drivers/thermal/qcom/msm_lmh_dcvs.c - kernel/msm-4.9 - Gitiles

 /* Copyright (c) 2016-2017, The Linux Foundation. All rights reserved.
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 2 and
  * only version 2 as published by the Free Software Foundation.
  *
  * This program is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  * GNU General Public License for more details.
  */

 #define pr_fmt(fmt) "%s:%s " fmt, KBUILD_MODNAME, __func__

 #include <linux/module.h>
 #include <linux/slab.h>
 #include <linux/thermal.h>
 #include <linux/of.h>
 #include <linux/of_device.h>
 #include <linux/of_irq.h>
 #include <linux/platform_device.h>
 #include <linux/sched.h>
 #include <linux/io.h>
 #include <linux/interrupt.h>
 #include <linux/timer.h>
 #include <linux/pm_opp.h>
 #include <linux/cpu_cooling.h>
 #include <linux/atomic.h>

 #include <asm/smp_plat.h>
 #include <asm/cacheflush.h>

 #include <soc/qcom/scm.h>

 #include "../thermal_core.h"

 #define LIMITS_DCVSH                0x10
 #define LIMITS_PROFILE_CHANGE       0x01
 #define LIMITS_NODE_DCVS            0x44435653

 #define LIMITS_SUB_FN_THERMAL       0x54484D4C
 #define LIMITS_SUB_FN_CRNT          0x43524E54
 #define LIMITS_SUB_FN_REL           0x52454C00
 #define LIMITS_SUB_FN_BCL           0x42434C00
 #define LIMITS_SUB_FN_GENERAL       0x47454E00

 #define LIMITS_ALGO_MODE_ENABLE     0x454E424C

 #define LIMITS_HI_THRESHOLD         0x48494748
 #define LIMITS_LOW_THRESHOLD        0x4C4F5700
 #define LIMITS_ARM_THRESHOLD        0x41524D00

 #define LIMITS_CLUSTER_0            0x6370302D
 #define LIMITS_CLUSTER_1            0x6370312D

 #define LIMITS_DOMAIN_MAX           0x444D4158
 #define LIMITS_DOMAIN_MIN           0x444D494E

 #define LIMITS_TEMP_DEFAULT         75000
 #define LIMITS_LOW_THRESHOLD_OFFSET 500
 #define LIMITS_POLLING_DELAY_MS     10
 #define LIMITS_CLUSTER_0_REQ        0x179C1B04
 #define LIMITS_CLUSTER_1_REQ        0x179C3B04
 #define LIMITS_CLUSTER_0_INT_CLR    0x179CE808
 #define LIMITS_CLUSTER_1_INT_CLR    0x179CC808
 #define LIMITS_CLUSTER_0_MIN_FREQ   0x17D78BC0
 #define LIMITS_CLUSTER_1_MIN_FREQ   0x17D70BC0
 #define dcvsh_get_frequency(_val, _max) do { \
 	_max = (_val) & 0x3FF; \
 	_max *= 19200; \
 } while (0)
 #define FREQ_KHZ_TO_HZ(_val) ((_val) * 1000)
 #define FREQ_HZ_TO_KHZ(_val) ((_val) / 1000)

 enum lmh_hw_trips {
 	LIMITS_TRIP_ARM,
 	LIMITS_TRIP_HI,
 	LIMITS_TRIP_MAX,
 };

 struct __limits_cdev_data {
 	struct thermal_cooling_device *cdev;
 	u32 max_freq;
 	u32 min_freq;
 };

 struct limits_dcvs_hw {
 	char sensor_name[THERMAL_NAME_LENGTH];
 	uint32_t affinity;
 	uint32_t temp_limits[LIMITS_TRIP_MAX];
 	int irq_num;
 	void *osm_hw_reg;
 	void *int_clr_reg;
 	void *min_freq_reg;
 	cpumask_t core_map;
 	struct timer_list poll_timer;
 	unsigned long max_freq;
 	unsigned long min_freq;
 	unsigned long hw_freq_limit;
 	struct list_head list;
 	atomic_t is_irq_enabled;
 	struct mutex access_lock;
 	struct __limits_cdev_data *cdev_data;
 };

 LIST_HEAD(lmh_dcvs_hw_list);

 static int limits_dcvs_get_freq_limits(uint32_t cpu, unsigned long *max_freq,
 					 unsigned long *min_freq)
 {
 	unsigned long freq_ceil = UINT_MAX, freq_floor = 0;
 	struct device *cpu_dev = NULL;
 	int ret = 0;

 	cpu_dev = get_cpu_device(cpu);
 	if (!cpu_dev) {
 		pr_err("Error in get CPU%d device\n", cpu);
 		return -ENODEV;
 	}

 	rcu_read_lock();
 	dev_pm_opp_find_freq_floor(cpu_dev, &freq_ceil);
 	dev_pm_opp_find_freq_ceil(cpu_dev, &freq_floor);
 	rcu_read_unlock();

