[mirror_ubuntu-bionic-kernel.git] / drivers / cpuidle / cpuidle-powernv.c

// SPDX-License-Identifier: GPL-2.0
/*
 *  cpuidle-powernv - idle state cpuidle driver.
 *  Adapted from drivers/cpuidle/cpuidle-pseries
 *
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/moduleparam.h>
#include <linux/cpuidle.h>
#include <linux/cpu.h>
#include <linux/notifier.h>
#include <linux/clockchips.h>
#include <linux/of.h>
#include <linux/slab.h>

#include <asm/machdep.h>
#include <asm/firmware.h>
#include <asm/opal.h>
#include <asm/runlatch.h>
#include <asm/cpuidle.h>

/*
 * Expose only those Hardware idle states via the cpuidle framework
 * that have latency value below POWERNV_THRESHOLD_LATENCY_NS.
 */
#define POWERNV_THRESHOLD_LATENCY_NS 200000

static struct cpuidle_driver powernv_idle_driver = {
	.name             = "powernv_idle",
	.owner            = THIS_MODULE,
};

static int max_idle_state __read_mostly;
static struct cpuidle_state *cpuidle_state_table __read_mostly;

struct stop_psscr_table {
	u64 val;
	u64 mask;
};

static struct stop_psscr_table stop_psscr_table[CPUIDLE_STATE_MAX] __read_mostly;

static u64 snooze_timeout __read_mostly;
static bool snooze_timeout_en __read_mostly;

static int snooze_loop(struct cpuidle_device *dev,
			struct cpuidle_driver *drv,
			int index)
{
	u64 snooze_exit_time;

	set_thread_flag(TIF_POLLING_NRFLAG);

	local_irq_enable();

	snooze_exit_time = get_tb() + snooze_timeout;
	ppc64_runlatch_off();
	HMT_very_low();
	while (!need_resched()) {
		if (likely(snooze_timeout_en) && get_tb() > snooze_exit_time) {
			/*
			 * Task has not woken up but we are exiting the polling
			 * loop anyway. Require a barrier after polling is
			 * cleared to order subsequent test of need_resched().
			 */
			clear_thread_flag(TIF_POLLING_NRFLAG);
			smp_mb();
			break;
		}
	}

	HMT_medium();
	ppc64_runlatch_on();
	clear_thread_flag(TIF_POLLING_NRFLAG);

	return index;
}

static int nap_loop(struct cpuidle_device *dev,
			struct cpuidle_driver *drv,
			int index)
{
	power7_idle_type(PNV_THREAD_NAP);

	return index;
}

/* Register for fastsleep only in oneshot mode of broadcast */
#ifdef CONFIG_TICK_ONESHOT
static int fastsleep_loop(struct cpuidle_device *dev,
				struct cpuidle_driver *drv,
				int index)
{
	unsigned long old_lpcr = mfspr(SPRN_LPCR);
	unsigned long new_lpcr;

	if (unlikely(system_state < SYSTEM_RUNNING))
		return index;

	new_lpcr = old_lpcr;
	/* Do not exit powersave upon decrementer as we've setup the timer
	 * offload.
	 */
	new_lpcr &= ~LPCR_PECE1;

	mtspr(SPRN_LPCR, new_lpcr);

	power7_idle_type(PNV_THREAD_SLEEP);

	mtspr(SPRN_LPCR, old_lpcr);

	return index;
}
#endif

static int stop_loop(struct cpuidle_device *dev,
		     struct cpuidle_driver *drv,
		     int index)
{
	power9_idle_type(stop_psscr_table[index].val,
			 stop_psscr_table[index].mask);
	return index;
}

/*
 * States for dedicated partition case.
 */
static struct cpuidle_state powernv_states[CPUIDLE_STATE_MAX] = {
	{ /* Snooze */
		.name = "snooze",
		.desc = "snooze",
		.exit_latency = 0,
		.target_residency = 0,
		.enter = snooze_loop },
};

static int powernv_cpuidle_cpu_online(unsigned int cpu)
{
	struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);

	if (dev && cpuidle_get_driver()) {
		cpuidle_pause_and_lock();
		cpuidle_enable_device(dev);
		cpuidle_resume_and_unlock();
	}
	return 0;
}

static int powernv_cpuidle_cpu_dead(unsigned int cpu)
{
	struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);

	if (dev && cpuidle_get_driver()) {
		cpuidle_pause_and_lock();
		cpuidle_disable_device(dev);
		cpuidle_resume_and_unlock();
	}
	return 0;
}

