diff options
Diffstat (limited to 'arch/arm/kernel')
-rw-r--r-- | arch/arm/kernel/Makefile | 4 | ||||
-rw-r--r-- | arch/arm/kernel/perf_event.c | 391 | ||||
-rw-r--r-- | arch/arm/kernel/perf_event_cpu.c | 351 | ||||
-rw-r--r-- | arch/arm/kernel/perf_event_v6.c | 152 | ||||
-rw-r--r-- | arch/arm/kernel/perf_event_v7.c | 330 | ||||
-rw-r--r-- | arch/arm/kernel/perf_event_xscale.c | 179 | ||||
-rw-r--r-- | arch/arm/kernel/pmu.c | 36 | ||||
-rw-r--r-- | arch/arm/kernel/topology.c | 118 |
8 files changed, 912 insertions, 649 deletions
diff --git a/arch/arm/kernel/Makefile b/arch/arm/kernel/Makefile index 7ad2d5cf700..786ea76a3e2 100644 --- a/arch/arm/kernel/Makefile +++ b/arch/arm/kernel/Makefile @@ -69,8 +69,8 @@ obj-$(CONFIG_CPU_XSC3) += xscale-cp0.o obj-$(CONFIG_CPU_MOHAWK) += xscale-cp0.o obj-$(CONFIG_CPU_PJ4) += pj4-cp0.o obj-$(CONFIG_IWMMXT) += iwmmxt.o -obj-$(CONFIG_CPU_HAS_PMU) += pmu.o -obj-$(CONFIG_HW_PERF_EVENTS) += perf_event.o +obj-$(CONFIG_HW_PERF_EVENTS) += perf_event.o perf_event_cpu.o +obj-$(CONFIG_HW_PERF_EVENTS) += perf_event_xscale.o perf_event_v6.o perf_event_v7.o AFLAGS_iwmmxt.o := -Wa,-mcpu=iwmmxt obj-$(CONFIG_ARM_CPU_TOPOLOGY) += topology.o diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c index ab243b87118..26dd5c0cf65 100644 --- a/arch/arm/kernel/perf_event.c +++ b/arch/arm/kernel/perf_event.c @@ -12,68 +12,15 @@ */ #define pr_fmt(fmt) "hw perfevents: " fmt -#include <linux/bitmap.h> -#include <linux/interrupt.h> #include <linux/kernel.h> -#include <linux/export.h> -#include <linux/perf_event.h> #include <linux/platform_device.h> -#include <linux/spinlock.h> +#include <linux/pm_runtime.h> #include <linux/uaccess.h> -#include <asm/cputype.h> -#include <asm/irq.h> #include <asm/irq_regs.h> #include <asm/pmu.h> #include <asm/stacktrace.h> -/* - * ARMv6 supports a maximum of 3 events, starting from index 0. If we add - * another platform that supports more, we need to increase this to be the - * largest of all platforms. - * - * ARMv7 supports up to 32 events: - * cycle counter CCNT + 31 events counters CNT0..30. - * Cortex-A8 has 1+4 counters, Cortex-A9 has 1+6 counters. - */ -#define ARMPMU_MAX_HWEVENTS 32 - -static DEFINE_PER_CPU(struct perf_event * [ARMPMU_MAX_HWEVENTS], hw_events); -static DEFINE_PER_CPU(unsigned long [BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)], used_mask); -static DEFINE_PER_CPU(struct pmu_hw_events, cpu_hw_events); - -#define to_arm_pmu(p) (container_of(p, struct arm_pmu, pmu)) - -/* Set at runtime when we know what CPU type we are. */ -static struct arm_pmu *cpu_pmu; - -const char *perf_pmu_name(void) -{ - if (!cpu_pmu) - return NULL; - - return cpu_pmu->pmu.name; -} -EXPORT_SYMBOL_GPL(perf_pmu_name); - -int perf_num_counters(void) -{ - int max_events = 0; - - if (cpu_pmu != NULL) - max_events = cpu_pmu->num_events; - - return max_events; -} -EXPORT_SYMBOL_GPL(perf_num_counters); - -#define HW_OP_UNSUPPORTED 0xFFFF - -#define C(_x) \ - PERF_COUNT_HW_CACHE_##_x - -#define CACHE_OP_UNSUPPORTED 0xFFFF - static int armpmu_map_cache_event(const unsigned (*cache_map) [PERF_COUNT_HW_CACHE_MAX] @@ -104,7 +51,7 @@ armpmu_map_cache_event(const unsigned (*cache_map) } static int -armpmu_map_event(const unsigned (*event_map)[PERF_COUNT_HW_MAX], u64 config) +armpmu_map_hw_event(const unsigned (*event_map)[PERF_COUNT_HW_MAX], u64 config) { int mapping = (*event_map)[config]; return mapping == HW_OP_UNSUPPORTED ? -ENOENT : mapping; @@ -116,19 +63,20 @@ armpmu_map_raw_event(u32 raw_event_mask, u64 config) return (int)(config & raw_event_mask); } -static int map_cpu_event(struct perf_event *event, - const unsigned (*event_map)[PERF_COUNT_HW_MAX], - const unsigned (*cache_map) - [PERF_COUNT_HW_CACHE_MAX] - [PERF_COUNT_HW_CACHE_OP_MAX] - [PERF_COUNT_HW_CACHE_RESULT_MAX], - u32 raw_event_mask) +int +armpmu_map_event(struct perf_event *event, + const unsigned (*event_map)[PERF_COUNT_HW_MAX], + const unsigned (*cache_map) + [PERF_COUNT_HW_CACHE_MAX] + [PERF_COUNT_HW_CACHE_OP_MAX] + [PERF_COUNT_HW_CACHE_RESULT_MAX], + u32 raw_event_mask) { u64 config = event->attr.config; switch (event->attr.type) { case PERF_TYPE_HARDWARE: - return armpmu_map_event(event_map, config); + return armpmu_map_hw_event(event_map, config); case PERF_TYPE_HW_CACHE: return armpmu_map_cache_event(cache_map, config); case PERF_TYPE_RAW: @@ -138,12 +86,10 @@ static int map_cpu_event(struct perf_event *event, return -ENOENT; } -int -armpmu_event_set_period(struct perf_event *event, - struct hw_perf_event *hwc, - int idx) +int armpmu_event_set_period(struct perf_event *event) { struct arm_pmu *armpmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; s64 left = local64_read(&hwc->period_left); s64 period = hwc->sample_period; int ret = 0; @@ -167,24 +113,22 @@ armpmu_event_set_period(struct perf_event *event, local64_set(&hwc->prev_count, (u64)-left); - armpmu->write_counter(idx, (u64)(-left) & 0xffffffff); + armpmu->write_counter(event, (u64)(-left) & 0xffffffff); perf_event_update_userpage(event); return ret; } -u64 -armpmu_event_update(struct perf_event *event, - struct hw_perf_event *hwc, - int idx) +u64 armpmu_event_update(struct perf_event *event) { struct arm_pmu *armpmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; u64 delta, prev_raw_count, new_raw_count; again: prev_raw_count = local64_read(&hwc->prev_count); - new_raw_count = armpmu->read_counter(idx); + new_raw_count = armpmu->read_counter(event); if (local64_cmpxchg(&hwc->prev_count, prev_raw_count, new_raw_count) != prev_raw_count) @@ -207,7 +151,7 @@ armpmu_read(struct perf_event *event) if (hwc->idx < 0) return; - armpmu_event_update(event, hwc, hwc->idx); + armpmu_event_update(event); } static void @@ -221,15 +165,13 @@ armpmu_stop(struct perf_event *event, int flags) * PERF_EF_UPDATE, see comments in armpmu_start(). */ if (!(hwc->state & PERF_HES_STOPPED)) { - armpmu->disable(hwc, hwc->idx); - barrier(); /* why? */ - armpmu_event_update(event, hwc, hwc->idx); + armpmu->disable(event); + armpmu_event_update(event); hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE; } } -static void -armpmu_start(struct perf_event *event, int flags) +static void armpmu_start(struct perf_event *event, int flags) { struct arm_pmu *armpmu = to_arm_pmu(event->pmu); struct hw_perf_event *hwc = &event->hw; @@ -249,8 +191,8 @@ armpmu_start(struct perf_event *event, int flags) * get an interrupt too soon or *way* too late if the overflow has * happened since disabling. */ - armpmu_event_set_period(event, hwc, hwc->idx); - armpmu->enable(hwc, hwc->idx); + armpmu_event_set_period(event); + armpmu->enable(event); } static void @@ -282,7 +224,7 @@ armpmu_add(struct perf_event *event, int flags) perf_pmu_disable(event->pmu); /* If we don't have