[deliverable/linux.git] / arch / x86 / events / amd / power.c

/*
 * Performance events - AMD Processor Power Reporting Mechanism
 *
 * Copyright (C) 2016 Advanced Micro Devices, Inc.
 *
 * Author: Huang Rui <ray.huang@amd.com>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

#include <linux/module.h>
#include <linux/slab.h>
#include <linux/perf_event.h>
#include <asm/cpu_device_id.h>
#include "../perf_event.h"

#define MSR_F15H_CU_PWR_ACCUMULATOR     0xc001007a
#define MSR_F15H_CU_MAX_PWR_ACCUMULATOR 0xc001007b
#define MSR_F15H_PTSC			0xc0010280

/* Event code: LSB 8 bits, passed in attr->config any other bit is reserved. */
#define AMD_POWER_EVENT_MASK		0xFFULL

/*
 * Accumulated power status counters.
 */
#define AMD_POWER_EVENTSEL_PKG		1

/*
 * The ratio of compute unit power accumulator sample period to the
 * PTSC period.
 */
static unsigned int cpu_pwr_sample_ratio;

/* Maximum accumulated power of a compute unit. */
static u64 max_cu_acc_power;

static struct pmu pmu_class;

/*
 * Accumulated power represents the sum of each compute unit's (CU) power
 * consumption. On any core of each CU we read the total accumulated power from
 * MSR_F15H_CU_PWR_ACCUMULATOR. cpu_mask represents CPU bit map of all cores
 * which are picked to measure the power for the CUs they belong to.
 */
static cpumask_t cpu_mask;

static void event_update(struct perf_event *event)
{
	struct hw_perf_event *hwc = &event->hw;
	u64 prev_pwr_acc, new_pwr_acc, prev_ptsc, new_ptsc;
	u64 delta, tdelta;

	prev_pwr_acc = hwc->pwr_acc;
	prev_ptsc = hwc->ptsc;
	rdmsrl(MSR_F15H_CU_PWR_ACCUMULATOR, new_pwr_acc);
	rdmsrl(MSR_F15H_PTSC, new_ptsc);

	/*
	 * Calculate the CU power consumption over a time period, the unit of
	 * final value (delta) is micro-Watts. Then add it to the event count.
	 */
	if (new_pwr_acc < prev_pwr_acc) {
		delta = max_cu_acc_power + new_pwr_acc;
		delta -= prev_pwr_acc;
	} else
		delta = new_pwr_acc - prev_pwr_acc;

	delta *= cpu_pwr_sample_ratio * 1000;
	tdelta = new_ptsc - prev_ptsc;

	do_div(delta, tdelta);
	local64_add(delta, &event->count);
}

static void __pmu_event_start(struct perf_event *event)
{
	if (WARN_ON_ONCE(!(event->hw.state & PERF_HES_STOPPED)))
		return;

	event->hw.state = 0;

	rdmsrl(MSR_F15H_PTSC, event->hw.ptsc);
	rdmsrl(MSR_F15H_CU_PWR_ACCUMULATOR, event->hw.pwr_acc);
}

static void pmu_event_start(struct perf_event *event, int mode)
{
	__pmu_event_start(event);
}

static void pmu_event_stop(struct perf_event *event, int mode)
{
	struct hw_perf_event *hwc = &event->hw;

	/* Mark event as deactivated and stopped. */
	if (!(hwc->state & PERF_HES_STOPPED))
		hwc->state |= PERF_HES_STOPPED;

	/* Check if software counter update is necessary. */
	if ((mode & PERF_EF_UPDATE) && !(hwc->state & PERF_HES_UPTODATE)) {
		/*
		 * Drain the remaining delta count out of an event
		 * that we are disabling:
		 */
		event_update(event);
		hwc->state |= PERF_HES_UPTODATE;
	}
}

static int pmu_event_add(struct perf_event *event, int mode)
{
	struct hw_perf_event *hwc = &event->hw;

	hwc->state = PERF_HES_UPTODATE | PERF_HES_STOPPED;

	if (mode & PERF_EF_START)
		__pmu_event_start(event);

	return 0;
}

static void pmu_event_del(struct perf_event *event, int flags)
{
	pmu_event_stop(event, PERF_EF_UPDATE);
}

static int pmu_event_init(struct perf_event *event)
{
	u64 cfg = event->attr.config & AMD_POWER_EVENT_MASK;

