[mirror_ubuntu-bionic-kernel.git] / drivers / base / arch_topology.c

/*
 * Arch specific cpu topology information
 *
 * Copyright (C) 2016, ARM Ltd.
 * Written by: Juri Lelli, ARM Ltd.
 *
 * This file is subject to the terms and conditions of the GNU General Public
 * License.  See the file "COPYING" in the main directory of this archive
 * for more details.
 *
 * Released under the GPLv2 only.
 * SPDX-License-Identifier: GPL-2.0
 */

#include <linux/acpi.h>
#include <linux/arch_topology.h>
#include <linux/cpu.h>
#include <linux/cpufreq.h>
#include <linux/device.h>
#include <linux/of.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/sched/topology.h>

DEFINE_PER_CPU(unsigned long, freq_scale) = SCHED_CAPACITY_SCALE;

void arch_set_freq_scale(struct cpumask *cpus, unsigned long cur_freq,
			 unsigned long max_freq)
{
	unsigned long scale;
	int i;

	scale = (cur_freq << SCHED_CAPACITY_SHIFT) / max_freq;

	for_each_cpu(i, cpus)
		per_cpu(freq_scale, i) = scale;
}

static DEFINE_MUTEX(cpu_scale_mutex);
DEFINE_PER_CPU(unsigned long, cpu_scale) = SCHED_CAPACITY_SCALE;

void topology_set_cpu_scale(unsigned int cpu, unsigned long capacity)
{
	per_cpu(cpu_scale, cpu) = capacity;
}

static ssize_t cpu_capacity_show(struct device *dev,
				 struct device_attribute *attr,
				 char *buf)
{
	struct cpu *cpu = container_of(dev, struct cpu, dev);

	return sprintf(buf, "%lu\n", topology_get_cpu_scale(NULL, cpu->dev.id));
}

static ssize_t cpu_capacity_store(struct device *dev,
				  struct device_attribute *attr,
				  const char *buf,
				  size_t count)
{
	struct cpu *cpu = container_of(dev, struct cpu, dev);
	int this_cpu = cpu->dev.id;
	int i;
	unsigned long new_capacity;
	ssize_t ret;

	if (!count)
		return 0;

	ret = kstrtoul(buf, 0, &new_capacity);
	if (ret)
		return ret;
	if (new_capacity > SCHED_CAPACITY_SCALE)
		return -EINVAL;

	mutex_lock(&cpu_scale_mutex);
	for_each_cpu(i, &cpu_topology[this_cpu].core_sibling)
		topology_set_cpu_scale(i, new_capacity);
	mutex_unlock(&cpu_scale_mutex);

	return count;
}

static DEVICE_ATTR_RW(cpu_capacity);

static int register_cpu_capacity_sysctl(void)
{
	int i;
	struct device *cpu;

	for_each_possible_cpu(i) {
		cpu = get_cpu_device(i);
		if (!cpu) {
			pr_err("%s: too early to get CPU%d device!\n",
			       __func__, i);
			continue;
		}
		device_create_file(cpu, &dev_attr_cpu_capacity);
	}

	return 0;
}
subsys_initcall(register_cpu_capacity_sysctl);

static u32 capacity_scale;
static u32 *raw_capacity;

static int free_raw_capacity(void)
{
	kfree(raw_capacity);
	raw_capacity = NULL;

	return 0;
}

void topology_normalize_cpu_scale(void)
{
	u64 capacity;
	int cpu;

	if (!raw_capacity)
		return;

	pr_debug("cpu_capacity: capacity_scale=%u\n", capacity_scale);
	mutex_lock(&cpu_scale_mutex);
	for_each_possible_cpu(cpu) {
		pr_debug("cpu_capacity: cpu=%d raw_capacity=%u\n",
			 cpu, raw_capacity[cpu]);
		capacity = (raw_capacity[cpu] << SCHED_CAPACITY_SHIFT)
			/ capacity_scale;
		topology_set_cpu_scale(cpu, capacity);
		pr_debug("cpu_capacity: CPU%d cpu_capacity=%lu\n",
			cpu, topology_get_cpu_scale(NULL, cpu));
	}
	mutex_unlock(&cpu_scale_mutex);
}