 	*max_freq = freq_ceil / 1000;
 	*min_freq = freq_floor / 1000;

 	return ret;
 }

 static unsigned long limits_mitigation_notify(struct limits_dcvs_hw *hw)
 {
 	uint32_t val = 0;
 	struct device *cpu_dev = NULL;
 	unsigned long freq_val, max_limit = 0;
 	struct dev_pm_opp *opp_entry;

 	val = readl_relaxed(hw->osm_hw_reg);
 	dcvsh_get_frequency(val, max_limit);
 	cpu_dev = get_cpu_device(cpumask_first(&hw->core_map));
 	if (!cpu_dev) {
 		pr_err("Error in get CPU%d device\n",
 			cpumask_first(&hw->core_map));
 		goto notify_exit;
 	}

 	freq_val = FREQ_KHZ_TO_HZ(max_limit);
 	rcu_read_lock();
 	opp_entry = dev_pm_opp_find_freq_floor(cpu_dev, &freq_val);
 	/*
 	 * Hardware mitigation frequency can be lower than the lowest
 	 * possible CPU frequency. In that case freq floor call will
 	 * fail with -ERANGE and we need to match to the lowest
 	 * frequency using freq_ceil.
 	 */
 	if (IS_ERR(opp_entry) && PTR_ERR(opp_entry) == -ERANGE) {
 		opp_entry = dev_pm_opp_find_freq_ceil(cpu_dev, &freq_val);
 		if (IS_ERR(opp_entry))
 			dev_err(cpu_dev, "frequency:%lu. opp error:%ld\n",
 					freq_val, PTR_ERR(opp_entry));
 	}
 	rcu_read_unlock();
 	max_limit = FREQ_HZ_TO_KHZ(freq_val);

 	sched_update_cpu_freq_min_max(&hw->core_map, 0, max_limit);

 notify_exit:
 	hw->hw_freq_limit = max_limit;
 	return max_limit;
 }

 static void limits_dcvs_poll(unsigned long data)
 {
 	unsigned long max_limit = 0;
 	struct limits_dcvs_hw *hw = (struct limits_dcvs_hw *)data;

 	if (hw->max_freq == UINT_MAX)
 		limits_dcvs_get_freq_limits(cpumask_first(&hw->core_map),
 			&hw->max_freq, &hw->min_freq);
 	max_limit = limits_mitigation_notify(hw);
 	if (max_limit >= hw->max_freq) {
 		del_timer(&hw->poll_timer);
 		writel_relaxed(0xFF, hw->int_clr_reg);
 		atomic_set(&hw->is_irq_enabled, 1);
 		enable_irq(hw->irq_num);
 	} else {
 		mod_timer(&hw->poll_timer, jiffies + msecs_to_jiffies(
 			LIMITS_POLLING_DELAY_MS));
 	}
 }

 static void lmh_dcvs_notify(struct limits_dcvs_hw *hw)
 {
 	if (atomic_dec_and_test(&hw->is_irq_enabled)) {
 		disable_irq_nosync(hw->irq_num);
 		limits_mitigation_notify(hw);
 		mod_timer(&hw->poll_timer, jiffies + msecs_to_jiffies(
 			LIMITS_POLLING_DELAY_MS));
 	}
 }

 static irqreturn_t lmh_dcvs_handle_isr(int irq, void *data)
 {
 	struct limits_dcvs_hw *hw = data;

 	lmh_dcvs_notify(hw);

 	return IRQ_HANDLED;
 }

 static int limits_dcvs_write(uint32_t node_id, uint32_t fn,
 			      uint32_t setting, uint32_t val)
 {
 	int ret;
 	struct scm_desc desc_arg;
 	uint32_t *payload = NULL;

 	payload = kzalloc(sizeof(uint32_t) * 5, GFP_KERNEL);
 	if (!payload)
 		return -ENOMEM;

 	payload[0] = fn; /* algorithm */
 	payload[1] = 0; /* unused sub-algorithm */
 	payload[2] = setting;
 	payload[3] = 1; /* number of values */
 	payload[4] = val;

 	desc_arg.args[0] = SCM_BUFFER_PHYS(payload);
 	desc_arg.args[1] = sizeof(uint32_t) * 5;
 	desc_arg.args[2] = LIMITS_NODE_DCVS;
 	desc_arg.args[3] = node_id;
 	desc_arg.args[4] = 0; /* version */
 	desc_arg.arginfo = SCM_ARGS(5, SCM_RO, SCM_VAL, SCM_VAL,
 					SCM_VAL, SCM_VAL);

 	dmac_flush_range(payload, (void *)payload + 5 * (sizeof(uint32_t)));
 	ret = scm_call2(SCM_SIP_FNID(SCM_SVC_LMH, LIMITS_DCVSH), &desc_arg);

 	kfree(payload);

 	return ret;
 }

 static int lmh_get_temp(void *data, int *val)
 {
 	/*
 	 * LMH DCVSh hardware doesn't support temperature read.
 	 * return a default value for the thermal core to aggregate
 	 * the thresholds
 	 */
 	*val = LIMITS_TEMP_DEFAULT;

 	return 0;
 }

 static int lmh_set_trips(void *data, int low, int high)
 {
 	struct limits_dcvs_hw *hw = (struct limits_dcvs_hw *)data;
 	int ret = 0;

 	if (high < LIMITS_LOW_THRESHOLD_OFFSET || low < 0) {
 		pr_err("Value out of range low:%d high:%d\n",
 				low, high);
 		return -EINVAL;
 	}