	/* Only look at AMD power events. */
	if (event->attr.type != pmu_class.type)
		return -ENOENT;

	/* Unsupported modes and filters. */
	if (event->attr.exclude_user   ||
	    event->attr.exclude_kernel ||
	    event->attr.exclude_hv     ||
	    event->attr.exclude_idle   ||
	    event->attr.exclude_host   ||
	    event->attr.exclude_guest  ||
	    /* no sampling */
	    event->attr.sample_period)
		return -EINVAL;

	if (cfg != AMD_POWER_EVENTSEL_PKG)
		return -EINVAL;

	return 0;
}

static void pmu_event_read(struct perf_event *event)
{
	event_update(event);
}

static ssize_t
get_attr_cpumask(struct device *dev, struct device_attribute *attr, char *buf)
{
	return cpumap_print_to_pagebuf(true, buf, &cpu_mask);
}

static DEVICE_ATTR(cpumask, S_IRUGO, get_attr_cpumask, NULL);

static struct attribute *pmu_attrs[] = {
	&dev_attr_cpumask.attr,
	NULL,
};

static struct attribute_group pmu_attr_group = {
	.attrs = pmu_attrs,
};

/*
 * Currently it only supports to report the power of each
 * processor/package.
 */
EVENT_ATTR_STR(power-pkg, power_pkg, "event=0x01");

EVENT_ATTR_STR(power-pkg.unit, power_pkg_unit, "mWatts");

/* Convert the count from micro-Watts to milli-Watts. */
EVENT_ATTR_STR(power-pkg.scale, power_pkg_scale, "1.000000e-3");

static struct attribute *events_attr[] = {
	EVENT_PTR(power_pkg),
	EVENT_PTR(power_pkg_unit),
	EVENT_PTR(power_pkg_scale),
	NULL,
};

static struct attribute_group pmu_events_group = {
	.name	= "events",
	.attrs	= events_attr,
};

PMU_FORMAT_ATTR(event, "config:0-7");

static struct attribute *formats_attr[] = {
	&format_attr_event.attr,
	NULL,
};

static struct attribute_group pmu_format_group = {
	.name	= "format",
	.attrs	= formats_attr,
};

static const struct attribute_group *attr_groups[] = {
	&pmu_attr_group,
	&pmu_format_group,
	&pmu_events_group,
	NULL,
};

static struct pmu pmu_class = {
	.attr_groups	= attr_groups,
	/* system-wide only */
	.task_ctx_nr	= perf_invalid_context,
	.event_init	= pmu_event_init,
	.add		= pmu_event_add,
	.del		= pmu_event_del,
	.start		= pmu_event_start,
	.stop		= pmu_event_stop,
	.read		= pmu_event_read,
};

static int power_cpu_exit(unsigned int cpu)
{
	int target;

	if (!cpumask_test_and_clear_cpu(cpu, &cpu_mask))
		return 0;

	/*
	 * Find a new CPU on the same compute unit, if was set in cpumask
	 * and still some CPUs on compute unit. Then migrate event and
	 * context to new CPU.
	 */
	target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
	if (target < nr_cpumask_bits) {
		cpumask_set_cpu(target, &cpu_mask);
		perf_pmu_migrate_context(&pmu_class, cpu, target);
	}
	return 0;
}

static int power_cpu_init(unsigned int cpu)
{
	int target;

	/*
	 * 1) If any CPU is set at cpu_mask in the same compute unit, do
	 * nothing.
	 * 2) If no CPU is set at cpu_mask in the same compute unit,
	 * set current ONLINE CPU.
	 *
	 * Note: if there is a CPU aside of the new one already in the
	 * sibling mask, then it is also in cpu_mask.
	 */
	target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
	if (target >= nr_cpumask_bits)
		cpumask_set_cpu(cpu, &cpu_mask);
	return 0;
}

static const struct x86_cpu_id cpu_match[] = {
	{ .vendor = X86_VENDOR_AMD, .family = 0x15 },
	{},
};

static int __init amd_power_pmu_init(void)
{
	int ret;

	if (!x86_match_cpu(cpu_match))
		return 0;

	if (!boot_cpu_has(X86_FEATURE_ACC_POWER))
		return -ENODEV;

	cpu_pwr_sample_ratio = cpuid_ecx(0x80000007);