bool __init topology_parse_cpu_capacity(struct device_node *cpu_node, int cpu)
{
	static bool cap_parsing_failed;
	int ret;
	u32 cpu_capacity;

	if (cap_parsing_failed)
		return false;

	ret = of_property_read_u32(cpu_node, "capacity-dmips-mhz",
				   &cpu_capacity);
	if (!ret) {
		if (!raw_capacity) {
			raw_capacity = kcalloc(num_possible_cpus(),
					       sizeof(*raw_capacity),
					       GFP_KERNEL);
			if (!raw_capacity) {
				pr_err("cpu_capacity: failed to allocate memory for raw capacities\n");
				cap_parsing_failed = true;
				return false;
			}
		}
		capacity_scale = max(cpu_capacity, capacity_scale);
		raw_capacity[cpu] = cpu_capacity;
		pr_debug("cpu_capacity: %pOF cpu_capacity=%u (raw)\n",
			cpu_node, raw_capacity[cpu]);
	} else {
		if (raw_capacity) {
			pr_err("cpu_capacity: missing %pOF raw capacity\n",
				cpu_node);
			pr_err("cpu_capacity: partial information: fallback to 1024 for all CPUs\n");
		}
		cap_parsing_failed = true;
		free_raw_capacity();
	}

	return !ret;
}

#ifdef CONFIG_CPU_FREQ
static cpumask_var_t cpus_to_visit;
static void parsing_done_workfn(struct work_struct *work);
static DECLARE_WORK(parsing_done_work, parsing_done_workfn);

static int
init_cpu_capacity_callback(struct notifier_block *nb,
			   unsigned long val,
			   void *data)
{
	struct cpufreq_policy *policy = data;
	int cpu;

	if (!raw_capacity)
		return 0;

	if (val != CPUFREQ_NOTIFY)
		return 0;

	pr_debug("cpu_capacity: init cpu capacity for CPUs [%*pbl] (to_visit=%*pbl)\n",
		 cpumask_pr_args(policy->related_cpus),
		 cpumask_pr_args(cpus_to_visit));

	cpumask_andnot(cpus_to_visit, cpus_to_visit, policy->related_cpus);

	for_each_cpu(cpu, policy->related_cpus) {
		raw_capacity[cpu] = topology_get_cpu_scale(NULL, cpu) *
				    policy->cpuinfo.max_freq / 1000UL;
		capacity_scale = max(raw_capacity[cpu], capacity_scale);
	}

	if (cpumask_empty(cpus_to_visit)) {
		topology_normalize_cpu_scale();
		free_raw_capacity();
		pr_debug("cpu_capacity: parsing done\n");
		schedule_work(&parsing_done_work);
	}

	return 0;
}

static struct notifier_block init_cpu_capacity_notifier = {
	.notifier_call = init_cpu_capacity_callback,
};

static int __init register_cpufreq_notifier(void)
{
	int ret;

	/*
	 * on ACPI-based systems we need to use the default cpu capacity
	 * until we have the necessary code to parse the cpu capacity, so
	 * skip registering cpufreq notifier.
	 */
	if (!acpi_disabled || !raw_capacity)
		return -EINVAL;

	if (!alloc_cpumask_var(&cpus_to_visit, GFP_KERNEL)) {
		pr_err("cpu_capacity: failed to allocate memory for cpus_to_visit\n");
		return -ENOMEM;
	}

	cpumask_copy(cpus_to_visit, cpu_possible_mask);

	ret = cpufreq_register_notifier(&init_cpu_capacity_notifier,
					CPUFREQ_POLICY_NOTIFIER);

	if (ret)
		free_cpumask_var(cpus_to_visit);

	return ret;
}
core_initcall(register_cpufreq_notifier);

static void parsing_done_workfn(struct work_struct *work)
{
	cpufreq_unregister_notifier(&init_cpu_capacity_notifier,
					 CPUFREQ_POLICY_NOTIFIER);
	free_cpumask_var(cpus_to_visit);
}