a space for the counter then finish early. */ - idx = armpmu->get_event_idx(hw_events, hwc); + idx = armpmu->get_event_idx(hw_events, event); if (idx < 0) { err = idx; goto out; @@ -293,7 +235,7 @@ armpmu_add(struct perf_event *event, int flags) * sure it is disabled. */ event->hw.idx = idx; - armpmu->disable(hwc, idx); + armpmu->disable(event); hw_events->events[idx] = event; hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE; @@ -313,13 +255,12 @@ validate_event(struct pmu_hw_events *hw_events, struct perf_event *event) { struct arm_pmu *armpmu = to_arm_pmu(event->pmu); - struct hw_perf_event fake_event = event->hw; struct pmu *leader_pmu = event->group_leader->pmu; if (event->pmu != leader_pmu || event->state <= PERF_EVENT_STATE_OFF) return 1; - return armpmu->get_event_idx(hw_events, &fake_event) >= 0; + return armpmu->get_event_idx(hw_events, event) >= 0; } static int @@ -350,99 +291,41 @@ validate_group(struct perf_event *event) return 0; } -static irqreturn_t armpmu_platform_irq(int irq, void *dev) +static irqreturn_t armpmu_dispatch_irq(int irq, void *dev) { struct arm_pmu *armpmu = (struct arm_pmu *) dev; struct platform_device *plat_device = armpmu->plat_device; struct arm_pmu_platdata *plat = dev_get_platdata(&plat_device->dev); - return plat->handle_irq(irq, dev, armpmu->handle_irq); + if (plat && plat->handle_irq) + return plat->handle_irq(irq, dev, armpmu->handle_irq); + else + return armpmu->handle_irq(irq, dev); } static void armpmu_release_hardware(struct arm_pmu *armpmu) { - int i, irq, irqs; - struct platform_device *pmu_device = armpmu->plat_device; - struct arm_pmu_platdata *plat = - dev_get_platdata(&pmu_device->dev); - - irqs = min(pmu_device->num_resources, num_possible_cpus()); - - for (i = 0; i < irqs; ++i) { - if (!cpumask_test_and_clear_cpu(i, &armpmu->active_irqs)) - continue; - irq = platform_get_irq(pmu_device, i); - if (irq >= 0) { - if (plat && plat->disable_irq) - plat->disable_irq(irq); - free_irq(irq, armpmu); - } - } - - release_pmu(armpmu->type); + armpmu->free_irq(armpmu); + pm_runtime_put_sync(&armpmu->plat_device->dev); } static int armpmu_reserve_hardware(struct arm_pmu *armpmu) { - struct arm_pmu_platdata *plat; - irq_handler_t handle_irq; - int i, err, irq, irqs; + int err; struct platform_device *pmu_device = armpmu->plat_device; if (!pmu_device) return -ENODEV; - err = reserve_pmu(armpmu->type); + pm_runtime_get_sync(&pmu_device->dev); + err = armpmu->request_irq(armpmu, armpmu_dispatch_irq); if (err) { - pr_warning("unable to reserve pmu\n"); + armpmu_release_hardware(armpmu); return err; } - plat = dev_get_platdata(&pmu_device->dev); - if (plat && plat->handle_irq) - handle_irq = armpmu_platform_irq; - else - handle_irq = armpmu->handle_irq; - - irqs = min(pmu_device->num_resources, num_possible_cpus()); - if (irqs < 1) { - pr_err("no irqs for PMUs defined\n"); - return -ENODEV; - } - - for (i = 0; i < irqs; ++i) { - err = 0; - irq = platform_get_irq(pmu_device, i); - if (irq < 0) - continue; - - /* - * If we have a single PMU interrupt that we can't shift, - * assume that we're running on a uniprocessor machine and - * continue. Otherwise, continue without this interrupt. - */ - if (irq_set_affinity(irq, cpumask_of(i)) && irqs > 1) { - pr_warning("unable to set irq affinity (irq=%d, cpu=%u)\n", - irq, i); - continue; - } - - err = request_irq(irq, handle_irq, - IRQF_DISABLED | IRQF_NOBALANCING, - "arm-pmu", armpmu); - if (err) { - pr_err("unable to request IRQ%d for ARM PMU counters\n", - irq); - armpmu_release_hardware(armpmu); - return err; - } else if (plat && plat->enable_irq) - plat->enable_irq(irq); - - cpumask_set_cpu(i, &armpmu->active_irqs); - } - return 0; } @@ -536,6 +419,10 @@ static int armpmu_event_init(struct perf_event *event) int err = 0; atomic_t *active_events = &armpmu->active_events; + pr_debug("cpu %d events 0x%llx\n", event->cpu, event->attr.config); + if (event->cpu != -1 && !cpumask_test_cpu(event->cpu, &armpmu->cpus)) + return -ENOENT; + /* does not support taken branch sampling */ if (has_branch_stack(event)) return -EOPNOTSUPP; @@ -572,14 +459,40 @@ static void armpmu_enable(struct pmu *pmu) int enabled = bitmap_weight(hw_events->used_mask, armpmu->num_events); if (enabled) - armpmu->start(); + armpmu->start(armpmu); } static void armpmu_disable(struct pmu *pmu) { struct arm_pmu *armpmu = to_arm_pmu(pmu); - armpmu->stop(); + armpmu->stop(armpmu); +} + +#ifdef CONFIG_PM_RUNTIME +static int armpmu_runtime_resume(struct device *dev) +{ + struct arm_pmu_platdata *plat = dev_get_platdata(dev); + + if (plat && plat->runtime_resume) + return plat->runtime_resume(dev); + + return 0; +} + +static int armpmu_runtime_suspend(struct device *dev) +{ + struct arm_pmu_platdata *plat = dev_get_platdata(dev); + + if (plat && plat->runtime_suspend) + return plat->runtime_suspend(dev); + + return 0; } +#endif + +const struct dev_pm_ops armpmu_dev_pm_ops = { + SET_RUNTIME_PM_OPS(armpmu_runtime_suspend, armpmu_runtime_resume, NULL) +}; static void __init armpmu_init(struct arm_pmu *armpmu) { @@ -598,174 +511,14 @@ static void __init armpmu_init(struct arm_pmu *armpmu) }; } -int __init armpmu_register(struct arm_pmu *armpmu, char *name, int type) +int armpmu_register(struct arm_pmu *armpmu, char *name, int type) { armpmu_init(armpmu); + pr_info("enabled with %s PMU driver, %d counters available\n", + armpmu->name, armpmu->num_events); return perf_pmu_register(&armpmu->pmu, name, type); } -/* Include the PMU-specific implementations. */ -#include "perf_event_xscale.c" -#include "perf_event_v6.c" -#include "perf_event_v7.c" - -/* - * Ensure the PMU has sane values out of reset. - * This requires SMP to be available, so exists as a separate initcall. - */ -static int __init -cpu_pmu_reset(void) -{ - if (cpu_pmu && cpu_pmu->reset) - return on_each_cpu(cpu_pmu->reset, NULL, 1); - return 0; -} -arch_initcall(cpu_pmu_reset); - -/* - * PMU platform driver and devicetree bindings. - */ -static struct of_device_id armpmu_of_device_ids[] = { - {.compatible = "arm,cortex-a9-pmu"}, - {.compatible = "arm,cortex-a8-pmu"}, - {.compatible = "arm,arm1136-pmu"}, - {.compatible = "arm,arm1176-pmu"}, - {}, -}; - -static struct platform_device_id armpmu_plat_device_ids[] = { - {.name = "arm-pmu"}, - {}, -}; - -static int __devinit armpmu_device_probe(struct platform_device *pdev) -{ - if (!cpu_pmu) - return -ENODEV; - - cpu_pmu->plat_device = pdev; - return 0; -} - -static struct platform_driver armpmu_driver = { - .driver = { - .name = "arm-pmu", - .of_match_table = armpmu_of_device_ids, - }, - .probe = armpmu_device_probe, - .id_table = armpmu_plat_device_ids, -}; - -static int __init register_pmu_driver(void) -{ - return