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

Commit 2d02e0aa authored by Ram Chandrasekar's avatar Ram Chandrasekar
Browse files

drivers: thermal: cpu-cooling: Avoid deadlock 



There is a possibility of deadlock when hotplug is triggered
externally. The hotplug path is waiting on core isolation mutex and
simultaneously if thermal does core isolation, it would have acquired
the core isolation lock and will wait on the hotplug lock.

Core isolation lock is used to protect the current mitigation state
variable and it is unnecessary to hold it when trying to do schedule
isolate and unisolate. So reduce the critical section in the cooling
device update path by moving schedule isolation and unisolate calls
out of critical section.

Change-Id: I68c6fe3ce1ac403ec0584e831e3c631a24cc8bc5
Signed-off-by: default avatarRam Chandrasekar <rkumbako@codeaurora.org>
parent 47f9b50d

drivers/thermal/cpu_cooling.c

+10 −10
Original line number Diff line number Diff line
@@ -237,11 +237,17 @@ static int cpufreq_cooling_pm_notify(struct notifier_block *nb, 
	case PM_POST_RESTORE:

	case PM_POST_SUSPEND:

		mutex_lock(&cooling_list_lock);

		mutex_lock(&core_isolate_lock);

		list_for_each_entry(cpufreq_dev, &cpufreq_dev_list, node) {

			mutex_lock(&core_isolate_lock);

			if (cpufreq_dev->cpufreq_state ==

				cpufreq_dev->max_level) {

				cpu = cpumask_any(&cpufreq_dev->allowed_cpus);

				/*

				 * Unlock this lock before calling

				 * schedule_isolate. as this could lead to

				 * deadlock with hotplug path.

				 */

				mutex_unlock(&core_isolate_lock);

				if (cpu_online(cpu) &&

					!cpumask_test_and_set_cpu(cpu,

					&cpus_isolated_by_thermal)) {
@@ -249,9 +255,10 @@ static int cpufreq_cooling_pm_notify(struct notifier_block *nb, 
						cpumask_clear_cpu(cpu,

						&cpus_isolated_by_thermal);

				}

			}

				continue;

			}

			mutex_unlock(&core_isolate_lock);

		}

		mutex_unlock(&cooling_list_lock);



		atomic_set(&in_suspend, 0);
@@ -727,6 +734,7 @@ static int cpufreq_set_cur_state(struct thermal_cooling_device *cdev, 
	mutex_lock(&core_isolate_lock);

	prev_state = cpufreq_device->cpufreq_state;

	cpufreq_device->cpufreq_state = state;

	mutex_unlock(&core_isolate_lock);

	/* If state is the last, isolate the CPU */

	if (state == cpufreq_device->max_level) {

		if (cpu_online(cpu) &&
@@ -736,18 +744,11 @@ static int cpufreq_set_cur_state(struct thermal_cooling_device *cdev, 
				cpumask_clear_cpu(cpu,

					&cpus_isolated_by_thermal);

		}

		mutex_unlock(&core_isolate_lock);

		return ret;

	} else if ((prev_state == cpufreq_device->max_level)

			&& (state < cpufreq_device->max_level)) {

		if (cpumask_test_and_clear_cpu(cpu, &cpus_pending_online)) {

			cpu_dev = get_cpu_device(cpu);

			mutex_unlock(&core_isolate_lock);

			/*

			 * Unlock before calling the device_online.

			 * Else, this will lead to deadlock, since the hp

			 * online callback will be blocked on this mutex.

			 */

			ret = device_online(cpu_dev);

			if (ret)

				pr_err("CPU:%d online error:%d\n", cpu, ret);
@@ -757,7 +758,6 @@ static int cpufreq_set_cur_state(struct thermal_cooling_device *cdev, 
			sched_unisolate_cpu(cpu);

		}

	}

	mutex_unlock(&core_isolate_lock);

update_frequency:

	clip_freq = cpufreq_device->freq_table[state];

	cpufreq_device->clipped_freq = clip_freq;