 	/* Sanity check limits before writing to the hardware */
 	if (low >= high)
 		return -EINVAL;

 	hw->temp_limits[LIMITS_TRIP_HI] = (uint32_t)high;
 	hw->temp_limits[LIMITS_TRIP_ARM] = (uint32_t)low;

 	ret =  limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
 				  LIMITS_ARM_THRESHOLD, low);
 	if (ret)
 		return ret;
 	ret =  limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
 				  LIMITS_HI_THRESHOLD, high);
 	if (ret)
 		return ret;
 	ret =  limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
 				  LIMITS_LOW_THRESHOLD,
 				  high - LIMITS_LOW_THRESHOLD_OFFSET);
 	if (ret)
 		return ret;

 	return ret;
 }

 static struct thermal_zone_of_device_ops limits_sensor_ops = {
 	.get_temp   = lmh_get_temp,
 	.set_trips  = lmh_set_trips,
 };

 static struct limits_dcvs_hw *get_dcvsh_hw_from_cpu(int cpu)
 {
 	struct limits_dcvs_hw *hw;

 	list_for_each_entry(hw, &lmh_dcvs_hw_list, list) {
 		if (cpumask_test_cpu(cpu, &hw->core_map))
 			return hw;
 	}

 	return NULL;
 }

 static int enable_lmh(void)
 {
 	int ret = 0;
 	struct scm_desc desc_arg;

 	desc_arg.args[0] = 1;
 	desc_arg.arginfo = SCM_ARGS(1, SCM_VAL);
 	ret = scm_call2(SCM_SIP_FNID(SCM_SVC_LMH, LIMITS_PROFILE_CHANGE),
 			&desc_arg);
 	if (ret) {
 		pr_err("Error switching profile:[1]. err:%d\n", ret);
 		return ret;
 	}

 	return ret;
 }

 static int lmh_set_max_limit(int cpu, u32 freq)
 {
 	struct limits_dcvs_hw *hw = get_dcvsh_hw_from_cpu(cpu);
 	int ret = 0, cpu_idx, idx = 0;
 	u32 max_freq = U32_MAX;

 	if (!hw)
 		return -EINVAL;

 	mutex_lock(&hw->access_lock);
 	for_each_cpu(cpu_idx, &hw->core_map) {
 		if (cpu_idx == cpu)
 			hw->cdev_data[idx].max_freq = freq;
 		if (max_freq > hw->cdev_data[idx].max_freq)
 			max_freq = hw->cdev_data[idx].max_freq;
 		idx++;
 	}
 	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_GENERAL,
 				  LIMITS_DOMAIN_MAX, max_freq);
 	mutex_unlock(&hw->access_lock);

 	return ret;
 }

 static int lmh_set_min_limit(int cpu, u32 freq)
 {
 	struct limits_dcvs_hw *hw = get_dcvsh_hw_from_cpu(cpu);
 	int cpu_idx, idx = 0;
 	u32 min_freq = 0;

 	if (!hw)
 		return -EINVAL;

 	mutex_lock(&hw->access_lock);
 	for_each_cpu(cpu_idx, &hw->core_map) {
 		if (cpu_idx == cpu)
 			hw->cdev_data[idx].min_freq = freq;
 		if (min_freq < hw->cdev_data[idx].min_freq)
 			min_freq = hw->cdev_data[idx].min_freq;
 		idx++;
 	}
 	if (min_freq != hw->min_freq)
 		writel_relaxed(0x01, hw->min_freq_reg);
 	else
 		writel_relaxed(0x00, hw->min_freq_reg);
 	mutex_unlock(&hw->access_lock);

 	return 0;
 }
 static struct cpu_cooling_ops cd_ops = {
 	.ceil_limit = lmh_set_max_limit,
 	.floor_limit = lmh_set_min_limit,
 };

 static int limits_dcvs_probe(struct platform_device *pdev)
 {
 	int ret;
 	int affinity = -1;
 	struct limits_dcvs_hw *hw;
 	struct thermal_zone_device *tzdev;
 	struct device_node *dn = pdev->dev.of_node;
 	struct device_node *cpu_node, *lmh_node;
 	uint32_t request_reg, clear_reg, min_reg;
 	unsigned long max_freq, min_freq;
 	int cpu, idx;
 	cpumask_t mask = { CPU_BITS_NONE };

 	for_each_possible_cpu(cpu) {
 		cpu_node = of_cpu_device_node_get(cpu);
 		if (!cpu_node)
 			continue;
 		lmh_node = of_parse_phandle(cpu_node, "qcom,lmh-dcvs", 0);
 		if (lmh_node == dn) {
 			/*set the cpumask*/
 			cpumask_set_cpu(cpu, &(mask));
 		}
 		of_node_put(cpu_node);
 		of_node_put(lmh_node);
 	}