platform_driver_register(&armpmu_driver); -} -device_initcall(register_pmu_driver); - -static struct pmu_hw_events *armpmu_get_cpu_events(void) -{ - return &__get_cpu_var(cpu_hw_events); -} - -static void __init cpu_pmu_init(struct arm_pmu *armpmu) -{ - int cpu; - for_each_possible_cpu(cpu) { - struct pmu_hw_events *events = &per_cpu(cpu_hw_events, cpu); - events->events = per_cpu(hw_events, cpu); - events->used_mask = per_cpu(used_mask, cpu); - raw_spin_lock_init(&events->pmu_lock); - } - armpmu->get_hw_events = armpmu_get_cpu_events; - armpmu->type = ARM_PMU_DEVICE_CPU; -} - -/* - * PMU hardware loses all context when a CPU goes offline. - * When a CPU is hotplugged back in, since some hardware registers are - * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading - * junk values out of them. - */ -static int __cpuinit pmu_cpu_notify(struct notifier_block *b, - unsigned long action, void *hcpu) -{ - if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING) - return NOTIFY_DONE; - - if (cpu_pmu && cpu_pmu->reset) - cpu_pmu->reset(NULL); - - return NOTIFY_OK; -} - -static struct notifier_block __cpuinitdata pmu_cpu_notifier = { - .notifier_call = pmu_cpu_notify, -}; - -/* - * CPU PMU identification and registration. - */ -static int __init -init_hw_perf_events(void) -{ - unsigned long cpuid = read_cpuid_id(); - unsigned long implementor = (cpuid & 0xFF000000) >> 24; - unsigned long part_number = (cpuid & 0xFFF0); - - /* ARM Ltd CPUs. */ - if (0x41 == implementor) { - switch (part_number) { - case 0xB360: /* ARM1136 */ - case 0xB560: /* ARM1156 */ - case 0xB760: /* ARM1176 */ - cpu_pmu = armv6pmu_init(); - break; - case 0xB020: /* ARM11mpcore */ - cpu_pmu = armv6mpcore_pmu_init(); - break; - case 0xC080: /* Cortex-A8 */ - cpu_pmu = armv7_a8_pmu_init(); - break; - case 0xC090: /* Cortex-A9 */ - cpu_pmu = armv7_a9_pmu_init(); - break; - case 0xC050: /* Cortex-A5 */ - cpu_pmu = armv7_a5_pmu_init(); - break; - case 0xC0F0: /* Cortex-A15 */ - cpu_pmu = armv7_a15_pmu_init(); - break; - case 0xC070: /* Cortex-A7 */ - cpu_pmu = armv7_a7_pmu_init(); - break; - } - /* Intel CPUs [xscale]. */ - } else if (0x69 == implementor) { - part_number = (cpuid >> 13) & 0x7; - switch (part_number) { - case 1: - cpu_pmu = xscale1pmu_init(); - break; - case 2: - cpu_pmu = xscale2pmu_init(); - break; - } - } - - if (cpu_pmu) { - pr_info("enabled with %s PMU driver, %d counters available\n", - cpu_pmu->name, cpu_pmu->num_events); - cpu_pmu_init(cpu_pmu); - register_cpu_notifier(&pmu_cpu_notifier); - armpmu_register(cpu_pmu, cpu_pmu->name, PERF_TYPE_RAW); - } else { - pr_info("no hardware support available\n"); - } - - return 0; -} -early_initcall(init_hw_perf_events); - /* * Callchain handling code. */ diff --git a/arch/arm/kernel/perf_event_cpu.c b/arch/arm/kernel/perf_event_cpu.c new file mode 100644 index 00000000000..da2bfb32e27 --- /dev/null +++ b/arch/arm/kernel/perf_event_cpu.c @@ -0,0 +1,351 @@ +/* + * 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. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * 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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * Copyright (C) 2012 ARM Limited + * + * Author: Will Deacon <will.deacon@arm.com> + */ +#define pr_fmt(fmt) "CPU PMU: " fmt + +#include <linux/bitmap.h> +#include <linux/cpu_pm.h> +#include <linux/export.h> +#include <linux/kernel.h> +#include <linux/of.h> +#include <linux/platform_device.h> +#include <linux/slab.h> +#include <linux/spinlock.h> + +#include <asm/cputype.h> +#include <asm/irq_regs.h> +#include <asm/pmu.h> + +static DEFINE_PER_CPU(struct perf_event * [ARMPMU_MAX_HWEVENTS], hw_events); +static DEFINE_PER_CPU(unsigned long [BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)], used_mask); +static DEFINE_PER_CPU(struct pmu_hw_events, cpu_hw_events); +static DEFINE_PER_CPU(struct arm_pmu *, armcpu_pmu); +static LIST_HEAD(cpupmu_list); +static DEFINE_PER_CPU(struct cpupmu_regs, cpu_pmu_regs); + +/* + * Despite the names, these two functions are CPU-specific and are used + * by the OProfile/perf code. + */ +const char *perf_pmu_name(void) +{ + struct arm_pmu *cpu_pmu = per_cpu(armcpu_pmu, 0); + if (!cpu_pmu) + return NULL; + + return cpu_pmu->pmu.name; +} +EXPORT_SYMBOL_GPL(perf_pmu_name); + +int perf_num_counters(void) +{ + struct arm_pmu *cpu_pmu = per_cpu(armcpu_pmu, 0); + int max_events = 0; + + if (cpu_pmu != NULL) + max_events = cpu_pmu->num_events; + + return max_events; +} +EXPORT_SYMBOL_GPL(perf_num_counters); + +static struct pmu_hw_events *cpu_pmu_get_cpu_events(void) +{ + return &__get_cpu_var(cpu_hw_events); +} + +static void cpu_pmu_free_irq(struct arm_pmu *cpu_pmu) +{ + int i, irq, irqs; + struct platform_device *pmu_device = cpu_pmu->plat_device; + int cpu = -1; + + irqs = min(pmu_device->num_resources, num_possible_cpus()); + + for (i = 0; i < irqs; ++i) { + cpu = cpumask_next(cpu, &cpu_pmu->cpus); + if (!cpumask_test_and_clear_cpu(cpu, &cpu_pmu->active_irqs)) + continue; + irq = platform_get_irq(pmu_device, i); + if (irq >= 0) + free_irq(irq, cpu_pmu); + } +} + +static int cpu_pmu_request_irq(struct arm_pmu *cpu_pmu, irq_handler_t handler) +{ + int i, err, irq, irqs; + struct platform_device *pmu_device = cpu_pmu->plat_device; + int cpu = -1; + + if (!pmu_device) + return -ENODEV; + + irqs = min(pmu_device->num_resources, num_possible_cpus()); + if (irqs < 1) { + pr_err("no irqs for PMUs defined\n"); + return -ENODEV; + } + + for (i = 0; i < irqs; ++i) { + err = 0; + cpu = cpumask_next(cpu, &cpu_pmu->cpus); + irq = platform_get_irq(pmu_device, i); + if (irq < 0) + continue; + + /* + * If we have a single PMU interrupt that we can't shift, + * assume that we're running on a uniprocessor machine and + * continue. Otherwise, continue without this interrupt. + */ + if (irq_set_affinity(irq, cpumask_of(cpu)) && irqs > 1) { + pr_warning("unable to set irq affinity (irq=%d, cpu=%u)\n", + irq, i); + continue; + } + + err = request_irq(irq, handler, IRQF_NOBALANCING, "arm-pmu", + cpu_pmu); + if (err) { + pr_err("unable to request IRQ%d for ARM PMU counters\n", + irq); + return err; + } + + cpumask_set_cpu(cpu, &cpu_pmu->active_irqs); + } + + return 0; +} + +static void __devinit cpu_pmu_init(struct arm_pmu *cpu_pmu) +{ + int cpu; + for_each_possible_cpu(cpu) { + struct pmu_hw_events *events = &per_cpu(cpu_hw_events, cpu); + events->events = per_cpu(hw_events, cpu); + events->used_mask = per_cpu(used_mask, cpu); + raw_spin_lock_init(&events->pmu_lock); + } + + cpu_pmu->get_hw_events = cpu_pmu_get_cpu_events; + cpu_pmu->request_irq = cpu_pmu_request_irq; + cpu_pmu->free_irq = cpu_pmu_free_irq; +} + +/* + * PMU hardware loses all context when a CPU goes offline. + * When a CPU is hotplugged back in, since some hardware registers