diff options
Diffstat (limited to 'drivers/powercap')
| -rw-r--r-- | drivers/powercap/Kconfig | 13 | ||||
| -rw-r--r-- | drivers/powercap/Makefile | 1 | ||||
| -rw-r--r-- | drivers/powercap/intel_rapl.c | 1395 |
3 files changed, 1409 insertions, 0 deletions
diff --git a/drivers/powercap/Kconfig b/drivers/powercap/Kconfig index a37055e..a7c81b5 100644 --- a/drivers/powercap/Kconfig +++ b/drivers/powercap/Kconfig @@ -15,5 +15,18 @@ menuconfig POWERCAP if POWERCAP # Client driver configurations go here. +config INTEL_RAPL + tristate "Intel RAPL Support" + depends on X86 + default n + ---help--- + This enables support for the Intel Running Average Power Limit (RAPL) + technology which allows power limits to be enforced and monitored on + modern Intel processors (Sandy Bridge and later). + + In RAPL, the platform level settings are divided into domains for + fine grained control. These domains include processor package, DRAM + controller, CPU core (Power Plance 0), graphics uncore (Power Plane + 1), etc. endif diff --git a/drivers/powercap/Makefile b/drivers/powercap/Makefile index 6defbc8..0a21ef3 100644 --- a/drivers/powercap/Makefile +++ b/drivers/powercap/Makefile @@ -1 +1,2 @@ obj-$(CONFIG_POWERCAP) += powercap_sys.o +obj-$(CONFIG_INTEL_RAPL) += intel_rapl.o diff --git a/drivers/powercap/intel_rapl.c b/drivers/powercap/intel_rapl.c new file mode 100644 index 0000000..2a786c5 --- /dev/null +++ b/drivers/powercap/intel_rapl.c @@ -0,0 +1,1395 @@ +/* + * Intel Running Average Power Limit (RAPL) Driver + * Copyright (c) 2013, Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope 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. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc. + * + */ +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/list.h> +#include <linux/types.h> +#include <linux/device.h> +#include <linux/slab.h> +#include <linux/log2.h> +#include <linux/bitmap.h> +#include <linux/delay.h> +#include <linux/sysfs.h> +#include <linux/cpu.h> +#include <linux/powercap.h> + +#include <asm/processor.h> +#include <asm/cpu_device_id.h> + +/* bitmasks for RAPL MSRs, used by primitive access functions */ +#define ENERGY_STATUS_MASK 0xffffffff + +#define POWER_LIMIT1_MASK 0x7FFF +#define POWER_LIMIT1_ENABLE BIT(15) +#define POWER_LIMIT1_CLAMP BIT(16) + +#define POWER_LIMIT2_MASK (0x7FFFULL<<32) +#define POWER_LIMIT2_ENABLE BIT_ULL(47) +#define POWER_LIMIT2_CLAMP BIT_ULL(48) +#define POWER_PACKAGE_LOCK BIT_ULL(63) +#define POWER_PP_LOCK BIT(31) + +#define TIME_WINDOW1_MASK (0x7FULL<<17) +#define TIME_WINDOW2_MASK (0x7FULL<<49) + +#define POWER_UNIT_OFFSET 0 +#define POWER_UNIT_MASK 0x0F + +#define ENERGY_UNIT_OFFSET 0x08 +#define ENERGY_UNIT_MASK 0x1F00 + +#define TIME_UNIT_OFFSET 0x10 +#define TIME_UNIT_MASK 0xF0000 + +#define POWER_INFO_MAX_MASK (0x7fffULL<<32) +#define POWER_INFO_MIN_MASK (0x7fffULL<<16) +#define POWER_INFO_MAX_TIME_WIN_MASK (0x3fULL<<48) +#define POWER_INFO_THERMAL_SPEC_MASK 0x7fff + +#define PERF_STATUS_THROTTLE_TIME_MASK 0xffffffff +#define PP_POLICY_MASK 0x1F + +/* Non HW constants */ +#define RAPL_PRIMITIVE_DERIVED BIT(1) /* not from raw data */ +#define RAPL_PRIMITIVE_DUMMY BIT(2) + +/* scale RAPL units to avoid floating point math inside kernel */ +#define POWER_UNIT_SCALE (1000000) +#define ENERGY_UNIT_SCALE (1000000) +#define TIME_UNIT_SCALE (1000000) + +#define TIME_WINDOW_MAX_MSEC 40000 +#define TIME_WINDOW_MIN_MSEC 250 + +enum unit_type { + ARBITRARY_UNIT, /* no translation */ + POWER_UNIT, + ENERGY_UNIT, + TIME_UNIT, +}; + +enum rapl_domain_type { + RAPL_DOMAIN_PACKAGE, /* entire package/socket */ + RAPL_DOMAIN_PP0, /* core power plane */ + RAPL_DOMAIN_PP1, /* graphics uncore */ + RAPL_DOMAIN_DRAM,/* DRAM control_type */ + RAPL_DOMAIN_MAX, +}; + +enum rapl_domain_msr_id { + RAPL_DOMAIN_MSR_LIMIT, + RAPL_DOMAIN_MSR_STATUS, + RAPL_DOMAIN_MSR_PERF, + RAPL_DOMAIN_MSR_POLICY, + RAPL_DOMAIN_MSR_INFO, + RAPL_DOMAIN_MSR_MAX, +}; + +/* per domain data, some are optional */ +enum rapl_primitives { + ENERGY_COUNTER, + POWER_LIMIT1, + POWER_LIMIT2, + FW_LOCK, + + PL1_ENABLE, /* power limit 1, aka long term */ + PL1_CLAMP, /* allow frequency to go below OS request */ + PL2_ENABLE, /* power limit 2, aka short term, instantaneous */ + PL2_CLAMP, + + TIME_WINDOW1, /* long term */ + TIME_WINDOW2, /* short term */ + THERMAL_SPEC_POWER, + MAX_POWER, + + MIN_POWER, + MAX_TIME_WINDOW, + THROTTLED_TIME, + PRIORITY_LEVEL, + + /* below are not raw primitive data */ + AVERAGE_POWER, + NR_RAPL_PRIMITIVES, +}; + +#define NR_RAW_PRIMITIVES (NR_RAPL_PRIMITIVES - 2) + +/* Can be expanded to include events, etc.