#else
core_initcall(free_raw_capacity);
#endif
Commit	Line	Data
2ef7a295 JL	1	/*
	2	* Arch specific cpu topology information
	3	*
	4	* Copyright (C) 2016, ARM Ltd.
	5	* Written by: Juri Lelli, ARM Ltd.
	6	*
	7	* This file is subject to the terms and conditions of the GNU General Public
	8	* License. See the file "COPYING" in the main directory of this archive
	9	* for more details.
	10	*
	11	* Released under the GPLv2 only.
	12	* SPDX-License-Identifier: GPL-2.0
	13	*/
	14
	15	#include <linux/acpi.h>
615ffd63	16	#include <linux/arch_topology.h>
2ef7a295 JL	17	#include <linux/cpu.h>
	18	#include <linux/cpufreq.h>
	19	#include <linux/device.h>
	20	#include <linux/of.h>
	21	#include <linux/slab.h>
	22	#include <linux/string.h>
	23	#include <linux/sched/topology.h>
	24
0e27c567	25	DEFINE_PER_CPU(unsigned long, freq_scale) = SCHED_CAPACITY_SCALE;
2ef7a295	26
0e27c567 DE	27	void arch_set_freq_scale(struct cpumask *cpus, unsigned long cur_freq,
0e27c567 DE	28	unsigned long max_freq)
2ef7a295	29	{
0e27c567 DE	30	unsigned long scale;
	31	int i;
	32
	33	scale = (cur_freq << SCHED_CAPACITY_SHIFT) / max_freq;
	34
	35	for_each_cpu(i, cpus)
	36	per_cpu(freq_scale, i) = scale;
2ef7a295 JL	37	}
2ef7a295 JL	38
2ef7a295	39	static DEFINE_MUTEX(cpu_scale_mutex);
8216f588	40	DEFINE_PER_CPU(unsigned long, cpu_scale) = SCHED_CAPACITY_SCALE;
2ef7a295	41
4ca4f26a	42	void topology_set_cpu_scale(unsigned int cpu, unsigned long capacity)
2ef7a295 JL	43	{
	44	per_cpu(cpu_scale, cpu) = capacity;
	45	}
	46
	47	static ssize_t cpu_capacity_show(struct device *dev,
	48	struct device_attribute *attr,
	49	char *buf)
	50	{
	51	struct cpu *cpu = container_of(dev, struct cpu, dev);
	52
3eeba1a2	53	return sprintf(buf, "%lu\n", topology_get_cpu_scale(NULL, cpu->dev.id));
2ef7a295 JL	54	}
	55
	56	static ssize_t cpu_capacity_store(struct device *dev,
	57	struct device_attribute *attr,
	58	const char *buf,
	59	size_t count)
	60	{
	61	struct cpu *cpu = container_of(dev, struct cpu, dev);
	62	int this_cpu = cpu->dev.id;
	63	int i;
	64	unsigned long new_capacity;
	65	ssize_t ret;
	66
	67	if (!count)
	68	return 0;
	69
	70	ret = kstrtoul(buf, 0, &new_capacity);
	71	if (ret)
	72	return ret;
	73	if (new_capacity > SCHED_CAPACITY_SCALE)
	74	return -EINVAL;
	75
	76	mutex_lock(&cpu_scale_mutex);
	77	for_each_cpu(i, &cpu_topology[this_cpu].core_sibling)
4ca4f26a	78	topology_set_cpu_scale(i, new_capacity);
2ef7a295 JL	79	mutex_unlock(&cpu_scale_mutex);
	80
	81	return count;
	82	}
	83
	84	static DEVICE_ATTR_RW(cpu_capacity);
	85
	86	static int register_cpu_capacity_sysctl(void)
	87	{
	88	int i;
	89	struct device *cpu;
	90