are + * UNKNOWN at reset, the PMU must be explicitly reset to avoid reading + * junk values out of them. + */ +static int __cpuinit cpu_pmu_notify(struct notifier_block *b, + unsigned long action, void *hcpu) +{ + int cpu = smp_processor_id(); + struct arm_pmu *cpu_pmu = per_cpu(armcpu_pmu, cpu); + if ((action & ~CPU_TASKS_FROZEN) != CPU_STARTING) + return NOTIFY_DONE; + + if (cpu_pmu && cpu_pmu->reset) + cpu_pmu->reset(cpu_pmu); + + return NOTIFY_OK; +} + +static int cpu_pmu_pm_notify(struct notifier_block *b, + unsigned long action, void *hcpu) +{ + int cpu = smp_processor_id(); + struct arm_pmu *cpu_pmu = per_cpu(armcpu_pmu, cpu); + struct cpupmu_regs *pmuregs = &per_cpu(cpu_pmu_regs, cpu); + + if (action == CPU_PM_ENTER && cpu_pmu->save_regs) { + cpu_pmu->save_regs(cpu_pmu, pmuregs); + cpu_pmu->reset(cpu_pmu); + } else if (action == CPU_PM_EXIT && cpu_pmu->restore_regs) { + cpu_pmu->restore_regs(cpu_pmu, pmuregs); + } + + return NOTIFY_OK; +} + +static struct notifier_block __cpuinitdata cpu_pmu_hotplug_notifier = { + .notifier_call = cpu_pmu_notify, +}; + +static struct notifier_block __cpuinitdata cpu_pmu_pm_notifier = { + .notifier_call = cpu_pmu_pm_notify, +}; + +int cpu_pmu_register(struct cpu_pmu_info *cpupmu_info) +{ + if (!strlen(cpupmu_info->compatible) || !cpupmu_info->init + || (!cpupmu_info->impl && !cpupmu_info->part)) + return -EINVAL; + list_add(&cpupmu_info->entry, &cpupmu_list); + return 0; +} + +struct cpu_pmu_info *cpupmu_match_compatible(const char *compatible) +{ + struct cpu_pmu_info *temp, *pmu_info = ERR_PTR(-ENODEV); + list_for_each_entry(temp, &cpupmu_list, entry) { + if (!strcmp(compatible, temp->compatible)) { + pmu_info = temp; + break; + } + } + return pmu_info; +} + +struct cpu_pmu_info *cpupmu_match_impl_part(unsigned long impl, + unsigned long part) +{ + struct cpu_pmu_info *temp, *pmu_info = ERR_PTR(-ENODEV); + list_for_each_entry(temp, &cpupmu_list, entry) { + if (impl == temp->impl && part == temp->part) { + pmu_info = temp; + break; + } + } + return pmu_info; +} + +/* + * PMU platform driver and devicetree bindings. + */ +static struct of_device_id __devinitdata cpu_pmu_of_device_ids[] = { + {.compatible = "arm,cortex-a15-pmu"}, + {.compatible = "arm,cortex-a9-pmu"}, + {.compatible = "arm,cortex-a8-pmu"}, + {.compatible = "arm,cortex-a7-pmu"}, + {.compatible = "arm,cortex-a5-pmu"}, + {.compatible = "arm,arm11mpcore-pmu"}, + {.compatible = "arm,arm1176-pmu"}, + {.compatible = "arm,arm1136-pmu"}, + {}, +}; + +static struct platform_device_id __devinitdata cpu_pmu_plat_device_ids[] = { + {.name = "arm-pmu"}, + {}, +}; + +/* + * CPU PMU identification and probing. + */ +static int __devinit probe_current_pmu(struct arm_pmu *pmu) +{ + int cpu = get_cpu(); + unsigned long cpuid = read_cpuid_id(); + unsigned long implementor = (cpuid & 0xFF000000) >> 24; + unsigned long part_number = (cpuid & 0xFFF0); + struct cpu_pmu_info *pmu_info = NULL; + int ret = -ENODEV; + + pr_info("probing PMU on CPU %d\n", cpu); + + /* Intel CPUs [xscale]. */ + if (0x69 == implementor) + part_number = (cpuid >> 13) & 0x7; + + pmu_info = cpupmu_match_impl_part(implementor, part_number); + if (!IS_ERR_OR_NULL(pmu_info)) + ret = pmu_info->init(pmu); + put_cpu(); + return ret; +} + +static int __devinit cpu_pmu_device_probe(struct platform_device *pdev) +{ + const struct of_device_id *of_id; + struct