	if (rdmsrl_safe(MSR_F15H_CU_MAX_PWR_ACCUMULATOR, &max_cu_acc_power)) {
		pr_err("Failed to read max compute unit power accumulator MSR\n");
		return -ENODEV;
	}


	cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
			  "AP_PERF_X86_AMD_POWER_ONLINE",
			  power_cpu_init, power_cpu_exit);

	ret = perf_pmu_register(&pmu_class, "power", -1);
	if (WARN_ON(ret)) {
		pr_warn("AMD Power PMU registration failed\n");
		return ret;
	}

	pr_info("AMD Power PMU detected\n");
	return ret;
}
module_init(amd_power_pmu_init);

static void __exit amd_power_pmu_exit(void)
{
	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE);
	perf_pmu_unregister(&pmu_class);
}
module_exit(amd_power_pmu_exit);

MODULE_AUTHOR("Huang Rui <ray.huang@amd.com>");
MODULE_DESCRIPTION("AMD Processor Power Reporting Mechanism");
MODULE_LICENSE("GPL v2");
Commit	Line	Data
c7ab62bf HR	1	/*
	2	* Performance events - AMD Processor Power Reporting Mechanism
	3	*
	4	* Copyright (C) 2016 Advanced Micro Devices, Inc.
	5	*
	6	* Author: Huang Rui <ray.huang@amd.com>
	7	*
	8	* This program is free software; you can redistribute it and/or modify
	9	* it under the terms of the GNU General Public License version 2 as
	10	* published by the Free Software Foundation.
	11	*/
	12
	13	#include <linux/module.h>
	14	#include <linux/slab.h>
	15	#include <linux/perf_event.h>
	16	#include <asm/cpu_device_id.h>
	17	#include "../perf_event.h"
	18
	19	#define MSR_F15H_CU_PWR_ACCUMULATOR 0xc001007a
	20	#define MSR_F15H_CU_MAX_PWR_ACCUMULATOR 0xc001007b
	21	#define MSR_F15H_PTSC 0xc0010280
	22
	23	/* Event code: LSB 8 bits, passed in attr->config any other bit is reserved. */
	24	#define AMD_POWER_EVENT_MASK 0xFFULL
	25
	26	/*
	27	* Accumulated power status counters.
	28	*/
	29	#define AMD_POWER_EVENTSEL_PKG 1
	30
	31	/*
	32	* The ratio of compute unit power accumulator sample period to the
	33	* PTSC period.
	34	*/
	35	static unsigned int cpu_pwr_sample_ratio;
	36
	37	/* Maximum accumulated power of a compute unit. */
	38	static u64 max_cu_acc_power;
	39
	40	static struct pmu pmu_class;
	41
	42	/*
	43	* Accumulated power represents the sum of each compute unit's (CU) power
	44	* consumption. On any core of each CU we read the total accumulated power from
	45	* MSR_F15H_CU_PWR_ACCUMULATOR. cpu_mask represents CPU bit map of all cores
	46	* which are picked to measure the power for the CUs they belong to.
	47	*/
	48	static cpumask_t cpu_mask;
	49
	50	static void event_update(struct perf_event *event)
	51	{
	52	struct hw_perf_event *hwc = &event->hw;
	53	u64 prev_pwr_acc, new_pwr_acc, prev_ptsc, new_ptsc;
	54	u64 delta, tdelta;
	55
	56	prev_pwr_acc = hwc->pwr_acc;
	57	prev_ptsc = hwc->ptsc;
	58	rdmsrl(MSR_F15H_CU_PWR_ACCUMULATOR, new_pwr_acc);
	59	rdmsrl(MSR_F15H_PTSC, new_ptsc);
	60
	61	/*
	62	* Calculate the CU power consumption over a time period, the unit of
	63	* final value (delta) is micro-Watts. Then add it to the event count.
	64	*/
65	if (new_pwr_acc < prev_pwr_acc) {
66	delta = max_cu_acc_power + new_pwr_acc;
67	delta -= prev_pwr_acc;
68	} else
69	delta = new_pwr_acc - prev_pwr_acc;
70
71	delta = cpu_pwr_sample_ratio 1000;
72	tdelta = new_ptsc - prev_ptsc;
73
74	do_div(delta, tdelta);
75	local64_add(delta, &event->count);
76	}
77
78	static void __pmu_event_start(struct perf_event *event)