/*
 * powernv_cpuidle_driver_init()
 */
static int powernv_cpuidle_driver_init(void)
{
	int idle_state;
	struct cpuidle_driver *drv = &powernv_idle_driver;

	drv->state_count = 0;

	for (idle_state = 0; idle_state < max_idle_state; ++idle_state) {
		/* Is the state not enabled? */
		if (cpuidle_state_table[idle_state].enter == NULL)
			continue;

		drv->states[drv->state_count] =	/* structure copy */
			cpuidle_state_table[idle_state];

		drv->state_count += 1;
	}

	/*
	 * On the PowerNV platform cpu_present may be less than cpu_possible in
	 * cases when firmware detects the CPU, but it is not available to the
	 * OS.  If CONFIG_HOTPLUG_CPU=n, then such CPUs are not hotplugable at
	 * run time and hence cpu_devices are not created for those CPUs by the
	 * generic topology_init().
	 *
	 * drv->cpumask defaults to cpu_possible_mask in
	 * __cpuidle_driver_init().  This breaks cpuidle on PowerNV where
	 * cpu_devices are not created for CPUs in cpu_possible_mask that
	 * cannot be hot-added later at run time.
	 *
	 * Trying cpuidle_register_device() on a CPU without a cpu_device is
	 * incorrect, so pass a correct CPU mask to the generic cpuidle driver.
	 */

	drv->cpumask = (struct cpumask *)cpu_present_mask;

	return 0;
}

static inline void add_powernv_state(int index, const char *name,
				     unsigned int flags,
				     int (*idle_fn)(struct cpuidle_device *,
						    struct cpuidle_driver *,
						    int),
				     unsigned int target_residency,
				     unsigned int exit_latency,
				     u64 psscr_val, u64 psscr_mask)
{
	strlcpy(powernv_states[index].name, name, CPUIDLE_NAME_LEN);
	strlcpy(powernv_states[index].desc, name, CPUIDLE_NAME_LEN);
	powernv_states[index].flags = flags;
	powernv_states[index].target_residency = target_residency;
	powernv_states[index].exit_latency = exit_latency;
	powernv_states[index].enter = idle_fn;
	stop_psscr_table[index].val = psscr_val;
	stop_psscr_table[index].mask = psscr_mask;
}

/*
 * Returns 0 if prop1_len == prop2_len. Else returns -1
 */
static inline int validate_dt_prop_sizes(const char *prop1, int prop1_len,
					 const char *prop2, int prop2_len)
{
	if (prop1_len == prop2_len)
		return 0;

	pr_warn("cpuidle-powernv: array sizes don't match for %s and %s\n",
		prop1, prop2);
	return -1;
}

extern u32 pnv_get_supported_cpuidle_states(void);
static int powernv_add_idle_states(void)
{
	struct device_node *power_mgt;
	int nr_idle_states = 1; /* Snooze */
	int dt_idle_states, count;
	u32 latency_ns[CPUIDLE_STATE_MAX];
	u32 residency_ns[CPUIDLE_STATE_MAX];
	u32 flags[CPUIDLE_STATE_MAX];
	u64 psscr_val[CPUIDLE_STATE_MAX];
	u64 psscr_mask[CPUIDLE_STATE_MAX];
	const char *names[CPUIDLE_STATE_MAX];
	u32 has_stop_states = 0;
	int i, rc;
	u32 supported_flags = pnv_get_supported_cpuidle_states();


	/* Currently we have snooze statically defined */

	power_mgt = of_find_node_by_path("/ibm,opal/power-mgt");
	if (!power_mgt) {
		pr_warn("opal: PowerMgmt Node not found\n");
		goto out;
	}

	/* Read values of any property to determine the num of idle states */
	dt_idle_states = of_property_count_u32_elems(power_mgt, "ibm,cpu-idle-state-flags");
	if (dt_idle_states < 0) {
		pr_warn("cpuidle-powernv: no idle states found in the DT\n");
		goto out;
	}

	count = of_property_count_u32_elems(power_mgt,
					    "ibm,cpu-idle-state-latencies-ns");

	if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags", dt_idle_states,
				   "ibm,cpu-idle-state-latencies-ns",
				   count) != 0)
		goto out;

	count = of_property_count_strings(power_mgt,
					  "ibm,cpu-idle-state-names");
	if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags", dt_idle_states,
				   "ibm,cpu-idle-state-names",
				   count) != 0)
		goto out;

	/*
	 * Since snooze is used as first idle state, max idle states allowed is
	 * CPUIDLE_STATE_MAX -1
	 */
	if (dt_idle_states > CPUIDLE_STATE_MAX - 1) {
		pr_warn("cpuidle-powernv: discovered idle states more than allowed");
		dt_idle_states = CPUIDLE_STATE_MAX - 1;
	}

	if (of_property_read_u32_array(power_mgt,
			"ibm,cpu-idle-state-flags", flags, dt_idle_states)) {
		pr_warn("cpuidle-powernv : missing ibm,cpu-idle-state-flags in DT\n");
		goto out;
	}

	if (of_property_read_u32_array(power_mgt,
		"ibm,cpu-idle-state-latencies-ns", latency_ns,
		dt_idle_states)) {
		pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-latencies-ns in DT\n");
		goto out;
	}
	if (of_property_read_string_array(power_mgt,
		"ibm,cpu-idle-state-names", names, dt_idle_states) < 0) {
		pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-names in DT\n");
		goto out;
	}