*/ +struct rapl_domain_data { + u64 primitives[NR_RAPL_PRIMITIVES]; + unsigned long timestamp; +}; + + +#define DOMAIN_STATE_INACTIVE BIT(0) +#define DOMAIN_STATE_POWER_LIMIT_SET BIT(1) +#define DOMAIN_STATE_BIOS_LOCKED BIT(2) + +#define NR_POWER_LIMITS (2) +struct rapl_power_limit { + struct powercap_zone_constraint *constraint; + int prim_id; /* primitive ID used to enable */ + struct rapl_domain *domain; + const char *name; +}; + +static const char pl1_name[] = "long_term"; +static const char pl2_name[] = "short_term"; + +struct rapl_domain { + const char *name; + enum rapl_domain_type id; + int msrs[RAPL_DOMAIN_MSR_MAX]; + struct powercap_zone power_zone; + struct rapl_domain_data rdd; + struct rapl_power_limit rpl[NR_POWER_LIMITS]; + u64 attr_map; /* track capabilities */ + unsigned int state; + int package_id; +}; +#define power_zone_to_rapl_domain(_zone) \ + container_of(_zone, struct rapl_domain, power_zone) + + +/* Each physical package contains multiple domains, these are the common + * data across RAPL domains within a package. + */ +struct rapl_package { + unsigned int id; /* physical package/socket id */ + unsigned int nr_domains; + unsigned long domain_map; /* bit map of active domains */ + unsigned int power_unit_divisor; + unsigned int energy_unit_divisor; + unsigned int time_unit_divisor; + struct rapl_domain *domains; /* array of domains, sized at runtime */ + struct powercap_zone *power_zone; /* keep track of parent zone */ + int nr_cpus; /* active cpus on the package, topology info is lost during + * cpu hotplug. so we have to track ourselves. + */ + unsigned long power_limit_irq; /* keep track of package power limit + * notify interrupt enable status. + */ + struct list_head plist; +}; +#define PACKAGE_PLN_INT_SAVED BIT(0) +#define MAX_PRIM_NAME (32) + +/* per domain data. used to describe individual knobs such that access function + * can be consolidated into one instead of many inline functions. + */ +struct rapl_primitive_info { + const char *name; + u64 mask; + int shift; + enum rapl_domain_msr_id id; + enum unit_type unit; + u32 flag; +}; + +#define PRIMITIVE_INFO_INIT(p, m, s, i, u, f) { \ + .name = #p, \ + .mask = m, \ + .shift = s, \ + .id = i, \ + .unit = u, \ + .flag = f \ + } + +static void rapl_init_domains(struct rapl_package *rp); +static int rapl_read_data_raw(struct rapl_domain *rd, + enum rapl_primitives prim, + bool xlate, u64 *data); +static int rapl_write_data_raw(struct rapl_domain *rd, + enum rapl_primitives prim, + unsigned long long value); +static u64 rapl_unit_xlate(int package, enum unit_type type, u64 value, + int to_raw); +static void package_power_limit_irq_save(int package_id); + +static LIST_HEAD(rapl_packages); /* guarded by CPU hotplug lock */ + +static const char * const rapl_domain_names[] = { + "package", + "core", + "uncore", + "dram", +}; + +static struct powercap_control_type *control_type; /* PowerCap Controller */ + +/* caller to ensure CPU hotplug lock is held */ +static struct rapl_package *find_package_by_id(int id) +{ + struct rapl_package *rp; + + list_for_each_entry(rp, &rapl_packages, plist) { + if (rp->id == id) + return rp; + } + + return NULL; +} + +/* caller to ensure CPU hotplug lock is held */ +static int find_active_cpu_on_package(int package_id) +{ + int i; + + for_each_online_cpu(i) { + if (topology_physical_package_id(i) == package_id) + return i; + } + /* all CPUs on this package are offline */ + + return -ENODEV; +} + +/* caller must hold cpu hotplug lock */ +static void rapl_cleanup_data(void) +{ + struct rapl_package *p, *tmp; + + list_for_each_entry_safe(p, tmp, &rapl_packages, plist) { + kfree(p->domains); + list_del(&p->plist); + kfree(p); + } +} + +static int get_energy_counter(struct powercap_zone *power_zone, u64 *energy_raw) +{ + struct rapl_domain *rd; + u64 energy_now; + + /* prevent CPU hotplug, make sure the RAPL domain does not go + * away while reading the counter. + */ + get_online_cpus(); + rd = power_zone_to_rapl_domain(power_zone); + + if (!rapl_read_data_raw(rd, ENERGY_COUNTER, true, &energy_now)) { + *energy_raw = energy_now; + put_online_cpus(); + + return 0; + } + put_online_cpus(); + + return -EIO; +} + +static int get_max_energy_counter(struct powercap_zone *pcd_dev, u64 *energy) +{ + *energy = rapl_unit_xlate(0, ENERGY_UNIT, ENERGY_STATUS_MASK, 0); + return 0; +} + +static int release_zone(struct powercap_zone *power_zone) +{ + struct rapl_domain *rd = power_zone_to_rapl_domain(power_zone); + struct rapl_package *rp; + + /* package zone is the last zone of a package, we can free + * memory here since all children has been unregistered. + */ + if (rd->id == RAPL_DOMAIN_PACKAGE) { + rp = find_package_by_id(rd->package_id); + if (!rp) { + dev_warn(&power_zone->dev, "no package id %s\n", + rd->name); + return -ENODEV; + } + kfree(rd); + rp->domains = NULL; + } + + return 0; + +} + +static