79	{
80	if (WARN_ON_ONCE(!(event->hw.state & PERF_HES_STOPPED)))
81	return;
82
83	event->hw.state = 0;
84
85	rdmsrl(MSR_F15H_PTSC, event->hw.ptsc);
86	rdmsrl(MSR_F15H_CU_PWR_ACCUMULATOR, event->hw.pwr_acc);
87	}
88
89	static void pmu_event_start(struct perf_event *event, int mode)
90	{
91	__pmu_event_start(event);
92	}
93
94	static void pmu_event_stop(struct perf_event *event, int mode)
95	{
96	struct hw_perf_event *hwc = &event->hw;
97
98	/* Mark event as deactivated and stopped. */
99	if (!(hwc->state & PERF_HES_STOPPED))
100	hwc->state \|= PERF_HES_STOPPED;
101
102	/* Check if software counter update is necessary. */
103	if ((mode & PERF_EF_UPDATE) && !(hwc->state & PERF_HES_UPTODATE)) {
104	/*
105	* Drain the remaining delta count out of an event
106	* that we are disabling:
107	*/
108	event_update(event);
109	hwc->state \|= PERF_HES_UPTODATE;
110	}
111	}
112
113	static int pmu_event_add(struct perf_event *event, int mode)
114	{
115	struct hw_perf_event *hwc = &event->hw;
116
117	hwc->state = PERF_HES_UPTODATE \| PERF_HES_STOPPED;
118
119	if (mode & PERF_EF_START)
120	__pmu_event_start(event);
121
122	return 0;
123	}
124
125	static void pmu_event_del(struct perf_event *event, int flags)
126	{
127	pmu_event_stop(event, PERF_EF_UPDATE);
128	}
129
130	static int pmu_event_init(struct perf_event *event)
131	{
132	u64 cfg = event->attr.config & AMD_POWER_EVENT_MASK;
133
134	/* Only look at AMD power events. */
135	if (event->attr.type != pmu_class.type)
136	return -ENOENT;
137
138	/* Unsupported modes and filters. */
139	if (event->attr.exclude_user \|\|
140	event->attr.exclude_kernel \|\|
141	event->attr.exclude_hv \|\|
142	event->attr.exclude_idle \|\|
143	event->attr.exclude_host \|\|
144	event->attr.exclude_guest \|\|
145	/* no sampling */
146	event->attr.sample_period)
147	return -EINVAL;
148
149	if (cfg != AMD_POWER_EVENTSEL_PKG)
150	return -EINVAL;
151
152	return 0;
153	}
154
155	static void pmu_event_read(struct perf_event *event)
156	{
157	event_update(event);
158	}
159
160	static ssize_t
161	get_attr_cpumask(struct device dev, struct device_attribute attr, char *buf)
162	{
163	return cpumap_print_to_pagebuf(true, buf, &cpu_mask);
164	}
165
166	static DEVICE_ATTR(cpumask, S_IRUGO, get_attr_cpumask, NULL);
167
168	static struct attribute *pmu_attrs[] = {
169	&dev_attr_cpumask.attr,
170	NULL,
171	};
172
173	static struct attribute_group pmu_attr_group = {
174	.attrs = pmu_attrs,
175	};
176
177	/*
178	* Currently it only supports to report the power of each
179	* processor/package.
180	*/
181	EVENT_ATTR_STR(power-pkg, power_pkg, "event=0x01");
182
183	EVENT_ATTR_STR(power-pkg.unit, power_pkg_unit, "mWatts");
184
185	/* Convert the count from micro-Watts to milli-Watts. */
186	EVENT_ATTR_STR(power-pkg.scale, power_pkg_scale, "1.000000e-3");
187
188	static struct attribute *events_attr[] = {
189	EVENT_PTR(power_pkg),
190	EVENT_PTR(power_pkg_unit),
191	EVENT_PTR(power_pkg_scale),
192	NULL,
193	};
194
195	static struct attribute_group pmu_events_group = {
196	.name = "events",
197	.attrs = events_attr,
198	};
199
200	PMU_FORMAT_ATTR(event, "config:0-7");
201
202	static struct attribute *formats_attr[] = {
203	&format_attr_event.attr,
204	NULL,
205	};
206
207	static struct attribute_group pmu_format_group = {