 	/*
 	 * We return error if none of the CPUs have
 	 * reference to our LMH node
 	 */
 	if (cpumask_empty(&mask))
 		return -EINVAL;

 	ret = limits_dcvs_get_freq_limits(cpumask_first(&mask), &max_freq,
 				     &min_freq);
 	if (ret)
 		return ret;
 	hw = devm_kzalloc(&pdev->dev, sizeof(*hw), GFP_KERNEL);
 	if (!hw)
 		return -ENOMEM;
 	hw->cdev_data = devm_kcalloc(&pdev->dev, cpumask_weight(&mask),
 				   sizeof(*hw->cdev_data),
 				   GFP_KERNEL);
 	if (!hw->cdev_data)
 		return -ENOMEM;

 	cpumask_copy(&hw->core_map, &mask);
 	ret = of_property_read_u32(dn, "qcom,affinity", &affinity);
 	if (ret)
 		return -ENODEV;
 	switch (affinity) {
 	case 0:
 		hw->affinity = LIMITS_CLUSTER_0;
 		break;
 	case 1:
 		hw->affinity = LIMITS_CLUSTER_1;
 		break;
 	default:
 		return -EINVAL;
 	};

 	/* Enable the thermal algorithm early */
 	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
 		 LIMITS_ALGO_MODE_ENABLE, 1);
 	if (ret)
 		return ret;
 	/* Enable the LMH outer loop algorithm */
 	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_CRNT,
 		 LIMITS_ALGO_MODE_ENABLE, 1);
 	if (ret)
 		return ret;
 	/* Enable the Reliability algorithm */
 	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_REL,
 		 LIMITS_ALGO_MODE_ENABLE, 1);
 	if (ret)
 		return ret;
 	/* Enable the BCL algorithm */
 	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_BCL,
 		 LIMITS_ALGO_MODE_ENABLE, 1);
 	if (ret)
 		return ret;
 	ret = enable_lmh();
 	if (ret)
 		return ret;

 	/*
 	 * Setup virtual thermal zones for each LMH-DCVS hardware
 	 * The sensor does not do actual thermal temperature readings
 	 * but does support setting thresholds for trips.
 	 * Let's register with thermal framework, so we have the ability
 	 * to set low/high thresholds.
 	 */
 	hw->temp_limits[LIMITS_TRIP_HI] = INT_MAX;
 	hw->temp_limits[LIMITS_TRIP_ARM] = 0;
 	hw->hw_freq_limit = hw->max_freq = max_freq;
 	hw->min_freq = min_freq;
 	snprintf(hw->sensor_name, sizeof(hw->sensor_name), "limits_sensor-%02d",
 			affinity);
 	tzdev = thermal_zone_of_sensor_register(&pdev->dev, 0, hw,
 			&limits_sensor_ops);
 	if (IS_ERR_OR_NULL(tzdev))
 		return PTR_ERR(tzdev);

 	/* Setup cooling devices to request mitigation states */
 	mutex_init(&hw->access_lock);
 	idx = 0;
 	for_each_cpu(cpu, &hw->core_map) {
 		cpumask_t cpu_mask  = { CPU_BITS_NONE };

 		cpumask_set_cpu(cpu, &cpu_mask);
 		hw->cdev_data[idx].cdev = cpufreq_platform_cooling_register(
 						&cpu_mask, &cd_ops);
 		if (IS_ERR_OR_NULL(hw->cdev_data[idx].cdev))
 			return PTR_ERR(hw->cdev_data[idx].cdev);
 		hw->cdev_data[idx].max_freq = U32_MAX;
 		hw->cdev_data[idx].min_freq = 0;
 		idx++;
 	}

 	switch (affinity) {
 	case 0:
 		request_reg = LIMITS_CLUSTER_0_REQ;
 		clear_reg = LIMITS_CLUSTER_0_INT_CLR;
 		min_reg = LIMITS_CLUSTER_0_MIN_FREQ;
 		break;
 	case 1:
 		request_reg = LIMITS_CLUSTER_1_REQ;
 		clear_reg = LIMITS_CLUSTER_1_INT_CLR;
 		min_reg = LIMITS_CLUSTER_1_MIN_FREQ;
 		break;
 	default:
 		return -EINVAL;
 	};

 	hw->osm_hw_reg = devm_ioremap(&pdev->dev, request_reg, 0x4);
 	if (!hw->osm_hw_reg) {
 		pr_err("register remap failed\n");
 		return -ENOMEM;
 	}
 	hw->int_clr_reg = devm_ioremap(&pdev->dev, clear_reg, 0x4);
 	if (!hw->int_clr_reg) {
 		pr_err("interrupt clear reg remap failed\n");
 		return -ENOMEM;
 	}
 	hw->min_freq_reg = devm_ioremap(&pdev->dev, min_reg, 0x4);
 	if (!hw->min_freq_reg) {
 		pr_err("min frequency enable register remap failed\n");
 		return -ENOMEM;
 	}
 	init_timer_deferrable(&hw->poll_timer);
 	hw->poll_timer.data = (unsigned long)hw;
 	hw->poll_timer.function = limits_dcvs_poll;

 	hw->irq_num = of_irq_get(pdev->dev.of_node, 0);
 	if (hw->irq_num < 0) {
 		ret = hw->irq_num;
 		pr_err("Error getting IRQ number. err:%d\n", ret);
 		return ret;
 	}
 	atomic_set(&hw->is_irq_enabled, 1);
 	ret = devm_request_threaded_irq(&pdev->dev, hw->irq_num, NULL,
 		lmh_dcvs_handle_isr, IRQF_TRIGGER_HIGH | IRQF_ONESHOT
 		| IRQF_NO_SUSPEND, hw->sensor_name, hw);
 	if (ret) {
 		pr_err("Error registering for irq. err:%d\n", ret);
 		return ret;
 	}