	/*
	 * If the idle states use stop instruction, probe for psscr values
	 * and psscr mask which are necessary to specify required stop level.
	 */
	has_stop_states = (flags[0] &
			   (OPAL_PM_STOP_INST_FAST | OPAL_PM_STOP_INST_DEEP));
	if (has_stop_states) {
		count = of_property_count_u64_elems(power_mgt,
						    "ibm,cpu-idle-state-psscr");
		if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags",
					   dt_idle_states,
					   "ibm,cpu-idle-state-psscr",
					   count) != 0)
			goto out;

		count = of_property_count_u64_elems(power_mgt,
						    "ibm,cpu-idle-state-psscr-mask");
		if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags",
					   dt_idle_states,
					   "ibm,cpu-idle-state-psscr-mask",
					   count) != 0)
			goto out;

		if (of_property_read_u64_array(power_mgt,
		    "ibm,cpu-idle-state-psscr", psscr_val, dt_idle_states)) {
			pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-psscr in DT\n");
			goto out;
		}

		if (of_property_read_u64_array(power_mgt,
					       "ibm,cpu-idle-state-psscr-mask",
						psscr_mask, dt_idle_states)) {
			pr_warn("cpuidle-powernv:Missing ibm,cpu-idle-state-psscr-mask in DT\n");
			goto out;
		}
	}

	count = of_property_count_u32_elems(power_mgt,
					    "ibm,cpu-idle-state-residency-ns");

	if (count < 0) {
		rc = count;
	} else if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags",
					  dt_idle_states,
					  "ibm,cpu-idle-state-residency-ns",
					  count) != 0) {
		goto out;
	} else {
		rc = of_property_read_u32_array(power_mgt,
						"ibm,cpu-idle-state-residency-ns",
						residency_ns, dt_idle_states);
	}

	for (i = 0; i < dt_idle_states; i++) {
		unsigned int exit_latency, target_residency;
		bool stops_timebase = false;

		/*
		 * Skip the platform idle state whose flag isn't in
		 * the supported_cpuidle_states flag mask.
		 */
		if ((flags[i] & supported_flags) != flags[i])
			continue;
		/*
		 * If an idle state has exit latency beyond
		 * POWERNV_THRESHOLD_LATENCY_NS then don't use it
		 * in cpu-idle.
		 */
		if (latency_ns[i] > POWERNV_THRESHOLD_LATENCY_NS)
			continue;
		/*
		 * Firmware passes residency and latency values in ns.
		 * cpuidle expects it in us.
		 */
		exit_latency = latency_ns[i] / 1000;
		if (!rc)
			target_residency = residency_ns[i] / 1000;
		else
			target_residency = 0;

		if (has_stop_states) {
			int err = validate_psscr_val_mask(&psscr_val[i],
							  &psscr_mask[i],
							  flags[i]);
			if (err) {
				report_invalid_psscr_val(psscr_val[i], err);
				continue;
			}
		}

		if (flags[i] & OPAL_PM_TIMEBASE_STOP)
			stops_timebase = true;

		/*
		 * For nap and fastsleep, use default target_residency
		 * values if f/w does not expose it.
		 */
		if (flags[i] & OPAL_PM_NAP_ENABLED) {
			if (!rc)
				target_residency = 100;
			/* Add NAP state */
			add_powernv_state(nr_idle_states, "Nap",
					  CPUIDLE_FLAG_NONE, nap_loop,
					  target_residency, exit_latency, 0, 0);
		} else if (has_stop_states && !stops_timebase) {
			add_powernv_state(nr_idle_states, names[i],
					  CPUIDLE_FLAG_NONE, stop_loop,
					  target_residency, exit_latency,
					  psscr_val[i], psscr_mask[i]);
		}

		/*
		 * All cpuidle states with CPUIDLE_FLAG_TIMER_STOP set must come
		 * within this config dependency check.
		 */
#ifdef CONFIG_TICK_ONESHOT
		else if (flags[i] & OPAL_PM_SLEEP_ENABLED ||
			 flags[i] & OPAL_PM_SLEEP_ENABLED_ER1) {
			if (!rc)
				target_residency = 300000;
			/* Add FASTSLEEP state */
			add_powernv_state(nr_idle_states, "FastSleep",
					  CPUIDLE_FLAG_TIMER_STOP,
					  fastsleep_loop,
					  target_residency, exit_latency, 0, 0);
		} else if (has_stop_states && stops_timebase) {
			add_powernv_state(nr_idle_states, names[i],
					  CPUIDLE_FLAG_TIMER_STOP, stop_loop,
					  target_residency, exit_latency,
					  psscr_val[i], psscr_mask[i]);
		}
#endif
		else
			continue;
		nr_idle_states++;
	}
out:
	return nr_idle_states;
}

