpd_dev: Genpd's device for which the performance-state needs to be found.
+ * @opp: struct dev_pm_opp of the OPP for which we need to find performance
+ *	state.
  *
- * Returns performance state corresponding to the "required-opps" property of
- * a DT node. This calls platform specific genpd->opp_to_performance_state()
- * callback to translate power domain OPP to performance state.
+ * Returns performance state encoded in the OPP of the genpd. This calls
+ * platform specific genpd->opp_to_performance_state() callback to translate
+ * power domain OPP to performance state.
  *
  * Returns performance state on success and 0 on failure.
  */
-unsigned int of_genpd_opp_to_performance_state(struct device *dev,
-					       struct device_node *np)
+unsigned int pm_genpd_opp_to_performance_state(struct device *genpd_dev,
+					       struct dev_pm_opp *opp)
 {
-	struct generic_pm_domain *genpd;
-	struct dev_pm_opp *opp;
-	int state = 0;
+	struct generic_pm_domain *genpd = NULL;
+	int state;
 
-	genpd = dev_to_genpd(dev);
-	if (IS_ERR(genpd))
-		return 0;
+	genpd = container_of(genpd_dev, struct generic_pm_domain, dev);
 
-	if (unlikely(!genpd->set_performance_state))
+	if (unlikely(!genpd->opp_to_performance_state))
 		return 0;
 
 	genpd_lock(genpd);
-
-	opp = of_dev_pm_opp_find_required_opp(&genpd->dev, np);
-	if (IS_ERR(opp)) {
-		dev_err(dev, "Failed to find required OPP: %ld\n",
-			PTR_ERR(opp));
-		goto unlock;
-	}
-
 	state = genpd->opp_to_performance_state(genpd, opp);
-	dev_pm_opp_put(opp);
-
-unlock:
 	genpd_unlock(genpd);
 
 	return state;
 }
-EXPORT_SYMBOL_GPL(of_genpd_opp_to_performance_state);
+EXPORT_SYMBOL_GPL(pm_genpd_opp_to_performance_state);
 
 static int __init genpd_bus_init(void)
 {
@@ -2671,7 +2749,7 @@ exit:
 	return 0;
 }
 
-static int genpd_summary_show(struct seq_file *s, void *data)
+static int summary_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd;
 	int ret = 0;
@@ -2694,7 +2772,7 @@ static int genpd_summary_show(struct seq_file *s, void *data)
 	return ret;
 }
 
-static int genpd_status_show(struct seq_file *s, void *data)
+static int status_show(struct seq_file *s, void *data)
 {
 	static const char * const status_lookup[] = {
 		[GPD_STATE_ACTIVE] = "on",
@@ -2721,7 +2799,7 @@ exit:
 	return ret;
 }
 
-static int genpd_sub_domains_show(struct seq_file *s, void *data)
+static int sub_domains_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd = s->private;
 	struct gpd_link *link;
@@ -2738,7 +2816,7 @@ static int genpd_sub_domains_show(struct seq_file *s, void *data)
 	return ret;
 }
 
-static int genpd_idle_states_show(struct seq_file *s, void *data)
+static int idle_states_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd = s->private;
 	unsigned int i;
@@ -2767,7 +2845,7 @@ static int genpd_idle_states_show(struct seq_file *s, void *data)
 	return ret;
 }
 
-static int genpd_active_time_show(struct seq_file *s, void *data)
+static int active_time_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd = s->private;
 	ktime_t delta = 0;
@@ -2787,7 +2865,7 @@ static int genpd_active_time_show(struct seq_file *s, void *data)
 	return ret;
 }
 
-static int genpd_total_idle_time_show(struct seq_file *s, void *data)
+static int total_idle_time_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd = s->private;
 	ktime_t delta = 0, total = 0;
@@ -2815,7 +2893,7 @@ static int genpd_total_idle_time_show(struct seq_file *s, void *data)
 }
 
 
-static int genpd_devices_show(struct seq_file *s, void *data)
+static int devices_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd = s->private;
 	struct pm_domain_data *pm_data;
@@ -2841,7 +2919,7 @@ static int genpd_devices_show(struct seq_file *s, void *data)
 	return ret;
 }
 
-static int genpd_perf_state_show(struct seq_file *s, void *data)
+static int perf_state_show(struct seq_file *s, void *data)
 {
 	struct generic_pm_domain *genpd = s->private;
 
@@ -2854,37 +2932,14 @@ static int genpd_perf_state_show(struct seq_file *s, void *data)
 	return 0;
 }
 