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

 	return ret;
 }

 static const struct of_device_id limits_dcvs_match[] = {
 	{ .compatible = "qcom,msm-hw-limits", },
 	{},
 };

 static struct platform_driver limits_dcvs_driver = {
 	.probe		= limits_dcvs_probe,
 	.driver		= {
 		.name = KBUILD_MODNAME,
 		.of_match_table = limits_dcvs_match,
 	},
 };
 builtin_platform_driver(limits_dcvs_driver);
	/* Copyright (c) 2016-2017, The Linux Foundation. All rights reserved.
	*
	* This program is free software; you can redistribute it and/or modify
	* it under the terms of the GNU General Public License version 2 and
	* only version 2 as published by the Free Software Foundation.
	*
	* This program is distributed in the hope that it will be useful,
	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	* GNU General Public License for more details.
	*/

	#define pr_fmt(fmt) "%s:%s " fmt, KBUILD_MODNAME, __func__

	#include <linux/module.h>
	#include <linux/slab.h>
	#include <linux/thermal.h>
	#include <linux/of.h>
	#include <linux/of_device.h>
	#include <linux/of_irq.h>
	#include <linux/platform_device.h>
	#include <linux/sched.h>
	#include <linux/io.h>
	#include <linux/interrupt.h>
	#include <linux/timer.h>
	#include <linux/pm_opp.h>
	#include <linux/cpu_cooling.h>
	#include <linux/atomic.h>

	#include <asm/smp_plat.h>
	#include <asm/cacheflush.h>

	#include <soc/qcom/scm.h>

	#include "../thermal_core.h"

	#define LIMITS_DCVSH 0x10
	#define LIMITS_PROFILE_CHANGE 0x01
	#define LIMITS_NODE_DCVS 0x44435653

	#define LIMITS_SUB_FN_THERMAL 0x54484D4C
	#define LIMITS_SUB_FN_CRNT 0x43524E54
	#define LIMITS_SUB_FN_REL 0x52454C00
	#define LIMITS_SUB_FN_BCL 0x42434C00
	#define LIMITS_SUB_FN_GENERAL 0x47454E00

	#define LIMITS_ALGO_MODE_ENABLE 0x454E424C

	#define LIMITS_HI_THRESHOLD 0x48494748
	#define LIMITS_LOW_THRESHOLD 0x4C4F5700
	#define LIMITS_ARM_THRESHOLD 0x41524D00

	#define LIMITS_CLUSTER_0 0x6370302D
	#define LIMITS_CLUSTER_1 0x6370312D

	#define LIMITS_DOMAIN_MAX 0x444D4158
	#define LIMITS_DOMAIN_MIN 0x444D494E

	#define LIMITS_TEMP_DEFAULT 75000
	#define LIMITS_LOW_THRESHOLD_OFFSET 500
	#define LIMITS_POLLING_DELAY_MS 10
	#define LIMITS_CLUSTER_0_REQ 0x179C1B04
	#define LIMITS_CLUSTER_1_REQ 0x179C3B04
	#define LIMITS_CLUSTER_0_INT_CLR 0x179CE808
	#define LIMITS_CLUSTER_1_INT_CLR 0x179CC808
	#define LIMITS_CLUSTER_0_MIN_FREQ 0x17D78BC0
	#define LIMITS_CLUSTER_1_MIN_FREQ 0x17D70BC0
	#define dcvsh_get_frequency(_val, _max) do { \
	_max = (_val) & 0x3FF; \
	_max *= 19200; \
	} while (0)
	#define FREQ_KHZ_TO_HZ(_val) ((_val) * 1000)
	#define FREQ_HZ_TO_KHZ(_val) ((_val) / 1000)

	enum lmh_hw_trips {
	LIMITS_TRIP_ARM,
	LIMITS_TRIP_HI,
	LIMITS_TRIP_MAX,
	};

	struct __limits_cdev_data {
	struct thermal_cooling_device *cdev;
	u32 max_freq;
	u32 min_freq;
	};

	struct limits_dcvs_hw {
	char sensor_name[THERMAL_NAME_LENGTH];
	uint32_t affinity;
	uint32_t temp_limits[LIMITS_TRIP_MAX];
	int irq_num;
	void *osm_hw_reg;
	void *int_clr_reg;
	void *min_freq_reg;
	cpumask_t core_map;
	struct timer_list poll_timer;
	unsigned long max_freq;
	unsigned long min_freq;
	unsigned long hw_freq_limit;
	struct list_head list;
	atomic_t is_irq_enabled;
	struct mutex access_lock;
	struct __limits_cdev_data *cdev_data;
	};

	LIST_HEAD(lmh_dcvs_hw_list);

	static int limits_dcvs_get_freq_limits(uint32_t cpu, unsigned long *max_freq,
	unsigned long *min_freq)
	{
	unsigned long freq_ceil = UINT_MAX, freq_floor = 0;
	struct device *cpu_dev = NULL;
	int ret = 0;

	cpu_dev = get_cpu_device(cpu);
	if (!cpu_dev) {
	pr_err("Error in get CPU%d device\n", cpu);
	return -ENODEV;
	}

	rcu_read_lock();
	dev_pm_opp_find_freq_floor(cpu_dev, &freq_ceil);
	dev_pm_opp_find_freq_ceil(cpu_dev, &freq_floor);
	rcu_read_unlock();