int find_nr_power_limit(struct rapl_domain *rd) +{ + int i; + + for (i = 0; i < NR_POWER_LIMITS; i++) { + if (rd->rpl[i].name == NULL) + break; + } + + return i; +} + +static int set_domain_enable(struct powercap_zone *power_zone, bool mode) +{ + struct rapl_domain *rd = power_zone_to_rapl_domain(power_zone); + int nr_powerlimit; + + if (rd->state & DOMAIN_STATE_BIOS_LOCKED) + return -EACCES; + get_online_cpus(); + nr_powerlimit = find_nr_power_limit(rd); + /* here we activate/deactivate the hardware for power limiting */ + rapl_write_data_raw(rd, PL1_ENABLE, mode); + /* always enable clamp such that p-state can go below OS requested + * range. power capping priority over guranteed frequency. + */ + rapl_write_data_raw(rd, PL1_CLAMP, mode); + /* some domains have pl2 */ + if (nr_powerlimit > 1) { + rapl_write_data_raw(rd, PL2_ENABLE, mode); + rapl_write_data_raw(rd, PL2_CLAMP, mode); + } + put_online_cpus(); + + return 0; +} + +static int get_domain_enable(struct powercap_zone *power_zone, bool *mode) +{ + struct rapl_domain *rd = power_zone_to_rapl_domain(power_zone); + u64 val; + + if (rd->state & DOMAIN_STATE_BIOS_LOCKED) { + *mode = false; + return 0; + } + get_online_cpus(); + if (rapl_read_data_raw(rd, PL1_ENABLE, true, &val)) { + put_online_cpus(); + return -EIO; + } + *mode = val; + put_online_cpus(); + + return 0; +} + +/* per RAPL domain ops, in the order of rapl_domain_type */ +static struct powercap_zone_ops zone_ops[] = { + /* RAPL_DOMAIN_PACKAGE */ + { + .get_energy_uj = get_energy_counter, + .get_max_energy_range_uj = get_max_energy_counter, + .release = release_zone, + .set_enable = set_domain_enable, + .get_enable = get_domain_enable, + }, + /* RAPL_DOMAIN_PP0 */ + { + .get_energy_uj = get_energy_counter, + .get_max_energy_range_uj = get_max_energy_counter, + .release = release_zone, + .set_enable = set_domain_enable, + .get_enable = get_domain_enable, + }, + /* RAPL_DOMAIN_PP1 */ + { + .get_energy_uj = get_energy_counter, + .get_max_energy_range_uj = get_max_energy_counter, + .release = release_zone, + .set_enable = set_domain_enable, + .get_enable = get_domain_enable, + }, + /* RAPL_DOMAIN_DRAM */ + { + .get_energy_uj = get_energy_counter, + .get_max_energy_range_uj = get_max_energy_counter, + .release = release_zone, + .set_enable = set_domain_enable, + .get_enable = get_domain_enable, + }, +}; + +static int set_power_limit(struct powercap_zone *power_zone, int id, + u64 power_limit) +{ + struct rapl_domain *rd; + struct rapl_package *rp; + int ret = 0; + + get_online_cpus(); + rd = power_zone_to_rapl_domain(power_zone); + rp = find_package_by_id(rd->package_id); + if (!rp) { + ret = -ENODEV; + goto set_exit; + } + + if (rd->state & DOMAIN_STATE_BIOS_LOCKED) { + dev_warn(&power_zone->dev, "%s locked by BIOS, monitoring only\n", + rd->name); + ret = -EACCES; + goto set_exit; + } + + switch (rd->rpl[id].prim_id) { + case PL1_ENABLE: + rapl_write_data_raw(rd, POWER_LIMIT1, power_limit); + break; + case PL2_ENABLE: + rapl_write_data_raw(rd, POWER_LIMIT2, power_limit); + break; + default: + ret = -EINVAL; + } + if (!ret) + package_power_limit_irq_save(rd->package_id); +set_exit: + put_online_cpus(); + return ret; +} + +static int get_current_power_limit(struct powercap_zone *power_zone, int id, + u64 *data) +{ + struct rapl_domain *rd; + u64 val; + int prim; + int ret = 0; + + get_online_cpus(); + rd = power_zone_to_rapl_domain(power_zone); + switch (rd->rpl[id].prim_id) { + case PL1_ENABLE: + prim = POWER_LIMIT1; + break; + case PL2_ENABLE: + prim = POWER_LIMIT2; + break; + default: + put_online_cpus(); + return -EINVAL; + } + if (rapl_read_data_raw(rd, prim, true, &val)) + ret = -EIO; + else + *data = val; + + put_online_cpus(); + + return ret; +} + +static int set_time_window(struct powercap_zone *power_zone, int id, + u64 window) +{ + struct rapl_domain *rd; + int ret = 0; + + get_online_cpus(); + rd = power_zone_to_rapl_domain(power_zone); + switch (rd->rpl[id].prim_id) { + case PL1_ENABLE: + rapl_write_data_raw(rd, TIME_WINDOW1, window); + break; + case PL2_ENABLE: + rapl_write_data_raw(rd, TIME_WINDOW2, window); + break; + default: + ret = -EINVAL; + } + put_online_cpus(); + return ret; +} + +static int get_time_window(struct powercap_zone *power_zone, int id, u64 *data) +{ + struct rapl_domain *rd; + u64 val; + int ret = 0; + + get_online_cpus(); + rd = power_zone_to_rapl_domain(power_zone); + switch (rd->rpl[id].prim_id) { + case PL1_ENABLE: + ret = rapl_read_data_raw(rd, TIME_WINDOW1, true, &val); + break; + case PL2_ENABLE: + ret = rapl_read_data_raw(rd, TIME_WINDOW2, true, &val); + break; + default: + put_online_cpus(); + return -EINVAL; + } + if (!ret) + *data = val; + put_online_cpus(); + + return ret; +} + +static const