Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit 386d4f46 authored by Rohit Gupta's avatar Rohit Gupta
Browse files

PM / devfreq: Handle offline CPUs while creating events 



If any of the CPUs monitored by a memlat device is offline while
trying to register for perf events, the registration fails
returning an error code to the governor's start_monitor function
which then fails the setting of governor and leaves the system in
a bad state.
To avoid this scenario, this patch allows for initialization errors
while registering for perf events and remembers the failing offline
CPUs. Perf event registration is tried for the failing CPUs when they
come back online. Once all the CPUs have registered for perf events
hotplug notifiers callbacks are unregistered.

Change-Id: I36c33d9dc627f60ab9d7db41913661927938e850
Signed-off-by: default avatarRohit Gupta <rohgup@codeaurora.org>
parent 5684450d

drivers/devfreq/arm-memlat-mon.c

+85 −10
Original line number Diff line number Diff line
@@ -45,12 +45,12 @@ enum ev_index { 
struct event_data {

	struct perf_event *pevent;

	unsigned long prev_count;

	bool overflow;

};



struct memlat_hwmon_data {

	struct event_data events[NUM_EVENTS];

	ktime_t prev_ts;

	bool init_pending;

};

static DEFINE_PER_CPU(struct memlat_hwmon_data, pm_data);
@@ -59,6 +59,12 @@ struct cpu_grp_info { 
	struct memlat_hwmon hw;

};



static DEFINE_MUTEX(notif_mutex);

static unsigned int notif_refcount;

static void notif_unregister_work_fn(struct work_struct *work);

static void hotplug_notif_unregister(void);

static DECLARE_WORK(notif_unregister_work, notif_unregister_work_fn);



static unsigned long compute_freq(struct memlat_hwmon_data *hw_data,

						unsigned long cyc_cnt)

{
@@ -101,6 +107,9 @@ static void read_perf_counters(int cpu, struct cpu_grp_info *cpu_grp) 
	struct memlat_hwmon *hw = &cpu_grp->hw;

	unsigned long cyc_cnt;



	if (hw_data->init_pending)

		return;



	cpu_idx = cpu - cpumask_first(&cpu_grp->cpus);



	hw->core_stats[cpu_idx].inst_count =
@@ -127,14 +136,17 @@ static unsigned long get_cnt(struct memlat_hwmon *hw) 


static void delete_events(struct memlat_hwmon_data *hw_data)

{

	perf_event_release_kernel(hw_data->events[INST_IDX].pevent);

	perf_event_release_kernel(hw_data->events[L2DM_IDX].pevent);

	perf_event_release_kernel(hw_data->events[CYC_IDX].pevent);

	int i;



	for (i = 0; i < NUM_EVENTS; i++) {

		hw_data->events[i].prev_count = 0;

		perf_event_release_kernel(hw_data->events[i].pevent);

	}

}



static void stop_hwmon(struct memlat_hwmon *hw)

{

	int cpu;

	int cpu, idx;

	struct memlat_hwmon_data *hw_data;

	struct cpu_grp_info *cpu_grp = container_of(hw,

					struct cpu_grp_info, hw);
@@ -142,7 +154,16 @@ static void stop_hwmon(struct memlat_hwmon *hw) 
	get_online_cpus();

	for_each_cpu(cpu, &cpu_grp->cpus) {

		hw_data = &per_cpu(pm_data, cpu);

		if (hw_data->init_pending)

			hotplug_notif_unregister();

		else

			delete_events(hw_data);



		/* Clear governor data */

		idx = cpu - cpumask_first(&cpu_grp->cpus);

		hw->core_stats[idx].inst_count = 0;

		hw->core_stats[idx].mem_count = 0;

		hw->core_stats[idx].freq = 0;

	}

	put_online_cpus();
@@ -198,6 +219,53 @@ err_out: 
	return err;

}



static int arm_memlat_cpu_callback(struct notifier_block *nb,

		unsigned long action, void *hcpu)

{

	unsigned long cpu = (unsigned long)hcpu;

	struct memlat_hwmon_data *hw_data = &per_cpu(pm_data, cpu);



	if ((action != CPU_ONLINE) || !hw_data->init_pending)

		return NOTIFY_OK;



	if (set_events(hw_data, cpu))

		pr_warn("Failed to create perf event for CPU%lu\n", cpu);



	hw_data->init_pending = false;

	hotplug_notif_unregister();



	return NOTIFY_OK;

}



static struct notifier_block arm_memlat_cpu_notifier = {

	.notifier_call = arm_memlat_cpu_callback,

};



static void notif_unregister_work_fn(struct work_struct *work)

{

	unregister_cpu_notifier(&arm_memlat_cpu_notifier);

}



static void hotplug_notif_register(void)

{

	mutex_lock(&notif_mutex);

	if (!notif_refcount) {

		flush_work(&notif_unregister_work);

		register_cpu_notifier(&arm_memlat_cpu_notifier);

	}

	notif_refcount++;

	mutex_unlock(&notif_mutex);

}



static void hotplug_notif_unregister(void)

{

	mutex_lock(&notif_mutex);

	notif_refcount--;

	if (!notif_refcount)

		schedule_work(&notif_unregister_work);

	mutex_unlock(&notif_mutex);

}



static int start_hwmon(struct memlat_hwmon *hw)

{

	int cpu, ret = 0;
@@ -209,12 +277,19 @@ static int start_hwmon(struct memlat_hwmon *hw) 
	for_each_cpu(cpu, &cpu_grp->cpus) {

		hw_data = &per_cpu(pm_data, cpu);

		ret = set_events(hw_data, cpu);

		if (ret)

			goto err;

		if (ret) {

			if (!cpu_online(cpu)) {

				hotplug_notif_register();

				hw_data->init_pending = true;

				ret = 0;

			} else {

				pr_warn("Perf event init failed on CPU%d\n",

					cpu);

				break;

			}

		}

	}

	pr_info("Group %d events created\n", cpumask_first(&cpu_grp->cpus));



err:

	put_online_cpus();

	return ret;

}