	*max_freq = freq_ceil / 1000;
	*min_freq = freq_floor / 1000;

	return ret;
	}

	static unsigned long limits_mitigation_notify(struct limits_dcvs_hw *hw)
	{
	uint32_t val = 0;
	struct device *cpu_dev = NULL;
	unsigned long freq_val, max_limit = 0;
	struct dev_pm_opp *opp_entry;

	val = readl_relaxed(hw->osm_hw_reg);
	dcvsh_get_frequency(val, max_limit);
	cpu_dev = get_cpu_device(cpumask_first(&hw->core_map));
	if (!cpu_dev) {
	pr_err("Error in get CPU%d device\n",
	cpumask_first(&hw->core_map));
	goto notify_exit;
	}

	freq_val = FREQ_KHZ_TO_HZ(max_limit);
	rcu_read_lock();
	opp_entry = dev_pm_opp_find_freq_floor(cpu_dev, &freq_val);
	/*
	* Hardware mitigation frequency can be lower than the lowest
	* possible CPU frequency. In that case freq floor call will
	* fail with -ERANGE and we need to match to the lowest
	* frequency using freq_ceil.
	*/
	if (IS_ERR(opp_entry) && PTR_ERR(opp_entry) == -ERANGE) {
	opp_entry = dev_pm_opp_find_freq_ceil(cpu_dev, &freq_val);
	if (IS_ERR(opp_entry))
	dev_err(cpu_dev, "frequency:%lu. opp error:%ld\n",
	freq_val, PTR_ERR(opp_entry));
	}
	rcu_read_unlock();
	max_limit = FREQ_HZ_TO_KHZ(freq_val);

	sched_update_cpu_freq_min_max(&hw->core_map, 0, max_limit);

	notify_exit:
	hw->hw_freq_limit = max_limit;
	return max_limit;
	}

	static void limits_dcvs_poll(unsigned long data)
	{
	unsigned long max_limit = 0;
	struct limits_dcvs_hw hw = (struct limits_dcvs_hw )data;

	if (hw->max_freq == UINT_MAX)
	limits_dcvs_get_freq_limits(cpumask_first(&hw->core_map),
	&hw->max_freq, &hw->min_freq);
	max_limit = limits_mitigation_notify(hw);
	if (max_limit >= hw->max_freq) {
	del_timer(&hw->poll_timer);
	writel_relaxed(0xFF, hw->int_clr_reg);
	atomic_set(&hw->is_irq_enabled, 1);
	enable_irq(hw->irq_num);
	} else {
	mod_timer(&hw->poll_timer, jiffies + msecs_to_jiffies(
	LIMITS_POLLING_DELAY_MS));
	}
	}

	static void lmh_dcvs_notify(struct limits_dcvs_hw *hw)
	{
	if (atomic_dec_and_test(&hw->is_irq_enabled)) {
	disable_irq_nosync(hw->irq_num);
	limits_mitigation_notify(hw);
	mod_timer(&hw->poll_timer, jiffies + msecs_to_jiffies(
	LIMITS_POLLING_DELAY_MS));
	}
	}

	static irqreturn_t lmh_dcvs_handle_isr(int irq, void *data)
	{
	struct limits_dcvs_hw *hw = data;

	lmh_dcvs_notify(hw);

	return IRQ_HANDLED;
	}

	static int limits_dcvs_write(uint32_t node_id, uint32_t fn,
	uint32_t setting, uint32_t val)
	{
	int ret;
	struct scm_desc desc_arg;
	uint32_t *payload = NULL;

	payload = kzalloc(sizeof(uint32_t) * 5, GFP_KERNEL);
	if (!payload)
	return -ENOMEM;

	payload[0] = fn; /* algorithm */
	payload[1] = 0; /* unused sub-algorithm */
	payload[2] = setting;
	payload[3] = 1; /* number of values */
	payload[4] = val;

	desc_arg.args[0] = SCM_BUFFER_PHYS(payload);
	desc_arg.args[1] = sizeof(uint32_t) * 5;
	desc_arg.args[2] = LIMITS_NODE_DCVS;
	desc_arg.args[3] = node_id;
	desc_arg.args[4] = 0; /* version */
	desc_arg.arginfo = SCM_ARGS(5, SCM_RO, SCM_VAL, SCM_VAL,
	SCM_VAL, SCM_VAL);

	dmac_flush_range(payload, (void )payload + 5 (sizeof(uint32_t)));
	ret = scm_call2(SCM_SIP_FNID(SCM_SVC_LMH, LIMITS_DCVSH), &desc_arg);

	kfree(payload);

	return ret;
	}

	static int lmh_get_temp(void data, int val)
	{
	/*
	* LMH DCVSh hardware doesn't support temperature read.
	* return a default value for the thermal core to aggregate
	* the thresholds
	*/
	*val = LIMITS_TEMP_DEFAULT;

	return 0;
	}

	static int lmh_set_trips(void *data, int low, int high)
	{
	struct limits_dcvs_hw hw = (struct limits_dcvs_hw )data;
	int ret = 0;

	if (high < LIMITS_LOW_THRESHOLD_OFFSET \|\| low < 0) {
	pr_err("Value out of range low:%d high:%d\n",
	low, high);
	return -EINVAL;
	}