char *get_constraint_name(struct powercap_zone *power_zone, int id) +{ + struct rapl_power_limit *rpl; + struct rapl_domain *rd; + + rd = power_zone_to_rapl_domain(power_zone); + rpl = (struct rapl_power_limit *) &rd->rpl[id]; + + return rpl->name; +} + + +static int get_max_power(struct powercap_zone *power_zone, int id, + u64 *data) +{ + struct rapl_domain *rd; + u64 val; + int prim; + int ret = 0; + + get_online_cpus(); + rd = power_zone_to_rapl_domain(power_zone); + switch (rd->rpl[id].prim_id) { + case PL1_ENABLE: + prim = THERMAL_SPEC_POWER; + break; + case PL2_ENABLE: + prim = MAX_POWER; + break; + default: + put_online_cpus(); + return -EINVAL; + } + if (rapl_read_data_raw(rd, prim, true, &val)) + ret = -EIO; + else + *data = val; + + put_online_cpus(); + + return ret; +} + +static struct powercap_zone_constraint_ops constraint_ops = { + .set_power_limit_uw = set_power_limit, + .get_power_limit_uw = get_current_power_limit, + .set_time_window_us = set_time_window, + .get_time_window_us = get_time_window, + .get_max_power_uw = get_max_power, + .get_name = get_constraint_name, +}; + +/* called after domain detection and package level data are set */ +static void rapl_init_domains(struct rapl_package *rp) +{ + int i; + struct rapl_domain *rd = rp->domains; + + for (i = 0; i < RAPL_DOMAIN_MAX; i++) { + unsigned int mask = rp->domain_map & (1 << i); + switch (mask) { + case BIT(RAPL_DOMAIN_PACKAGE): + rd->name = rapl_domain_names[RAPL_DOMAIN_PACKAGE]; + rd->id = RAPL_DOMAIN_PACKAGE; + rd->msrs[0] = MSR_PKG_POWER_LIMIT; + rd->msrs[1] = MSR_PKG_ENERGY_STATUS; + rd->msrs[2] = MSR_PKG_PERF_STATUS; + rd->msrs[3] = 0; + rd->msrs[4] = MSR_PKG_POWER_INFO; + rd->rpl[0].prim_id = PL1_ENABLE; + rd->rpl[0].name = pl1_name; + rd->rpl[1].prim_id = PL2_ENABLE; + rd->rpl[1].name = pl2_name; + break; + case BIT(RAPL_DOMAIN_PP0): + rd->name = rapl_domain_names[RAPL_DOMAIN_PP0]; + rd->id = RAPL_DOMAIN_PP0; + rd->msrs[0] = MSR_PP0_POWER_LIMIT; + rd->msrs[1] = MSR_PP0_ENERGY_STATUS; + rd->msrs[2] = 0; + rd->msrs[3] = MSR_PP0_POLICY; + rd->msrs[4] = 0; + rd->rpl[0].prim_id = PL1_ENABLE; + rd->rpl[0].name = pl1_name; + break; + case BIT(RAPL_DOMAIN_PP1): + rd->name = rapl_domain_names[RAPL_DOMAIN_PP1]; + rd->id = RAPL_DOMAIN_PP1; + rd->msrs[0] = MSR_PP1_POWER_LIMIT; + rd->msrs[1] = MSR_PP1_ENERGY_STATUS; + rd->msrs[2] = 0; + rd->msrs[3] = MSR_PP1_POLICY; + rd->msrs[4] = 0; + rd->rpl[0].prim_id = PL1_ENABLE; + rd->rpl[0].name = pl1_name; + break; + case BIT(RAPL_DOMAIN_DRAM): + rd->name = rapl_domain_names[RAPL_DOMAIN_DRAM]; + rd->id = RAPL_DOMAIN_DRAM; + rd->msrs[0] = MSR_DRAM_POWER_LIMIT; + rd->msrs[1] = MSR_DRAM_ENERGY_STATUS; + rd->msrs[2] = MSR_DRAM_PERF_STATUS; + rd->msrs[3] = 0; + rd->msrs[4] = MSR_DRAM_POWER_INFO; + rd->rpl[0].prim_id = PL1_ENABLE; + rd->rpl[0].name = pl1_name; + break; + } + if (mask) { + rd->package_id = rp->id; + rd++; + } + } +} + +static u64 rapl_unit_xlate(int package, enum unit_type type, u64 value, + int to_raw) +{ + u64 divisor = 1; + int scale = 1; /* scale to user friendly data without floating point */ + u64 f, y; /* fraction and exp. used for time unit */ + struct rapl_package *rp; + + rp = find_package_by_id(package); + if (!rp) + return value; + + switch (type) { + case POWER_UNIT: + divisor = rp->power_unit_divisor; + scale = POWER_UNIT_SCALE; + break; + case ENERGY_UNIT: + scale = ENERGY_UNIT_SCALE; + divisor = rp->energy_unit_divisor; + break; + case TIME_UNIT: + divisor = rp->time_unit_divisor; + scale = TIME_UNIT_SCALE; + /* special processing based on 2^Y*(1+F)/4 = val/divisor, refer + * to Intel Software Developer's manual Vol. 3a, CH 14.7.4. + */ + if (!to_raw) { + f = (value & 0x60) >> 5; + y = value & 0x1f; + value = (1 << y) * (4 + f) * scale / 4; + return div64_u64(value, divisor); + } else { + do_div(value, scale); + value *= divisor; + y = ilog2(value); + f = div64_u64(4 * (value - (1 << y)), 1 << y); + value = (y & 0x1f) | ((f & 0x3) << 5); + return value; + } + break; + case ARBITRARY_UNIT: + default: + return value; + }; + + if (to_raw) + return div64_u64(value * divisor, scale); + else + return div64_u64(value * scale, divisor); +} + +/* in the order of enum rapl_primitives */ +static struct rapl_primitive_info rpi[] = { + /* name, mask, shift, msr index, unit divisor */ + PRIMITIVE_INFO_INIT(ENERGY_COUNTER, ENERGY_STATUS_MASK, 0, + RAPL_DOMAIN_MSR_STATUS, ENERGY_UNIT, 0), + PRIMITIVE_INFO_INIT(POWER_LIMIT1, POWER_LIMIT1_MASK, 0, + RAPL_DOMAIN_MSR_LIMIT, POWER_UNIT, 0), + PRIMITIVE_INFO_INIT(POWER_LIMIT2, POWER_LIMIT2_MASK, 32, + RAPL_DOMAIN_MSR_LIMIT, POWER_UNIT, 0), + PRIMITIVE_INFO_INIT(FW_LOCK, POWER_PP_LOCK, 31, + RAPL_DOMAIN_MSR_LIMIT, ARBITRARY_UNIT, 0), + PRIMITIVE_INFO_INIT(PL1_ENABLE, POWER_LIMIT1_ENABLE, 