208	.name = "format",
209	.attrs = formats_attr,
210	};
211
212	static const struct attribute_group *attr_groups[] = {
213	&pmu_attr_group,
214	&pmu_format_group,
215	&pmu_events_group,
216	NULL,
217	};
218
219	static struct pmu pmu_class = {
220	.attr_groups = attr_groups,
221	/* system-wide only */
222	.task_ctx_nr = perf_invalid_context,
223	.event_init = pmu_event_init,
224	.add = pmu_event_add,
225	.del = pmu_event_del,
226	.start = pmu_event_start,
227	.stop = pmu_event_stop,
228	.read = pmu_event_read,
229	};
230
c6a84daa	231	static int power_cpu_exit(unsigned int cpu)
c7ab62bf HR	232	{
	233	int target;
	234
	235	if (!cpumask_test_and_clear_cpu(cpu, &cpu_mask))
c6a84daa	236	return 0;
c7ab62bf HR	237
	238	/*
	239	* Find a new CPU on the same compute unit, if was set in cpumask
	240	* and still some CPUs on compute unit. Then migrate event and
	241	* context to new CPU.
	242	*/
	243	target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
	244	if (target < nr_cpumask_bits) {
	245	cpumask_set_cpu(target, &cpu_mask);
	246	perf_pmu_migrate_context(&pmu_class, cpu, target);
	247	}
c6a84daa	248	return 0;
c7ab62bf HR	249	}
c7ab62bf HR	250
c6a84daa	251	static int power_cpu_init(unsigned int cpu)
c7ab62bf HR	252	{
	253	int target;
	254
	255	/*
	256	* 1) If any CPU is set at cpu_mask in the same compute unit, do
	257	* nothing.
	258	* 2) If no CPU is set at cpu_mask in the same compute unit,
8381f6a0	259	* set current ONLINE CPU.
c7ab62bf HR	260	*
	261	* Note: if there is a CPU aside of the new one already in the
	262	* sibling mask, then it is also in cpu_mask.
	263	*/
	264	target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
	265	if (target >= nr_cpumask_bits)
	266	cpumask_set_cpu(cpu, &cpu_mask);
c6a84daa	267	return 0;
c7ab62bf HR	268	}
c7ab62bf HR	269
c7ab62bf HR	270	static const struct x86_cpu_id cpu_match[] = {
	271	{ .vendor = X86_VENDOR_AMD, .family = 0x15 },
	272	{},
	273	};
	274
	275	static int __init amd_power_pmu_init(void)
	276	{
c6a84daa	277	int ret;
c7ab62bf HR	278
	279	if (!x86_match_cpu(cpu_match))
	280	return 0;
	281
	282	if (!boot_cpu_has(X86_FEATURE_ACC_POWER))
	283	return -ENODEV;
	284
	285	cpu_pwr_sample_ratio = cpuid_ecx(0x80000007);
	286
	287	if (rdmsrl_safe(MSR_F15H_CU_MAX_PWR_ACCUMULATOR, &max_cu_acc_power)) {
	288	pr_err("Failed to read max compute unit power accumulator MSR\n");
	289	return -ENODEV;
	290	}
	291
c7ab62bf	292
c6a84daa AMG	293	cpuhp_setup_state(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE,
	294	"AP_PERF_X86_AMD_POWER_ONLINE",
	295	power_cpu_init, power_cpu_exit);
c7ab62bf HR	296
	297	ret = perf_pmu_register(&pmu_class, "power", -1);
	298	if (WARN_ON(ret)) {
	299	pr_warn("AMD Power PMU registration failed\n");
c6a84daa	300	return ret;
c7ab62bf HR	301	}
c7ab62bf HR	302
c7ab62bf	303	pr_info("AMD Power PMU detected\n");
c7ab62bf HR	304	return ret;
	305	}
	306	module_init(amd_power_pmu_init);
	307
	308	static void __exit amd_power_pmu_exit(void)
	309	{
c6a84daa	310	cpuhp_remove_state_nocalls(CPUHP_AP_PERF_X86_AMD_POWER_ONLINE);
c7ab62bf HR	311	perf_pmu_unregister(&pmu_class);
	312	}
	313	module_exit(amd_power_pmu_exit);
	314
	315	MODULE_AUTHOR("Huang Rui <ray.huang@amd.com>");
	316	MODULE_DESCRIPTION("AMD Processor Power Reporting Mechanism");
	317	MODULE_LICENSE("GPL v2");