	/* Sanity check limits before writing to the hardware */
	if (low >= high)
	return -EINVAL;

	hw->temp_limits[LIMITS_TRIP_HI] = (uint32_t)high;
	hw->temp_limits[LIMITS_TRIP_ARM] = (uint32_t)low;

	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
	LIMITS_ARM_THRESHOLD, low);
	if (ret)
	return ret;
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
	LIMITS_HI_THRESHOLD, high);
	if (ret)
	return ret;
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
	LIMITS_LOW_THRESHOLD,
	high - LIMITS_LOW_THRESHOLD_OFFSET);
	if (ret)
	return ret;

	return ret;
	}

	static struct thermal_zone_of_device_ops limits_sensor_ops = {
	.get_temp = lmh_get_temp,
	.set_trips = lmh_set_trips,
	};

	static struct limits_dcvs_hw *get_dcvsh_hw_from_cpu(int cpu)
	{
	struct limits_dcvs_hw *hw;

	list_for_each_entry(hw, &lmh_dcvs_hw_list, list) {
	if (cpumask_test_cpu(cpu, &hw->core_map))
	return hw;
	}

	return NULL;
	}

	static int enable_lmh(void)
	{
	int ret = 0;
	struct scm_desc desc_arg;

	desc_arg.args[0] = 1;
	desc_arg.arginfo = SCM_ARGS(1, SCM_VAL);
	ret = scm_call2(SCM_SIP_FNID(SCM_SVC_LMH, LIMITS_PROFILE_CHANGE),
	&desc_arg);
	if (ret) {
	pr_err("Error switching profile:[1]. err:%d\n", ret);
	return ret;
	}

	return ret;
	}

	static int lmh_set_max_limit(int cpu, u32 freq)
	{
	struct limits_dcvs_hw *hw = get_dcvsh_hw_from_cpu(cpu);
	int ret = 0, cpu_idx, idx = 0;
	u32 max_freq = U32_MAX;

	if (!hw)
	return -EINVAL;

	mutex_lock(&hw->access_lock);
	for_each_cpu(cpu_idx, &hw->core_map) {
	if (cpu_idx == cpu)
	hw->cdev_data[idx].max_freq = freq;
	if (max_freq > hw->cdev_data[idx].max_freq)
	max_freq = hw->cdev_data[idx].max_freq;
	idx++;
	}
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_GENERAL,
	LIMITS_DOMAIN_MAX, max_freq);
	mutex_unlock(&hw->access_lock);

	return ret;
	}

	static int lmh_set_min_limit(int cpu, u32 freq)
	{
	struct limits_dcvs_hw *hw = get_dcvsh_hw_from_cpu(cpu);
	int cpu_idx, idx = 0;
	u32 min_freq = 0;

	if (!hw)
	return -EINVAL;

	mutex_lock(&hw->access_lock);
	for_each_cpu(cpu_idx, &hw->core_map) {
	if (cpu_idx == cpu)
	hw->cdev_data[idx].min_freq = freq;
	if (min_freq < hw->cdev_data[idx].min_freq)
	min_freq = hw->cdev_data[idx].min_freq;
	idx++;
	}
	if (min_freq != hw->min_freq)
	writel_relaxed(0x01, hw->min_freq_reg);
	else
	writel_relaxed(0x00, hw->min_freq_reg);
	mutex_unlock(&hw->access_lock);

	return 0;
	}
	static struct cpu_cooling_ops cd_ops = {
	.ceil_limit = lmh_set_max_limit,
	.floor_limit = lmh_set_min_limit,
	};

	static int limits_dcvs_probe(struct platform_device *pdev)
	{
	int ret;
	int affinity = -1;
	struct limits_dcvs_hw *hw;
	struct thermal_zone_device *tzdev;
	struct device_node *dn = pdev->dev.of_node;
	struct device_node cpu_node, lmh_node;
	uint32_t request_reg, clear_reg, min_reg;
	unsigned long max_freq, min_freq;
	int cpu, idx;
	cpumask_t mask = { CPU_BITS_NONE };

	for_each_possible_cpu(cpu) {
	cpu_node = of_cpu_device_node_get(cpu);
	if (!cpu_node)
	continue;
	lmh_node = of_parse_phandle(cpu_node, "qcom,lmh-dcvs", 0);
	if (lmh_node == dn) {
	/set the cpumask/
	cpumask_set_cpu(cpu, &(mask));
	}
	of_node_put(cpu_node);
	of_node_put(lmh_node);
	}

	/*
	* We return error if none of the CPUs have
	* reference to our LMH node
	*/
	if (cpumask_empty(&mask))
	return -EINVAL;

	ret = limits_dcvs_get_freq_limits(cpumask_first(&mask), &max_freq,
	&min_freq);
	if (ret)
	return ret;
	hw = devm_kzalloc(&pdev->dev, sizeof(*hw), GFP_KERNEL);
	if (!hw)
	return -ENOMEM;
	hw->cdev_data = devm_kcalloc(&pdev->dev, cpumask_weight(&mask),
	sizeof(*hw->cdev_data),
	GFP_KERNEL);
	if (!hw->cdev_data)
	return -ENOMEM;

	cpumask_copy(&hw->core_map, &mask);
	ret = of_property_read_u32(dn, "qcom,affinity", &affinity);
	if (ret)
	return -ENODEV;
	switch (affinity) {
	case 0:
	hw->affinity = LIMITS_CLUSTER_0;
	break;
	case 1:
	hw->affinity = LIMITS_CLUSTER_1;
	break;
	default:
	return -EINVAL;
	};