/*
 * powernv_idle_probe()
 * Choose state table for shared versus dedicated partition
 */
static int powernv_idle_probe(void)
{
	if (cpuidle_disable != IDLE_NO_OVERRIDE)
		return -ENODEV;

	if (firmware_has_feature(FW_FEATURE_OPAL)) {
		cpuidle_state_table = powernv_states;
		/* Device tree can indicate more idle states */
		max_idle_state = powernv_add_idle_states();
		if (max_idle_state > 1) {
			snooze_timeout_en = true;
			snooze_timeout = powernv_states[1].target_residency *
					 tb_ticks_per_usec;
		}
 	} else
 		return -ENODEV;

	return 0;
}

static int __init powernv_processor_idle_init(void)
{
	int retval;

	retval = powernv_idle_probe();
	if (retval)
		return retval;

	powernv_cpuidle_driver_init();
	retval = cpuidle_register(&powernv_idle_driver, NULL);
	if (retval) {
		printk(KERN_DEBUG "Registration of powernv driver failed.\n");
		return retval;
	}

	retval = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
					   "cpuidle/powernv:online",
					   powernv_cpuidle_cpu_online, NULL);
	WARN_ON(retval < 0);
	retval = cpuhp_setup_state_nocalls(CPUHP_CPUIDLE_DEAD,
					   "cpuidle/powernv:dead", NULL,
					   powernv_cpuidle_cpu_dead);
	WARN_ON(retval < 0);
	printk(KERN_DEBUG "powernv_idle_driver registered\n");
	return 0;
}