	91	for_each_possible_cpu(i) {
	92	cpu = get_cpu_device(i);
	93	if (!cpu) {
	94	pr_err("%s: too early to get CPU%d device!\n",
	95	__func__, i);
	96	continue;
	97	}
	98	device_create_file(cpu, &dev_attr_cpu_capacity);
	99	}
	100
	101	return 0;
	102	}
	103	subsys_initcall(register_cpu_capacity_sysctl);
	104
	105	static u32 capacity_scale;
	106	static u32 *raw_capacity;
62de1161	107
82d8ba71	108	static int free_raw_capacity(void)
62de1161 VK	109	{
	110	kfree(raw_capacity);
	111	raw_capacity = NULL;
	112
	113	return 0;
	114	}
2ef7a295	115
4ca4f26a	116	void topology_normalize_cpu_scale(void)
2ef7a295 JL	117	{
	118	u64 capacity;
	119	int cpu;
	120
62de1161	121	if (!raw_capacity)
2ef7a295 JL	122	return;
	123
	124	pr_debug("cpu_capacity: capacity_scale=%u\n", capacity_scale);
	125	mutex_lock(&cpu_scale_mutex);
	126	for_each_possible_cpu(cpu) {
	127	pr_debug("cpu_capacity: cpu=%d raw_capacity=%u\n",
	128	cpu, raw_capacity[cpu]);
	129	capacity = (raw_capacity[cpu] << SCHED_CAPACITY_SHIFT)
	130	/ capacity_scale;
4ca4f26a	131	topology_set_cpu_scale(cpu, capacity);
2ef7a295	132	pr_debug("cpu_capacity: CPU%d cpu_capacity=%lu\n",
4ca4f26a	133	cpu, topology_get_cpu_scale(NULL, cpu));
2ef7a295 JL	134	}
	135	mutex_unlock(&cpu_scale_mutex);
	136	}
	137
805df296	138	bool __init topology_parse_cpu_capacity(struct device_node *cpu_node, int cpu)
2ef7a295	139	{
62de1161	140	static bool cap_parsing_failed;
805df296	141	int ret;
2ef7a295 JL	142	u32 cpu_capacity;
	143
	144	if (cap_parsing_failed)
805df296	145	return false;
2ef7a295	146
3eeba1a2	147	ret = of_property_read_u32(cpu_node, "capacity-dmips-mhz",
2ef7a295 JL	148	&cpu_capacity);
	149	if (!ret) {
	150	if (!raw_capacity) {
	151	raw_capacity = kcalloc(num_possible_cpus(),
	152	sizeof(*raw_capacity),
	153	GFP_KERNEL);
	154	if (!raw_capacity) {
	155	pr_err("cpu_capacity: failed to allocate memory for raw capacities\n");
	156	cap_parsing_failed = true;
805df296	157	return false;
2ef7a295 JL	158	}
	159	}
	160	capacity_scale = max(cpu_capacity, capacity_scale);
	161	raw_capacity[cpu] = cpu_capacity;
6ef2541f RH	162	pr_debug("cpu_capacity: %pOF cpu_capacity=%u (raw)\n",
6ef2541f RH	163	cpu_node, raw_capacity[cpu]);
2ef7a295 JL	164	} else {
2ef7a295 JL	165	if (raw_capacity) {
6ef2541f RH	166	pr_err("cpu_capacity: missing %pOF raw capacity\n",
6ef2541f RH	167	cpu_node);
2ef7a295 JL	168	pr_err("cpu_capacity: partial information: fallback to 1024 for all CPUs\n");
	169	}
	170	cap_parsing_failed = true;
62de1161	171	free_raw_capacity();
2ef7a295 JL	172	}
	173
	174	return !ret;
	175	}
	176
	177	#ifdef CONFIG_CPU_FREQ
7ac358ed GI	178	static cpumask_var_t cpus_to_visit;