15, + RAPL_DOMAIN_MSR_LIMIT, ARBITRARY_UNIT, 0), + PRIMITIVE_INFO_INIT(PL1_CLAMP, POWER_LIMIT1_CLAMP, 16, + RAPL_DOMAIN_MSR_LIMIT, ARBITRARY_UNIT, 0), + PRIMITIVE_INFO_INIT(PL2_ENABLE, POWER_LIMIT2_ENABLE, 47, + RAPL_DOMAIN_MSR_LIMIT, ARBITRARY_UNIT, 0), + PRIMITIVE_INFO_INIT(PL2_CLAMP, POWER_LIMIT2_CLAMP, 48, + RAPL_DOMAIN_MSR_LIMIT, ARBITRARY_UNIT, 0), + PRIMITIVE_INFO_INIT(TIME_WINDOW1, TIME_WINDOW1_MASK, 17, + RAPL_DOMAIN_MSR_LIMIT, TIME_UNIT, 0), + PRIMITIVE_INFO_INIT(TIME_WINDOW2, TIME_WINDOW2_MASK, 49, + RAPL_DOMAIN_MSR_LIMIT, TIME_UNIT, 0), + PRIMITIVE_INFO_INIT(THERMAL_SPEC_POWER, POWER_INFO_THERMAL_SPEC_MASK, + 0, RAPL_DOMAIN_MSR_INFO, POWER_UNIT, 0), + PRIMITIVE_INFO_INIT(MAX_POWER, POWER_INFO_MAX_MASK, 32, + RAPL_DOMAIN_MSR_INFO, POWER_UNIT, 0), + PRIMITIVE_INFO_INIT(MIN_POWER, POWER_INFO_MIN_MASK, 16, + RAPL_DOMAIN_MSR_INFO, POWER_UNIT, 0), + PRIMITIVE_INFO_INIT(MAX_TIME_WINDOW, POWER_INFO_MAX_TIME_WIN_MASK, 48, + RAPL_DOMAIN_MSR_INFO, TIME_UNIT, 0), + PRIMITIVE_INFO_INIT(THROTTLED_TIME, PERF_STATUS_THROTTLE_TIME_MASK, 0, + RAPL_DOMAIN_MSR_PERF, TIME_UNIT, 0), + PRIMITIVE_INFO_INIT(PRIORITY_LEVEL, PP_POLICY_MASK, 0, + RAPL_DOMAIN_MSR_POLICY, ARBITRARY_UNIT, 0), + /* non-hardware */ + PRIMITIVE_INFO_INIT(AVERAGE_POWER, 0, 0, 0, POWER_UNIT, + RAPL_PRIMITIVE_DERIVED), + {NULL, 0, 0, 0}, +}; + +/* Read primitive data based on its related struct rapl_primitive_info. + * if xlate flag is set, return translated data based on data units, i.e. + * time, energy, and power. + * RAPL MSRs are non-architectual and are laid out not consistently across + * domains. Here we use primitive info to allow writing consolidated access + * functions. + * For a given primitive, it is processed by MSR mask and shift. Unit conversion + * is pre-assigned based on RAPL unit MSRs read at init time. + * 63-------------------------- 31--------------------------- 0 + * | xxxxx (mask) | + * | |<- shift ----------------| + * 63-------------------------- 31--------------------------- 0 + */ +static int rapl_read_data_raw(struct rapl_domain *rd, + enum rapl_primitives prim, + bool xlate, u64 *data) +{ + u64 value, final; + u32 msr; + struct rapl_primitive_info *rp = &rpi[prim]; + int cpu; + + if (!rp->name || rp->flag & RAPL_PRIMITIVE_DUMMY) + return -EINVAL; + + msr = rd->msrs[rp->id]; + if (!msr) + return -EINVAL; + /* use physical package id to look up active cpus */ + cpu = find_active_cpu_on_package(rd->package_id); + if (cpu < 0) + return cpu; + + /* special-case package domain, which uses a different bit*/ + if (prim == FW_LOCK && rd->id == RAPL_DOMAIN_PACKAGE) { + rp->mask = POWER_PACKAGE_LOCK; + rp->shift = 63; + } + /* non-hardware data are collected by the polling thread */ + if (rp->flag & RAPL_PRIMITIVE_DERIVED) { + *data = rd->rdd.primitives[prim]; + return 0; + } + + if (rdmsrl_safe_on_cpu(cpu, msr, &value)) { + pr_debug("failed to read msr 0x%x on cpu %d\n", msr, cpu); + return -EIO; + } + + final = value & rp->mask; + final = final >> rp->shift; + if (xlate) + *data = rapl_unit_xlate(rd->package_id, rp->unit, final, 0); + else + *data = final; + + return 0; +} + +/* Similar use of primitive info in the read counterpart */ +static int rapl_write_data_raw(struct rapl_domain *rd, + enum rapl_primitives prim, + unsigned long long value) +{ + u64 msr_val; + u32 msr; + struct rapl_primitive_info *rp = &rpi[prim]; + int cpu; + + cpu = find_active_cpu_on_package(rd->package_id); + if (cpu < 0) + return cpu; + msr = rd->msrs[rp->id]; + if (rdmsrl_safe_on_cpu(cpu, msr, &msr_val)) { + dev_dbg(&rd->power_zone.dev, + "failed to read msr 0x%x on cpu %d\n", msr, cpu); + return -EIO; + } + value = rapl_unit_xlate(rd->package_id, rp->unit, value, 1); + msr_val &= ~rp->mask; + msr_val |= value << rp->shift; + if (wrmsrl_safe_on_cpu(cpu, msr, msr_val)) { + dev_dbg(&rd->power_zone.dev, + "failed to write msr 0x%x on cpu %d\n", msr, cpu); + return -EIO; + } + + return 0; +} + +static int rapl_check_unit(struct rapl_package *rp, int cpu) +{ + u64 msr_val; + u32 value; + + if (rdmsrl_safe_on_cpu(cpu, MSR_RAPL_POWER_UNIT, &msr_val)) { + pr_err("Failed to read power unit MSR 0x%x on CPU %d, exit.