device_initcall(powernv_processor_idle_init);
Commit	Line	Data
b2441318	1	// SPDX-License-Identifier: GPL-2.0
2c2e6ecf DD	2	/*
	3	* cpuidle-powernv - idle state cpuidle driver.
	4	* Adapted from drivers/cpuidle/cpuidle-pseries
	5	*
	6	*/
	7
	8	#include <linux/kernel.h>
	9	#include <linux/module.h>
	10	#include <linux/init.h>
	11	#include <linux/moduleparam.h>
	12	#include <linux/cpuidle.h>
	13	#include <linux/cpu.h>
	14	#include <linux/notifier.h>
0d948730	15	#include <linux/clockchips.h>
0888839c	16	#include <linux/of.h>
92c83ff5	17	#include <linux/slab.h>
2c2e6ecf DD	18
	19	#include <asm/machdep.h>
	20	#include <asm/firmware.h>
8eb8ac89	21	#include <asm/opal.h>
591ac0cb	22	#include <asm/runlatch.h>
09206b60	23	#include <asm/cpuidle.h>
2c2e6ecf	24
9e9fc6f0 GS	25	/*
	26	* Expose only those Hardware idle states via the cpuidle framework
	27	* that have latency value below POWERNV_THRESHOLD_LATENCY_NS.
	28	*/
3005c597 SP	29	#define POWERNV_THRESHOLD_LATENCY_NS 200000
3005c597 SP	30
ed61390b	31	static struct cpuidle_driver powernv_idle_driver = {
2c2e6ecf DD	32	.name = "powernv_idle",
	33	.owner = THIS_MODULE,
	34	};
	35
624e46d0 NP	36	static int max_idle_state __read_mostly;
624e46d0 NP	37	static struct cpuidle_state *cpuidle_state_table __read_mostly;
3005c597	38
09206b60 GS	39	struct stop_psscr_table {
	40	u64 val;
	41	u64 mask;
	42	};
	43
624e46d0	44	static struct stop_psscr_table stop_psscr_table[CPUIDLE_STATE_MAX] __read_mostly;
3005c597	45
624e46d0 NP	46	static u64 snooze_timeout __read_mostly;
624e46d0 NP	47	static bool snooze_timeout_en __read_mostly;
2c2e6ecf DD	48
	49	static int snooze_loop(struct cpuidle_device *dev,
	50	struct cpuidle_driver *drv,
	51	int index)
	52	{
78eaa10f SB	53	u64 snooze_exit_time;
78eaa10f SB	54
2c2e6ecf DD	55	set_thread_flag(TIF_POLLING_NRFLAG);
2c2e6ecf DD	56
3fc5ee92 NP	57	local_irq_enable();
3fc5ee92 NP	58
78eaa10f	59	snooze_exit_time = get_tb() + snooze_timeout;
591ac0cb	60	ppc64_runlatch_off();
26eb48a9	61	HMT_very_low();
2c2e6ecf	62	while (!need_resched()) {
7ded4291 NP	63	if (likely(snooze_timeout_en) && get_tb() > snooze_exit_time) {
	64	/*
	65	* Task has not woken up but we are exiting the polling
	66	* loop anyway. Require a barrier after polling is
	67	* cleared to order subsequent test of need_resched().
	68	*/
	69	clear_thread_flag(TIF_POLLING_NRFLAG);
	70	smp_mb();
78eaa10f	71	break;
7ded4291	72	}
2c2e6ecf DD	73	}
	74
	75	HMT_medium();
591ac0cb	76	ppc64_runlatch_on();
2c2e6ecf	77	clear_thread_flag(TIF_POLLING_NRFLAG);
3fc5ee92	78
2c2e6ecf DD	79	return index;
	80	}
	81
	82	static int nap_loop(struct cpuidle_device *dev,
	83	struct cpuidle_driver *drv,
	84	int index)
	85	{
2201f994 NP	86	power7_idle_type(PNV_THREAD_NAP);
2201f994 NP	87
2c2e6ecf DD	88	return index;
	89	}
	90
cc5a2f7b	91	/* Register for fastsleep only in oneshot mode of broadcast */
cc5a2f7b	92	#ifdef CONFIG_TICK_ONESHOT
0d948730 PM	93	static int fastsleep_loop(struct cpuidle_device *dev,
	94	struct cpuidle_driver *drv,
	95	int index)
	96	{
	97	unsigned long old_lpcr = mfspr(SPRN_LPCR);
	98	unsigned long new_lpcr;
	99
	100	if (unlikely(system_state < SYSTEM_RUNNING))
	101	return index;
	102
	103	new_lpcr = old_lpcr;
9b6a68d9 MN	104	/* Do not exit powersave upon decrementer as we've setup the timer
9b6a68d9 MN	105	* offload.
0d948730	106	*/
9b6a68d9	107	new_lpcr &= ~LPCR_PECE1;
0d948730 PM	108
0d948730 PM	109	mtspr(SPRN_LPCR, new_lpcr);
2201f994 NP	110
2201f994 NP	111	power7_idle_type(PNV_THREAD_SLEEP);
0d948730 PM	112
	113	mtspr(SPRN_LPCR, old_lpcr);
	114
	115	return index;
	116	}
cc5a2f7b	117	#endif
3005c597 SP	118
	119	static int stop_loop(struct cpuidle_device *dev,
	120	struct cpuidle_driver *drv,
	121	int index)
	122	{
2201f994	123	power9_idle_type(stop_psscr_table[index].val,
09206b60	124	stop_psscr_table[index].mask);
3005c597 SP	125	return index;
	126	}
	127
2c2e6ecf DD	128	/*
	129	* States for dedicated partition case.
	130	*/
169f3fae	131	static struct cpuidle_state powernv_states[CPUIDLE_STATE_MAX] = {
2c2e6ecf DD	132	{ /* Snooze */
	133	.name = "snooze",
	134	.desc = "snooze",
2c2e6ecf DD	135	.exit_latency = 0,
2c2e6ecf DD	136	.target_residency = 0,
957efced	137	.enter = snooze_loop },