-#define define_genpd_open_function(name) \
-static int genpd_##name##_open(struct inode *inode, struct file *file) \
-{ \
-	return single_open(file, genpd_##name##_show, inode->i_private); \
-}
-
-define_genpd_open_function(summary);
-define_genpd_open_function(status);
-define_genpd_open_function(sub_domains);
-define_genpd_open_function(idle_states);
-define_genpd_open_function(active_time);
-define_genpd_open_function(total_idle_time);
-define_genpd_open_function(devices);
-define_genpd_open_function(perf_state);
-
-#define define_genpd_debugfs_fops(name) \
-static const struct file_operations genpd_##name##_fops = { \
-	.open = genpd_##name##_open, \
-	.read = seq_read, \
-	.llseek = seq_lseek, \
-	.release = single_release, \
-}
-
-define_genpd_debugfs_fops(summary);
-define_genpd_debugfs_fops(status);
-define_genpd_debugfs_fops(sub_domains);
-define_genpd_debugfs_fops(idle_states);
-define_genpd_debugfs_fops(active_time);
-define_genpd_debugfs_fops(total_idle_time);
-define_genpd_debugfs_fops(devices);
-define_genpd_debugfs_fops(perf_state);
+DEFINE_SHOW_ATTRIBUTE(summary);
+DEFINE_SHOW_ATTRIBUTE(status);
+DEFINE_SHOW_ATTRIBUTE(sub_domains);
+DEFINE_SHOW_ATTRIBUTE(idle_states);
+DEFINE_SHOW_ATTRIBUTE(active_time);
+DEFINE_SHOW_ATTRIBUTE(total_idle_time);
+DEFINE_SHOW_ATTRIBUTE(devices);
+DEFINE_SHOW_ATTRIBUTE(perf_state);
 
 static int __init genpd_debug_init(void)
 {
@@ -2897,7 +2952,7 @@ static int __init genpd_debug_init(void)
 		return -ENOMEM;
 
 	d = debugfs_create_file("pm_genpd_summary", S_IRUGO,
-			genpd_debugfs_dir, NULL, &genpd_summary_fops);
+			genpd_debugfs_dir, NULL, &summary_fops);
 	if (!d)
 		return -ENOMEM;
 
@@ -2907,20 +2962,20 @@ static int __init genpd_debug_init(void)
 			return -ENOMEM;
 
 		debugfs_create_file("current_state", 0444,
-				d, genpd, &genpd_status_fops);
+				d, genpd, &status_fops);
 		debugfs_create_file("sub_domains", 0444,
-				d, genpd, &genpd_sub_domains_fops);
+				d, genpd, &sub_domains_fops);
 		debugfs_create_file("idle_states", 0444,
-				d, genpd, &genpd_idle_states_fops);
+				d, genpd, &idle_states_fops);
 		debugfs_create_file("active_time", 0444,
-				d, genpd, &genpd_active_time_fops);
+				d, genpd, &active_time_fops);
 		debugfs_create_file("total_idle_time", 0444,
-				d, genpd, &genpd_total_idle_time_fops);
+				d, genpd, &total_idle_time_fops);
 		debugfs_create_file("devices", 0444,
-				d, genpd, &genpd_devices_fops);
+				d, genpd, &devices_fops);
 		if (genpd->set_performance_state)
 			debugfs_create_file("perf_state", 0444,
-					    d, genpd, &genpd_perf_state_fops);
+					    d, genpd, &perf_state_fops);
 	}
 
 	return 0;
diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
index beb85c31f3fa..70624695b6d5 100644
--- a/drivers/base/power/runtime.c
+++ b/drivers/base/power/runtime.c
@@ -8,6 +8,8 @@
  */
 
 #include <linux/sched/mm.h>
+#include <linux/ktime.h>
+#include <linux/hrtimer.h>
 #include <linux/export.h>
 #include <linux/pm_runtime.h>
 #include <linux/pm_wakeirq.h>
@@ -93,7 +95,7 @@ static void __update_runtime_status(struct device *dev, enum rpm_status status)
 static void pm_runtime_deactivate_timer(struct device *dev)
 {
 	if (dev->power.timer_expires > 0) {
-		del_timer(&dev->power.suspend_timer);
+		hrtimer_cancel(&dev->power.suspend_timer);
 		dev->power.timer_expires = 0;
 	}
 }
@@ -124,12 +126,11 @@ static void pm_runtime_cancel_pending(struct device *dev)
  * This function may be called either with or without dev->power.lock held.
  * Either way it can be racy, since power.last_busy may be updated at any time.
  */
-unsigned long pm_runtime_autosuspend_expiration(struct device *dev)
+u64 pm_runtime_autosuspend_expiration(struct device *dev)
 {
 	int autosuspend_delay;
-	long elapsed;
-	unsigned long last_busy;
-	unsigned long expires = 0;
+	u64 last_busy, expires = 0;
+	u64 now = ktime_to_ns(ktime_get());
 
 	if (!dev->power.use_autosuspend)
 		goto out;
@@ -139,19 +140,9 @@ unsigned long pm_runtime_autosuspend_expiration(struct device *dev)
 		goto out;
 
 	last_busy = READ_ONCE(dev->power.last_busy);
-	elapsed = jiffies - last_busy;
-	if (elapsed < 0)
-		goto out;	/* jiffies has wrapped around. */
 