\n", + MSR_RAPL_POWER_UNIT, cpu); + return -ENODEV; + } + + /* Raw RAPL data stored in MSRs are in certain scales. We need to + * convert them into standard units based on the divisors reported in + * the RAPL unit MSRs. + * i.e. + * energy unit: 1/enery_unit_divisor Joules + * power unit: 1/power_unit_divisor Watts + * time unit: 1/time_unit_divisor Seconds + */ + value = (msr_val & ENERGY_UNIT_MASK) >> ENERGY_UNIT_OFFSET; + rp->energy_unit_divisor = 1 << value; + + + value = (msr_val & POWER_UNIT_MASK) >> POWER_UNIT_OFFSET; + rp->power_unit_divisor = 1 << value; + + value = (msr_val & TIME_UNIT_MASK) >> TIME_UNIT_OFFSET; + rp->time_unit_divisor = 1 << value; + + pr_debug("Physical package %d units: energy=%d, time=%d, power=%d\n", + rp->id, + rp->energy_unit_divisor, + rp->time_unit_divisor, + rp->power_unit_divisor); + + return 0; +} + +/* REVISIT: + * When package power limit is set artificially low by RAPL, LVT + * thermal interrupt for package power limit should be ignored + * since we are not really exceeding the real limit. The intention + * is to avoid excessive interrupts while we are trying to save power. + * A useful feature might be routing the package_power_limit interrupt + * to userspace via eventfd. once we have a usecase, this is simple + * to do by adding an atomic notifier. + */ + +static void package_power_limit_irq_save(int package_id) +{ + u32 l, h = 0; + int cpu; + struct rapl_package *rp; + + rp = find_package_by_id(package_id); + if (!rp) + return; + + if (!boot_cpu_has(X86_FEATURE_PTS) || !boot_cpu_has(X86_FEATURE_PLN)) + return; + + cpu = find_active_cpu_on_package(package_id); + if (cpu < 0) + return; + /* save the state of PLN irq mask bit before disabling it */ + rdmsr_safe_on_cpu(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, &l, &h); + if (!(rp->power_limit_irq & PACKAGE_PLN_INT_SAVED)) { + rp->power_limit_irq = l & PACKAGE_THERM_INT_PLN_ENABLE; + rp->power_limit_irq |= PACKAGE_PLN_INT_SAVED; + } + l &= ~PACKAGE_THERM_INT_PLN_ENABLE; + wrmsr_on_cpu(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h); +} + +/* restore per package power limit interrupt enable state */ +static void package_power_limit_irq_restore(int package_id) +{ + u32 l, h; + int cpu; + struct rapl_package *rp; + + rp = find_package_by_id(package_id); + if (!rp) + return; + + if (!boot_cpu_has(X86_FEATURE_PTS) || !boot_cpu_has(X86_FEATURE_PLN)) + return; + + cpu = find_active_cpu_on_package(package_id); + if (cpu < 0) + return; + + /* irq enable state not saved, nothing to restore */ + if (!(rp->power_limit_irq & PACKAGE_PLN_INT_SAVED)) + return; + rdmsr_safe_on_cpu(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, &l, &h); + + if (rp->power_limit_irq & PACKAGE_THERM_INT_PLN_ENABLE) + l |= PACKAGE_THERM_INT_PLN_ENABLE; + else + l &= ~PACKAGE_THERM_INT_PLN_ENABLE; + + wrmsr_on_cpu(cpu, MSR_IA32_PACKAGE_THERM_INTERRUPT, l, h); +} + +static const struct x86_cpu_id rapl_ids[] = { + { X86_VENDOR_INTEL, 6, 0x2a},/* SNB */ + { X86_VENDOR_INTEL, 6, 0x2d},/* SNB EP */ + { X86_VENDOR_INTEL, 6, 0x3a},/* IVB */ + { X86_VENDOR_INTEL, 6, 0x45},/* HSW */ + /* TODO: Add more CPU IDs after testing */ + {} +}; +MODULE_DEVICE_TABLE(x86cpu, rapl_ids); + +/* read once for all raw primitive data for all packages, domains */ +static void rapl_update_domain_data(void) +{ + int dmn, prim; + u64 val; + struct rapl_package *rp; + + list_for_each_entry(rp, &rapl_packages, plist) { + for (dmn = 0; dmn < rp->nr_domains; dmn++) { + pr_debug("update package %d domain %s data\n", rp->id, + rp->domains[dmn].name); + /* exclude non-raw primitives */ + for (prim = 0; prim < NR_RAW_PRIMITIVES; prim++) + if (!rapl_read_data_raw(&rp->domains[dmn], prim, + rpi[prim].unit, + &val)) + rp->domains[dmn].rdd.primitives[prim] = + val; + } + } + +} + +static int rapl_unregister_powercap(void) +{ + struct rapl_package *rp; + struct rapl_domain *rd, *rd_package = NULL; + + /* unregister all active rapl packages from the powercap layer, + * hotplug lock held + */ + list_for_each_entry(rp, &rapl_packages, plist) { + package_power_limit_irq_restore(rp->id); + + for (rd = rp->domains; rd < rp->domains + rp->nr_domains; + rd++) { + pr_debug("remove package, undo power limit on %d: %s\n", + rp->id, rd->name); + rapl_write_data_raw(rd, PL1_ENABLE, 0); + rapl_write_data_raw(rd, PL2_ENABLE, 0); + rapl_write_data_raw(rd, PL1_CLAMP, 0); + rapl_write_data_raw(rd, PL2_CLAMP, 0); + if (rd->id == RAPL_DOMAIN_PACKAGE) { + rd_package = rd; + continue; + } + powercap_unregister_zone(control_type, &rd->power_zone); + } + /* do the package zone last */ + if (rd_package) + powercap_unregister_zone(control_type, + &rd_package->power_zone); + } + powercap_unregister_control_type(control_type); + + return 0; +} + +static int rapl_package_register_powercap(struct rapl_package *rp) +{ + struct rapl_domain *rd; + int ret = 0; + char dev_name[17]; /* max domain name = 7 + 1 + 8 for int + 1 for null*/ + struct powercap_zone *power_zone = NULL; + int nr_pl; + + /* first we register package domain as the parent zone*/ + for (rd = rp->domains; rd < rp->domains + rp->nr_domains; rd++) { + if (rd->id == RAPL_DOMAIN_PACKAGE) { + nr_pl = find_nr_power_limit(rd); + pr_debug("register socket %d package domain %s\n", + rp->id, rd->name); + memset(dev_name, 0, sizeof(dev_name)); + snprintf(dev_name, sizeof(dev_name), "%s-%d", + rd->name, rp->id); + power_zone = powercap_register_zone(&rd->power_zone, + control_type, + dev_name, NULL, + &zone_ops[rd->id], + nr_pl, + &constraint_ops); + if (IS_ERR(power_zone)) { + pr_debug("failed to register package, %d\n", + rp->id); + ret = PTR_ERR(power_zone); + goto exit_package; + } + /* track parent zone in per package/socket data */ + rp->power_zone = power_zone; + /* done, only one package domain per socket */ + break; + } + } + if (!power_zone) { + pr_err("no package domain found, unknown topology!