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

Commit c92c9422 authored by Manaf Meethalavalappu Pallikunhi's avatar Manaf Meethalavalappu Pallikunhi
Browse files

drivers: lmh_dcvsh: Defer cluster cooling register until core comes online 



If all cores for a cluster are offline or not brought up during
lmh dcvsh probe, it fails to register cooling device and it leads
to probe failure. If later these cores come online, lmh dcvsh doesn't
re-probe it and eventually that cluster lmh dcvsh is broken.

Register for cpu online notification and register cpu cooling device
in cpu online notification callback context. It makes sure that
both lmh dcvsh probing and cooling device register are proper.

Change-Id: Id5345a578aab920d12bd9d9e2ccf1d5059582de9
Signed-off-by: default avatarManaf Meethalavalappu Pallikunhi <manafm@codeaurora.org>
parent 9d94fcc7

drivers/thermal/msm_lmh_dcvs.c

+34 −5
Original line number Diff line number Diff line
@@ -82,6 +82,7 @@ struct msm_lmh_dcvs_hw { 
	uint32_t affinity;

	uint32_t temp_limits[LIMITS_TRIP_MAX];

	struct sensor_threshold default_lo, default_hi;

	struct thermal_cooling_device *cdev;

	int irq_num;

	void *osm_hw_reg;

	void *int_clr_reg;
@@ -377,13 +378,38 @@ int msm_lmh_dcvsh_sw_notify(int cpu) 
	return 0;

}



static int __ref lmh_dcvs_cpu_callback(struct notifier_block *nfb,

		unsigned long action, void *hcpu)

{

	uint32_t cpu = (uintptr_t)hcpu;

	struct msm_lmh_dcvs_hw *hw = get_dcvsh_hw_from_cpu(cpu);



	if (!hw || hw->cdev)

		return NOTIFY_OK;



	switch (action & ~CPU_TASKS_FROZEN) {

	case CPU_ONLINE:

		hw->cdev = cpufreq_platform_cooling_register(&hw->core_map,

				&cd_ops);

		if (IS_ERR_OR_NULL(hw->cdev))

			hw->cdev = NULL;

		break;

	default:

		break;

	}

	return NOTIFY_OK;

}



static struct notifier_block __refdata lmh_dcvs_cpu_notifier = {

	.notifier_call = lmh_dcvs_cpu_callback,

};



static int msm_lmh_dcvs_probe(struct platform_device *pdev)

{

	int ret;

	int affinity = -1;

	struct msm_lmh_dcvs_hw *hw;

	struct thermal_zone_device *tzdev;

	struct thermal_cooling_device *cdev;

	struct device_node *dn = pdev->dev.of_node;

	struct device_node *cpu_node, *lmh_node;

	uint32_t id, max_freq, request_reg, clear_reg;
@@ -458,10 +484,6 @@ static int msm_lmh_dcvs_probe(struct platform_device *pdev) 
	if (IS_ERR_OR_NULL(tzdev))

		return PTR_ERR(tzdev);



	/* Setup cooling devices to request mitigation states */

	cdev = cpufreq_platform_cooling_register(&hw->core_map, &cd_ops);

	if (IS_ERR_OR_NULL(cdev))

		return PTR_ERR(cdev);

	/*

	 * Driver defaults to for low and hi thresholds.

	 * Since we make a check for hi > lo value, set the hi threshold
@@ -531,9 +553,16 @@ static int msm_lmh_dcvs_probe(struct platform_device *pdev) 
		return ret;

	}



	if (list_empty(&lmh_dcvs_hw_list))

		register_cpu_notifier(&lmh_dcvs_cpu_notifier);



	INIT_LIST_HEAD(&hw->list);

	list_add(&hw->list, &lmh_dcvs_hw_list);



	/* Better register explicitly for 1st CPU of each HW */

	lmh_dcvs_cpu_callback(&lmh_dcvs_cpu_notifier, CPU_ONLINE,

			(void *)(long)cpumask_first(&hw->core_map));



	return ret;

}