	179	static void parsing_done_workfn(struct work_struct *work);
	180	static DECLARE_WORK(parsing_done_work, parsing_done_workfn);
2ef7a295	181
7ac358ed	182	static int
2ef7a295 JL	183	init_cpu_capacity_callback(struct notifier_block *nb,
	184	unsigned long val,
	185	void *data)
	186	{
	187	struct cpufreq_policy *policy = data;
	188	int cpu;
	189
d8bcf4db	190	if (!raw_capacity)
2ef7a295 JL	191	return 0;
2ef7a295 JL	192
93a57081 VK	193	if (val != CPUFREQ_NOTIFY)
	194	return 0;
	195
	196	pr_debug("cpu_capacity: init cpu capacity for CPUs [%pbl] (to_visit=%pbl)\n",
	197	cpumask_pr_args(policy->related_cpus),
	198	cpumask_pr_args(cpus_to_visit));
	199
	200	cpumask_andnot(cpus_to_visit, cpus_to_visit, policy->related_cpus);
	201
	202	for_each_cpu(cpu, policy->related_cpus) {
	203	raw_capacity[cpu] = topology_get_cpu_scale(NULL, cpu) *
	204	policy->cpuinfo.max_freq / 1000UL;
	205	capacity_scale = max(raw_capacity[cpu], capacity_scale);
2ef7a295	206	}
93a57081 VK	207
	208	if (cpumask_empty(cpus_to_visit)) {
	209	topology_normalize_cpu_scale();
62de1161	210	free_raw_capacity();
93a57081	211	pr_debug("cpu_capacity: parsing done\n");
93a57081 VK	212	schedule_work(&parsing_done_work);
	213	}
	214
2ef7a295 JL	215	return 0;
	216	}
	217
7ac358ed	218	static struct notifier_block init_cpu_capacity_notifier = {
2ef7a295 JL	219	.notifier_call = init_cpu_capacity_callback,
	220	};
	221
	222	static int __init register_cpufreq_notifier(void)
	223	{
5408211a DE	224	int ret;
5408211a DE	225
2ef7a295 JL	226	/*
	227	* on ACPI-based systems we need to use the default cpu capacity
	228	* until we have the necessary code to parse the cpu capacity, so
	229	* skip registering cpufreq notifier.
	230	*/
c105aa31	231	if (!acpi_disabled \|\| !raw_capacity)
2ef7a295 JL	232	return -EINVAL;
	233
	234	if (!alloc_cpumask_var(&cpus_to_visit, GFP_KERNEL)) {
	235	pr_err("cpu_capacity: failed to allocate memory for cpus_to_visit\n");
	236	return -ENOMEM;
	237	}
	238
	239	cpumask_copy(cpus_to_visit, cpu_possible_mask);
	240
5408211a DE	241	ret = cpufreq_register_notifier(&init_cpu_capacity_notifier,
	242	CPUFREQ_POLICY_NOTIFIER);
	243
	244	if (ret)
	245	free_cpumask_var(cpus_to_visit);
	246
	247	return ret;
2ef7a295 JL	248	}
	249	core_initcall(register_cpufreq_notifier);
	250
7ac358ed	251	static void parsing_done_workfn(struct work_struct *work)
2ef7a295 JL	252	{
	253	cpufreq_unregister_notifier(&init_cpu_capacity_notifier,
	254	CPUFREQ_POLICY_NOTIFIER);
5408211a	255	free_cpumask_var(cpus_to_visit);
2ef7a295 JL	256	}
	257
	258	#else
2ef7a295 JL	259	core_initcall(free_raw_capacity);
2ef7a295 JL	260	#endif