2c2e6ecf DD	138	};
2c2e6ecf DD	139
10fcca9d	140	static int powernv_cpuidle_cpu_online(unsigned int cpu)
2c2e6ecf	141	{
10fcca9d	142	struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);
2c2e6ecf DD	143
2c2e6ecf DD	144	if (dev && cpuidle_get_driver()) {
10fcca9d SAS	145	cpuidle_pause_and_lock();
	146	cpuidle_enable_device(dev);
	147	cpuidle_resume_and_unlock();
	148	}
	149	return 0;
	150	}
2c2e6ecf	151
10fcca9d SAS	152	static int powernv_cpuidle_cpu_dead(unsigned int cpu)
	153	{
	154	struct cpuidle_device *dev = per_cpu(cpuidle_devices, cpu);
2c2e6ecf	155
10fcca9d SAS	156	if (dev && cpuidle_get_driver()) {
	157	cpuidle_pause_and_lock();
	158	cpuidle_disable_device(dev);
	159	cpuidle_resume_and_unlock();
2c2e6ecf	160	}
10fcca9d	161	return 0;
2c2e6ecf DD	162	}
2c2e6ecf DD	163
2c2e6ecf DD	164	/*
	165	* powernv_cpuidle_driver_init()
	166	*/
	167	static int powernv_cpuidle_driver_init(void)
	168	{
	169	int idle_state;
	170	struct cpuidle_driver *drv = &powernv_idle_driver;
	171
	172	drv->state_count = 0;
	173
	174	for (idle_state = 0; idle_state < max_idle_state; ++idle_state) {
	175	/* Is the state not enabled? */
	176	if (cpuidle_state_table[idle_state].enter == NULL)
	177	continue;
	178
	179	drv->states[drv->state_count] = /* structure copy */
	180	cpuidle_state_table[idle_state];
	181
	182	drv->state_count += 1;
	183	}
	184
293d264f VS	185	/*
	186	* On the PowerNV platform cpu_present may be less than cpu_possible in
	187	* cases when firmware detects the CPU, but it is not available to the
	188	* OS. If CONFIG_HOTPLUG_CPU=n, then such CPUs are not hotplugable at
	189	* run time and hence cpu_devices are not created for those CPUs by the
	190	* generic topology_init().
	191	*
	192	* drv->cpumask defaults to cpu_possible_mask in
	193	* __cpuidle_driver_init(). This breaks cpuidle on PowerNV where
	194	* cpu_devices are not created for CPUs in cpu_possible_mask that
	195	* cannot be hot-added later at run time.
	196	*
	197	* Trying cpuidle_register_device() on a CPU without a cpu_device is
	198	* incorrect, so pass a correct CPU mask to the generic cpuidle driver.
	199	*/
	200
	201	drv->cpumask = (struct cpumask *)cpu_present_mask;
	202
2c2e6ecf DD	203	return 0;
	204	}
	205
9e9fc6f0 GS	206	static inline void add_powernv_state(int index, const char *name,
	207	unsigned int flags,
	208	int (idle_fn)(struct cpuidle_device ,
	209	struct cpuidle_driver *,
	210	int),
	211	unsigned int target_residency,
	212	unsigned int exit_latency,
09206b60	213	u64 psscr_val, u64 psscr_mask)
9e9fc6f0 GS	214	{
	215	strlcpy(powernv_states[index].name, name, CPUIDLE_NAME_LEN);
	216	strlcpy(powernv_states[index].desc, name, CPUIDLE_NAME_LEN);
	217	powernv_states[index].flags = flags;
	218	powernv_states[index].target_residency = target_residency;
	219	powernv_states[index].exit_latency = exit_latency;
	220	powernv_states[index].enter = idle_fn;
09206b60 GS	221	stop_psscr_table[index].val = psscr_val;
09206b60 GS	222	stop_psscr_table[index].mask = psscr_mask;
9e9fc6f0 GS	223	}
9e9fc6f0 GS	224
ecad4502 GS	225	/*
	226	* Returns 0 if prop1_len == prop2_len. Else returns -1
	227	*/
	228	static inline int validate_dt_prop_sizes(const char *prop1, int prop1_len,
	229	const char *prop2, int prop2_len)
	230	{
	231	if (prop1_len == prop2_len)
	232	return 0;
	233
	234	pr_warn("cpuidle-powernv: array sizes don't match for %s and %s\n",
	235	prop1, prop2);
	236	return -1;
	237	}
	238
785a12af	239	extern u32 pnv_get_supported_cpuidle_states(void);
0888839c PM	240	static int powernv_add_idle_states(void)
	241	{
	242	struct device_node *power_mgt;
0888839c	243	int nr_idle_states = 1; /* Snooze */
ecad4502	244	int dt_idle_states, count;
957efced SP	245	u32 latency_ns[CPUIDLE_STATE_MAX];
	246	u32 residency_ns[CPUIDLE_STATE_MAX];
	247	u32 flags[CPUIDLE_STATE_MAX];
3005c597	248	u64 psscr_val[CPUIDLE_STATE_MAX];
09206b60	249	u64 psscr_mask[CPUIDLE_STATE_MAX];
3005c597	250	const char *names[CPUIDLE_STATE_MAX];
09206b60	251	u32 has_stop_states = 0;
92c83ff5	252	int i, rc;
785a12af GS	253	u32 supported_flags = pnv_get_supported_cpuidle_states();
785a12af GS	254
0888839c PM	255
	256	/* Currently we have snooze statically defined */