\n"); + ret = -ENODEV; + goto exit_package; + } + /* now register domains as children of the socket/package*/ + for (rd = rp->domains; rd < rp->domains + rp->nr_domains; rd++) { + if (rd->id == RAPL_DOMAIN_PACKAGE) + continue; + /* number of power limits per domain varies */ + nr_pl = find_nr_power_limit(rd); + power_zone = powercap_register_zone(&rd->power_zone, + control_type, rd->name, + rp->power_zone, + &zone_ops[rd->id], nr_pl, + &constraint_ops); + + if (IS_ERR(power_zone)) { + pr_debug("failed to register power_zone, %d:%s:%s\n", + rp->id, rd->name, dev_name); + ret = PTR_ERR(power_zone); + goto err_cleanup; + } + } + +exit_package: + return ret; +err_cleanup: + /* clean up previously initialized domains within the package if we + * failed after the first domain setup. + */ + while (--rd >= rp->domains) { + pr_debug("unregister package %d domain %s\n", rp->id, rd->name); + powercap_unregister_zone(control_type, &rd->power_zone); + } + + return ret; +} + +static int rapl_register_powercap(void) +{ + struct rapl_domain *rd; + struct rapl_package *rp; + int ret = 0; + + control_type = powercap_register_control_type(NULL, "intel-rapl", NULL); + if (IS_ERR(control_type)) { + pr_debug("failed to register powercap control_type.\n"); + return PTR_ERR(control_type); + } + /* read the initial data */ + rapl_update_domain_data(); + list_for_each_entry(rp, &rapl_packages, plist) + if (rapl_package_register_powercap(rp)) + goto err_cleanup_package; + return ret; + +err_cleanup_package: + /* clean up previously initialized packages */ + list_for_each_entry_continue_reverse(rp, &rapl_packages, plist) { + for (rd = rp->domains; rd < rp->domains + rp->nr_domains; + rd++) { + pr_debug("unregister zone/package %d, %s domain\n", + rp->id, rd->name); + powercap_unregister_zone(control_type, &rd->power_zone); + } + } + + return ret; +} + +static int rapl_check_domain(int cpu, int domain) +{ + unsigned msr; + u64 val1, val2 = 0; + int retry = 0; + + switch (domain) { + case RAPL_DOMAIN_PACKAGE: + msr = MSR_PKG_ENERGY_STATUS; + break; + case RAPL_DOMAIN_PP0: + msr = MSR_PP0_ENERGY_STATUS; + break; + case RAPL_DOMAIN_PP1: + msr = MSR_PP1_ENERGY_STATUS; + break; + case RAPL_DOMAIN_DRAM: + msr = MSR_DRAM_ENERGY_STATUS; + break; + default: + pr_err("invalid domain id %d\n", domain); + return -EINVAL; + } + if (rdmsrl_safe_on_cpu(cpu, msr, &val1)) + return -ENODEV; + + /* energy counters roll slowly on some domains */ + while (++retry < 10) { + usleep_range(10000, 15000); + rdmsrl_safe_on_cpu(cpu, msr, &val2); + if ((val1 & ENERGY_STATUS_MASK) != (val2 & ENERGY_STATUS_MASK)) + return 0; + } + /* if energy counter does not change, report as bad domain */ + pr_info("domain %s energy ctr %llu:%llu not working, skip\n", + rapl_domain_names[domain], val1, val2); + + return -ENODEV; +} + +/* Detect active and valid domains for the given CPU, caller must + * ensure the CPU belongs to the targeted package and CPU hotlug is disabled. + */ +static int rapl_detect_domains(struct rapl_package *rp, int cpu) +{ + int i; + int ret = 0; + struct rapl_domain *rd; + u64 locked; + + for (i = 0; i < RAPL_DOMAIN_MAX; i++) { + /* use physical package id to read counters */ + if (!rapl_check_domain(cpu, i)) + rp->domain_map |= 1 << i; + } + rp->nr_domains = bitmap_weight(&rp->domain_map, RAPL_DOMAIN_MAX); + if (!rp->nr_domains) { + pr_err("no valid rapl domains found in package %d\n", rp->id); + ret = -ENODEV; + goto done; + } + pr_debug("found %d domains on package %d\n", rp->nr_domains, rp->id); + + rp->domains = kcalloc(rp->nr_domains + 1, sizeof(struct rapl_domain), + GFP_KERNEL); + if (!rp->domains) { + ret = -ENOMEM; + goto done; + } + rapl_init_domains(rp); + + for (rd = rp->domains; rd < rp->domains + rp->nr_domains; rd++) { + /* check if the domain is locked by BIOS */ + if (rapl_read_data_raw(rd, FW_LOCK, false, &locked)) { + pr_info("RAPL package %d domain %s locked by BIOS\n", + rp->id, rd->name); + rd->state |= DOMAIN_STATE_BIOS_LOCKED; + } + } + + +done: + return ret; +} + +static bool is_package_new(int package) +{ + struct rapl_package *rp; + + /* caller prevents cpu hotplug, there will be no new packages added + * or deleted while traversing the package list, no