-	/*
-	 * If the autosuspend_delay is >= 1 second, align the timer by rounding
-	 * up to the nearest second.
-	 */
-	expires = last_busy + msecs_to_jiffies(autosuspend_delay);
-	if (autosuspend_delay >= 1000)
-		expires = round_jiffies(expires);
-	expires += !expires;
-	if (elapsed >= expires - last_busy)
+	expires = last_busy + autosuspend_delay * NSEC_PER_MSEC;
+	if (expires <= now)
 		expires = 0;	/* Already expired. */
 
  out:
@@ -515,7 +506,7 @@ static int rpm_suspend(struct device *dev, int rpmflags)
 	/* If the autosuspend_delay time hasn't expired yet, reschedule. */
 	if ((rpmflags & RPM_AUTO)
 	    && dev->power.runtime_status != RPM_SUSPENDING) {
-		unsigned long expires = pm_runtime_autosuspend_expiration(dev);
+		u64 expires = pm_runtime_autosuspend_expiration(dev);
 
 		if (expires != 0) {
 			/* Pending requests need to be canceled. */
@@ -528,10 +519,20 @@ static int rpm_suspend(struct device *dev, int rpmflags)
 			 * expire; pm_suspend_timer_fn() will take care of the
 			 * rest.
 			 */
-			if (!(dev->power.timer_expires && time_before_eq(
-			    dev->power.timer_expires, expires))) {
+			if (!(dev->power.timer_expires &&
+					dev->power.timer_expires <= expires)) {
+				/*
+				 * We add a slack of 25% to gather wakeups
+				 * without sacrificing the granularity.
+				 */
+				u64 slack = READ_ONCE(dev->power.autosuspend_delay) *
+						    (NSEC_PER_MSEC >> 2);
+
 				dev->power.timer_expires = expires;
-				mod_timer(&dev->power.suspend_timer, expires);
+				hrtimer_start_range_ns(&dev->power.suspend_timer,
+						ns_to_ktime(expires),
+						slack,
+						HRTIMER_MODE_ABS);
 			}
 			dev->power.timer_autosuspends = 1;
 			goto out;
@@ -895,23 +896,25 @@ static void pm_runtime_work(struct work_struct *work)
  *
  * Check if the time is right and queue a suspend request.
  */
-static void pm_suspend_timer_fn(struct timer_list *t)
+static enum hrtimer_restart  pm_suspend_timer_fn(struct hrtimer *timer)
 {
-	struct device *dev = from_timer(dev, t, power.suspend_timer);
+	struct device *dev = container_of(timer, struct device, power.suspend_timer);
 	unsigned long flags;
-	unsigned long expires;
+	u64 expires;
 
 	spin_lock_irqsave(&dev->power.lock, flags);
 
 	expires = dev->power.timer_expires;
 	/* If 'expire' is after 'jiffies' we've been called too early. */
-	if (expires > 0 && !time_after(expires, jiffies)) {
+	if (expires > 0 && expires < ktime_to_ns(ktime_get())) {
 		dev->power.timer_expires = 0;
 		rpm_suspend(dev, dev->power.timer_autosuspends ?
 		    (RPM_ASYNC | RPM_AUTO) : RPM_ASYNC);
 	}
 
 	spin_unlock_irqrestore(&dev->power.lock, flags);
+
+	return HRTIMER_NORESTART;
 }
 
 /**
@@ -922,6 +925,7 @@ static void pm_suspend_timer_fn(struct timer_list *t)
 int pm_schedule_suspend(struct device *dev, unsigned int delay)
 {
 	unsigned long flags;
+	ktime_t expires;
 	int retval;
 
 	spin_lock_irqsave(&dev->power.lock, flags);
@@ -938,10 +942,10 @@ int pm_schedule_suspend(struct device *dev, unsigned int delay)
 	/* Other scheduled or pending requests need to be canceled. */
 	pm_runtime_cancel_pending(dev);
 
-	dev->power.timer_expires = jiffies + msecs_to_jiffies(delay);
-	dev->power.timer_expires += !dev->power.timer_expires;
+	expires = ktime_add(ktime_get(), ms_to_ktime(delay));
+	dev->power.timer_expires = ktime_to_ns(expires);
 	dev->power.timer_autosuspends = 0;
-	mod_timer(&dev->power.suspend_timer, dev->power.timer_expires);
+	hrtimer_start(&dev->power.suspend_timer, expires, HRTIMER_MODE_ABS);
 
  out:
 	spin_unlock_irqrestore(&dev->power.lock, flags);
@@ -1491,7 +1495,8 @@ void pm_runtime_init(struct device *dev)
 	INIT_WORK(&dev->power.work, pm_runtime_work);
 
 	dev->power.timer_expires = 0;
-	timer_setup(&dev->power.suspend_timer, pm_suspend_timer_fn, 0);
+	hrtimer_init(&dev->power.suspend_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS);
+	dev->power.suspend_timer.function = pm_suspend_timer_fn;
 
 	init_waitqueue_head(&dev->power.wait_queue);
 }