	257
	258	power_mgt = of_find_node_by_path("/ibm,opal/power-mgt");
	259	if (!power_mgt) {
	260	pr_warn("opal: PowerMgmt Node not found\n");
92c83ff5	261	goto out;
0888839c PM	262	}
0888839c PM	263
70734a78 PM	264	/* Read values of any property to determine the num of idle states */
	265	dt_idle_states = of_property_count_u32_elems(power_mgt, "ibm,cpu-idle-state-flags");
	266	if (dt_idle_states < 0) {
	267	pr_warn("cpuidle-powernv: no idle states found in the DT\n");
92c83ff5	268	goto out;
0888839c PM	269	}
0888839c PM	270
ecad4502 GS	271	count = of_property_count_u32_elems(power_mgt,
	272	"ibm,cpu-idle-state-latencies-ns");
	273
	274	if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags", dt_idle_states,
	275	"ibm,cpu-idle-state-latencies-ns",
	276	count) != 0)
	277	goto out;
	278
	279	count = of_property_count_strings(power_mgt,
	280	"ibm,cpu-idle-state-names");
	281	if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags", dt_idle_states,
	282	"ibm,cpu-idle-state-names",
	283	count) != 0)
	284	goto out;
	285
957efced SP	286	/*
	287	* Since snooze is used as first idle state, max idle states allowed is
	288	* CPUIDLE_STATE_MAX -1
	289	*/
	290	if (dt_idle_states > CPUIDLE_STATE_MAX - 1) {
	291	pr_warn("cpuidle-powernv: discovered idle states more than allowed");
	292	dt_idle_states = CPUIDLE_STATE_MAX - 1;
	293	}
	294
70734a78 PM	295	if (of_property_read_u32_array(power_mgt,
	296	"ibm,cpu-idle-state-flags", flags, dt_idle_states)) {
	297	pr_warn("cpuidle-powernv : missing ibm,cpu-idle-state-flags in DT\n");
957efced	298	goto out;
70734a78	299	}
92c83ff5	300
957efced SP	301	if (of_property_read_u32_array(power_mgt,
	302	"ibm,cpu-idle-state-latencies-ns", latency_ns,
	303	dt_idle_states)) {
92c83ff5	304	pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-latencies-ns in DT\n");
957efced	305	goto out;
74aa51b5	306	}
3005c597 SP	307	if (of_property_read_string_array(power_mgt,
	308	"ibm,cpu-idle-state-names", names, dt_idle_states) < 0) {
	309	pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-names in DT\n");
	310	goto out;
	311	}
	312
	313	/*
	314	* If the idle states use stop instruction, probe for psscr values
09206b60	315	* and psscr mask which are necessary to specify required stop level.
3005c597	316	*/
09206b60 GS	317	has_stop_states = (flags[0] &
	318	(OPAL_PM_STOP_INST_FAST \| OPAL_PM_STOP_INST_DEEP));
	319	if (has_stop_states) {
ecad4502 GS	320	count = of_property_count_u64_elems(power_mgt,
	321	"ibm,cpu-idle-state-psscr");
	322	if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags",
	323	dt_idle_states,
	324	"ibm,cpu-idle-state-psscr",
	325	count) != 0)
	326	goto out;
	327
	328	count = of_property_count_u64_elems(power_mgt,
	329	"ibm,cpu-idle-state-psscr-mask");
	330	if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags",
	331	dt_idle_states,
	332	"ibm,cpu-idle-state-psscr-mask",
	333	count) != 0)
	334	goto out;
	335
3005c597 SP	336	if (of_property_read_u64_array(power_mgt,
3005c597 SP	337	"ibm,cpu-idle-state-psscr", psscr_val, dt_idle_states)) {
09206b60	338	pr_warn("cpuidle-powernv: missing ibm,cpu-idle-state-psscr in DT\n");
3005c597 SP	339	goto out;
3005c597 SP	340	}
74aa51b5	341
09206b60 GS	342	if (of_property_read_u64_array(power_mgt,
	343	"ibm,cpu-idle-state-psscr-mask",
	344	psscr_mask, dt_idle_states)) {
	345	pr_warn("cpuidle-powernv:Missing ibm,cpu-idle-state-psscr-mask in DT\n");
	346	goto out;
	347	}
	348	}
	349
ecad4502 GS	350	count = of_property_count_u32_elems(power_mgt,
	351	"ibm,cpu-idle-state-residency-ns");
	352
	353	if (count < 0) {
	354	rc = count;
	355	} else if (validate_dt_prop_sizes("ibm,cpu-idle-state-flags",
	356	dt_idle_states,
	357	"ibm,cpu-idle-state-residency-ns",
	358	count) != 0) {
	359	goto out;
	360	} else {
	361	rc = of_property_read_u32_array(power_mgt,
	362	"ibm,cpu-idle-state-residency-ns",
	363	residency_ns, dt_idle_states);
	364	}
0888839c PM	365
0888839c PM	366	for (i = 0; i < dt_idle_states; i++) {
9e9fc6f0	367	unsigned int exit_latency, target_residency;
f9122ee4	368	bool stops_timebase = false;
785a12af GS	369
	370	/*
	371	* Skip the platform idle state whose flag isn't in
	372	* the supported_cpuidle_states flag mask.