need for locking. + */ + list_for_each_entry(rp, &rapl_packages, plist) + if (package == rp->id) + return false; + + return true; +} + +/* RAPL interface can be made of a two-level hierarchy: package level and domain + * level. We first detect the number of packages then domains of each package. + * We have to consider the possiblity of CPU online/offline due to hotplug and + * other scenarios. + */ +static int rapl_detect_topology(void) +{ + int i; + int phy_package_id; + struct rapl_package *new_package, *rp; + + for_each_online_cpu(i) { + phy_package_id = topology_physical_package_id(i); + if (is_package_new(phy_package_id)) { + new_package = kzalloc(sizeof(*rp), GFP_KERNEL); + if (!new_package) { + rapl_cleanup_data(); + return -ENOMEM; + } + /* add the new package to the list */ + new_package->id = phy_package_id; + new_package->nr_cpus = 1; + + /* check if the package contains valid domains */ + if (rapl_detect_domains(new_package, i) || + rapl_check_unit(new_package, i)) { + kfree(new_package->domains); + kfree(new_package); + /* free up the packages already initialized */ + rapl_cleanup_data(); + return -ENODEV; + } + INIT_LIST_HEAD(&new_package->plist); + list_add(&new_package->plist, &rapl_packages); + } else { + rp = find_package_by_id(phy_package_id); + if (rp) + ++rp->nr_cpus; + } + } + + return 0; +} + +/* called from CPU hotplug notifier, hotplug lock held */ +static void rapl_remove_package(struct rapl_package *rp) +{ + struct rapl_domain *rd, *rd_package = NULL; + + for (rd = rp->domains; rd < rp->domains + rp->nr_domains; rd++) { + if (rd->id == RAPL_DOMAIN_PACKAGE) { + rd_package = rd; + continue; + } + pr_debug("remove package %d, %s domain\n", rp->id, rd->name); + powercap_unregister_zone(control_type, &rd->power_zone); + } + /* do parent zone last */ + powercap_unregister_zone(control_type, &rd_package->power_zone); + list_del(&rp->plist); + kfree(rp); +} + +/* called from CPU hotplug notifier, hotplug lock held */ +static int rapl_add_package(int cpu) +{ + int ret = 0; + int phy_package_id; + struct rapl_package *rp; + + phy_package_id = topology_physical_package_id(cpu); + rp = kzalloc(sizeof(struct rapl_package), GFP_KERNEL); + if (!rp) + return -ENOMEM; + + /* add the new package to the list */ + rp->id = phy_package_id; + rp->nr_cpus = 1; + /* check if the package contains valid domains */ + if (rapl_detect_domains(rp, cpu) || + rapl_check_unit(rp, cpu)) { + ret = -ENODEV; + goto err_free_package; + } + if (!rapl_package_register_powercap(rp)) { + INIT_LIST_HEAD(&rp->plist); + list_add(&rp->plist, &rapl_packages); + return ret; + } + +err_free_package: + kfree(rp->domains); + kfree(rp); + + return ret; +} + +/* Handles CPU hotplug on multi-socket systems. + * If a CPU goes online as the first CPU of the physical package + * we add the RAPL package to the system. Similarly, when the last + * CPU of the package is removed, we remove the RAPL package and its + * associated domains. Cooling devices are handled accordingly at + * per-domain level. + */ +static int rapl_cpu_callback(struct notifier_block *nfb, + unsigned long action, void *hcpu) +{ + unsigned long cpu = (unsigned long)hcpu; + int phy_package_id; + struct rapl_package *rp; + + phy_package_id = topology_physical_package_id(cpu); + switch (action) { + case CPU_ONLINE: + case CPU_ONLINE_FROZEN: + case CPU_DOWN_FAILED: + case CPU_DOWN_FAILED_FROZEN: + rp = find_package_by_id(phy_package_id); + if (rp) + ++rp->nr_cpus; + else + rapl_add_package(cpu); + break; + case CPU_DOWN_PREPARE: + case CPU_DOWN_PREPARE_FROZEN: + rp = find_package_by_id(phy_package_id); + if (!rp) + break; + if (--rp->nr_cpus == 0) + rapl_remove_package(rp); + } + + return NOTIFY_OK; +} + +static struct notifier_block rapl_cpu_notifier = { + .notifier_call = rapl_cpu_callback, +}; + +static int __init rapl_init(void) +{ + int ret = 0; + + if (!x86_match_cpu(rapl_ids)) { + pr_err("driver does not support CPU family %d model %d\n", + boot_cpu_data.x86, boot_cpu_data.x86_model); + + return -ENODEV; + } + /* prevent CPU hotplug during detection */ + get_online_cpus(); + ret = rapl_detect_topology(); + if (ret) + goto done; + + if (rapl_register_powercap()) { + rapl_cleanup_data(); + ret = -ENODEV; + goto done; + } + register_hotcpu_notifier(&rapl_cpu_notifier); +done: + put_online_cpus(); + + return ret; +} + +static void __exit rapl_exit(void) +{ + get_online_cpus(); + unregister_hotcpu_notifier(&rapl_cpu_notifier); + rapl_unregister_powercap(); + rapl_cleanup_data(); + put_online_cpus(); +} + +module_init(rapl_init); +module_exit(rapl_exit); + +MODULE_DESCRIPTION("Driver for Intel RAPL (Running Average Power Limit)"); +MODULE_AUTHOR("Jacob Pan <jacob.jun.pan@intel.com>"); +MODULE_LICENSE("GPL v2"); |