	/* Enable the thermal algorithm early */
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_THERMAL,
	LIMITS_ALGO_MODE_ENABLE, 1);
	if (ret)
	return ret;
	/* Enable the LMH outer loop algorithm */
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_CRNT,
	LIMITS_ALGO_MODE_ENABLE, 1);
	if (ret)
	return ret;
	/* Enable the Reliability algorithm */
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_REL,
	LIMITS_ALGO_MODE_ENABLE, 1);
	if (ret)
	return ret;
	/* Enable the BCL algorithm */
	ret = limits_dcvs_write(hw->affinity, LIMITS_SUB_FN_BCL,
	LIMITS_ALGO_MODE_ENABLE, 1);
	if (ret)
	return ret;
	ret = enable_lmh();
	if (ret)
	return ret;

	/*
	* Setup virtual thermal zones for each LMH-DCVS hardware
	* The sensor does not do actual thermal temperature readings
	* but does support setting thresholds for trips.
	* Let's register with thermal framework, so we have the ability
	* to set low/high thresholds.
	*/
	hw->temp_limits[LIMITS_TRIP_HI] = INT_MAX;
	hw->temp_limits[LIMITS_TRIP_ARM] = 0;
	hw->hw_freq_limit = hw->max_freq = max_freq;
	hw->min_freq = min_freq;
	snprintf(hw->sensor_name, sizeof(hw->sensor_name), "limits_sensor-%02d",
	affinity);
	tzdev = thermal_zone_of_sensor_register(&pdev->dev, 0, hw,
	&limits_sensor_ops);
	if (IS_ERR_OR_NULL(tzdev))
	return PTR_ERR(tzdev);

	/* Setup cooling devices to request mitigation states */
	mutex_init(&hw->access_lock);
	idx = 0;
	for_each_cpu(cpu, &hw->core_map) {
	cpumask_t cpu_mask = { CPU_BITS_NONE };

	cpumask_set_cpu(cpu, &cpu_mask);
	hw->cdev_data[idx].cdev = cpufreq_platform_cooling_register(
	&cpu_mask, &cd_ops);
	if (IS_ERR_OR_NULL(hw->cdev_data[idx].cdev))
	return PTR_ERR(hw->cdev_data[idx].cdev);
	hw->cdev_data[idx].max_freq = U32_MAX;
	hw->cdev_data[idx].min_freq = 0;
	idx++;
	}

	switch (affinity) {
	case 0:
	request_reg = LIMITS_CLUSTER_0_REQ;
	clear_reg = LIMITS_CLUSTER_0_INT_CLR;
	min_reg = LIMITS_CLUSTER_0_MIN_FREQ;
	break;
	case 1:
	request_reg = LIMITS_CLUSTER_1_REQ;
	clear_reg = LIMITS_CLUSTER_1_INT_CLR;
	min_reg = LIMITS_CLUSTER_1_MIN_FREQ;
	break;
	default:
	return -EINVAL;
	};

	hw->osm_hw_reg = devm_ioremap(&pdev->dev, request_reg, 0x4);
	if (!hw->osm_hw_reg) {
	pr_err("register remap failed\n");
	return -ENOMEM;
	}
	hw->int_clr_reg = devm_ioremap(&pdev->dev, clear_reg, 0x4);
	if (!hw->int_clr_reg) {
	pr_err("interrupt clear reg remap failed\n");
	return -ENOMEM;
	}
	hw->min_freq_reg = devm_ioremap(&pdev->dev, min_reg, 0x4);
	if (!hw->min_freq_reg) {
	pr_err("min frequency enable register remap failed\n");
	return -ENOMEM;
	}
	init_timer_deferrable(&hw->poll_timer);
	hw->poll_timer.data = (unsigned long)hw;
	hw->poll_timer.function = limits_dcvs_poll;

	hw->irq_num = of_irq_get(pdev->dev.of_node, 0);
	if (hw->irq_num < 0) {
	ret = hw->irq_num;
	pr_err("Error getting IRQ number. err:%d\n", ret);
	return ret;
	}
	atomic_set(&hw->is_irq_enabled, 1);
	ret = devm_request_threaded_irq(&pdev->dev, hw->irq_num, NULL,
	lmh_dcvs_handle_isr, IRQF_TRIGGER_HIGH \| IRQF_ONESHOT
	\| IRQF_NO_SUSPEND, hw->sensor_name, hw);
	if (ret) {
	pr_err("Error registering for irq. err:%d\n", ret);
	return ret;
	}

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

	return ret;
	}

	static const struct of_device_id limits_dcvs_match[] = {
	{ .compatible = "qcom,msm-hw-limits", },
	{},
	};

	static struct platform_driver limits_dcvs_driver = {
	.probe = limits_dcvs_probe,
	.driver = {
	.name = KBUILD_MODNAME,
	.of_match_table = limits_dcvs_match,
	},
	};
	builtin_platform_driver(limits_dcvs_driver);