	373	*/
	374	if ((flags[i] & supported_flags) != flags[i])
	375	continue;
3005c597 SP	376	/*
	377	* If an idle state has exit latency beyond
	378	* POWERNV_THRESHOLD_LATENCY_NS then don't use it
	379	* in cpu-idle.
	380	*/
	381	if (latency_ns[i] > POWERNV_THRESHOLD_LATENCY_NS)
	382	continue;
9e9fc6f0 GS	383	/*
	384	* Firmware passes residency and latency values in ns.
	385	* cpuidle expects it in us.
	386	*/
	387	exit_latency = latency_ns[i] / 1000;
	388	if (!rc)
	389	target_residency = residency_ns[i] / 1000;
	390	else
	391	target_residency = 0;
0888839c	392
09206b60 GS	393	if (has_stop_states) {
	394	int err = validate_psscr_val_mask(&psscr_val[i],
	395	&psscr_mask[i],
	396	flags[i]);
	397	if (err) {
	398	report_invalid_psscr_val(psscr_val[i], err);
	399	continue;
	400	}
	401	}
	402
f9122ee4 GS	403	if (flags[i] & OPAL_PM_TIMEBASE_STOP)
	404	stops_timebase = true;
	405
92c83ff5	406	/*
9e9fc6f0 GS	407	* For nap and fastsleep, use default target_residency
9e9fc6f0 GS	408	* values if f/w does not expose it.
74aa51b5	409	*/
70734a78	410	if (flags[i] & OPAL_PM_NAP_ENABLED) {
9e9fc6f0 GS	411	if (!rc)
9e9fc6f0 GS	412	target_residency = 100;
0888839c	413	/* Add NAP state */
9e9fc6f0 GS	414	add_powernv_state(nr_idle_states, "Nap",
9e9fc6f0 GS	415	CPUIDLE_FLAG_NONE, nap_loop,
09206b60	416	target_residency, exit_latency, 0, 0);
f9122ee4	417	} else if (has_stop_states && !stops_timebase) {
9e9fc6f0 GS	418	add_powernv_state(nr_idle_states, names[i],
	419	CPUIDLE_FLAG_NONE, stop_loop,
	420	target_residency, exit_latency,
09206b60	421	psscr_val[i], psscr_mask[i]);
cc5a2f7b	422	}
	423
	424	/*
	425	* All cpuidle states with CPUIDLE_FLAG_TIMER_STOP set must come
	426	* within this config dependency check.
	427	*/
	428	#ifdef CONFIG_TICK_ONESHOT
f9122ee4 GS	429	else if (flags[i] & OPAL_PM_SLEEP_ENABLED \|\|
f9122ee4 GS	430	flags[i] & OPAL_PM_SLEEP_ENABLED_ER1) {
9e9fc6f0 GS	431	if (!rc)
9e9fc6f0 GS	432	target_residency = 300000;
0888839c	433	/* Add FASTSLEEP state */
9e9fc6f0 GS	434	add_powernv_state(nr_idle_states, "FastSleep",
	435	CPUIDLE_FLAG_TIMER_STOP,
	436	fastsleep_loop,
09206b60	437	target_residency, exit_latency, 0, 0);
f9122ee4	438	} else if (has_stop_states && stops_timebase) {
9e9fc6f0 GS	439	add_powernv_state(nr_idle_states, names[i],
	440	CPUIDLE_FLAG_TIMER_STOP, stop_loop,
	441	target_residency, exit_latency,
09206b60	442	psscr_val[i], psscr_mask[i]);
0888839c	443	}
cc5a2f7b	444	#endif
f9122ee4 GS	445	else
f9122ee4 GS	446	continue;
92c83ff5	447	nr_idle_states++;
0888839c	448	}
92c83ff5	449	out:
0888839c PM	450	return nr_idle_states;
	451	}
	452
2c2e6ecf DD	453	/*
	454	* powernv_idle_probe()
	455	* Choose state table for shared versus dedicated partition
	456	*/
	457	static int powernv_idle_probe(void)
	458	{
2c2e6ecf DD	459	if (cpuidle_disable != IDLE_NO_OVERRIDE)
	460	return -ENODEV;
	461
e4d54f71	462	if (firmware_has_feature(FW_FEATURE_OPAL)) {
2c2e6ecf	463	cpuidle_state_table = powernv_states;
0888839c PM	464	/* Device tree can indicate more idle states */
0888839c PM	465	max_idle_state = powernv_add_idle_states();
78eaa10f SB	466	if (max_idle_state > 1) {
	467	snooze_timeout_en = true;
	468	snooze_timeout = powernv_states[1].target_residency *
	469	tb_ticks_per_usec;
	470	}
2c2e6ecf DD	471	} else
	472	return -ENODEV;
	473
	474	return 0;
	475	}
	476
	477	static int __init powernv_processor_idle_init(void)
	478	{
	479	int retval;
	480
	481	retval = powernv_idle_probe();
	482	if (retval)
	483	return retval;
	484
	485	powernv_cpuidle_driver_init();
	486	retval = cpuidle_register(&powernv_idle_driver, NULL);
	487	if (retval) {
	488	printk(KERN_DEBUG "Registration of powernv driver failed.\n");
	489	return retval;
	490	}
	491
10fcca9d SAS	492	retval = cpuhp_setup_state_nocalls(CPUHP_AP_ONLINE_DYN,
	493	"cpuidle/powernv:online",
	494	powernv_cpuidle_cpu_online, NULL);
	495	WARN_ON(retval < 0);
	496	retval = cpuhp_setup_state_nocalls(CPUHP_CPUIDLE_DEAD,
	497	"cpuidle/powernv:dead", NULL,
	498	powernv_cpuidle_cpu_dead);
	499	WARN_ON(retval < 0);
2c2e6ecf DD	500	printk(KERN_DEBUG "powernv_idle_driver registered\n");
	501	return 0;
	502	}
	503
	504	device_initcall(powernv_processor_idle_init);