cpu_pmu_info *pmu_info = NULL; + struct device_node *node = pdev->dev.of_node, *ncluster; + struct arm_pmu *cpupmu; + int ret = 0; + int cluster = -1; + + cpupmu = kzalloc(sizeof(struct arm_pmu), GFP_KERNEL); + if (IS_ERR_OR_NULL(cpupmu)) { + pr_info("failed to allocate PMU device"); + return -ENOMEM; + } + + if (node && (of_id = of_match_node(cpu_pmu_of_device_ids, pdev->dev.of_node))) { + pmu_info = cpupmu_match_compatible(of_id->compatible); + if (!IS_ERR_OR_NULL(pmu_info)) { + int sibling; + cpumask_t sibling_mask; + smp_call_func_t init = (smp_call_func_t)pmu_info->init; + ncluster = of_parse_phandle(node, "cluster", 0); + if (ncluster) { + int len; + const u32 *hwid; + hwid = of_get_property(ncluster, "reg", &len); + if (hwid && len == 4) + cluster = be32_to_cpup(hwid); + } + /* set sibling mask to all cpu mask if socket is not specified */ + if (cluster == -1 || + cluster_to_logical_mask(cluster, &sibling_mask)) + cpumask_copy(&sibling_mask, cpu_possible_mask); + smp_call_function_any(&sibling_mask, init, cpupmu, 1); + for_each_cpu(sibling, &sibling_mask) + per_cpu(armcpu_pmu, sibling) = cpupmu; + cpumask_copy(&cpupmu->cpus, &sibling_mask); + /* Ensure the PMU has sane values out of reset */ + on_each_cpu_mask(&sibling_mask, cpupmu->reset, cpupmu, 1); + } + } else { + ret = probe_current_pmu(cpupmu); + } + + if (ret) { + pr_info("failed to register PMU devices!"); + kfree(cpupmu); + return ret; + } + + cpupmu->plat_device = pdev; + cpu_pmu_init(cpupmu); + if (!cluster) { + register_cpu_notifier(&cpu_pmu_hotplug_notifier); + cpu_pm_register_notifier(&cpu_pmu_pm_notifier); + } + armpmu_register(cpupmu, cpupmu->name, PERF_TYPE_RAW); + + return 0; +} + +static struct platform_driver cpu_pmu_driver = { + .driver = { + .name = "arm-pmu", + .pm = &armpmu_dev_pm_ops, + .of_match_table = cpu_pmu_of_device_ids, + }, + .probe = cpu_pmu_device_probe, + .id_table = cpu_pmu_plat_device_ids, +}; + +static int __init register_pmu_driver(void) +{ + return platform_driver_register(&cpu_pmu_driver); +} +device_initcall(register_pmu_driver); diff --git a/arch/arm/kernel/perf_event_v6.c b/arch/arm/kernel/perf_event_v6.c index c90fcb2b696..fc8cfef2a29 100644 --- a/arch/arm/kernel/perf_event_v6.c +++ b/arch/arm/kernel/perf_event_v6.c @@ -31,6 +31,15 @@ */ #if defined(CONFIG_CPU_V6) || defined(CONFIG_CPU_V6K) +#include <linux/interrupt.h> +#include <linux/kernel.h> +#include <linux/perf_event.h> +#include <linux/platform_device.h> +#include <linux/spinlock.h> + +#include <asm/irq_regs.h> +#include <asm/pmu.h> + enum armv6_perf_types { ARMV6_PERFCTR_ICACHE_MISS = 0x0, ARMV6_PERFCTR_IBUF_STALL = 0x1, @@ -401,9 +410,10 @@ armv6_pmcr_counter_has_overflowed(unsigned long pmcr, return ret; } -static inline u32 -armv6pmu_read_counter(int counter) +static inline u32 armv6pmu_read_counter(struct perf_event *event) { + struct hw_perf_event *hwc = &event->hw; + int counter = hwc->idx; unsigned long value = 0; if (ARMV6_CYCLE_COUNTER == counter) @@ -418,10 +428,11 @@ armv6pmu_read_counter(int counter) return value; } -static inline void -armv6pmu_write_counter(int counter, - u32 value) +static inline void armv6pmu_write_counter(struct perf_event *event, u32 value) { + struct hw_perf_event *hwc = &event->hw; + int counter = hwc->idx; + if (ARMV6_CYCLE_COUNTER == counter) asm volatile("mcr p15, 0, %0, c15, c12, 1" : : "r"(value)); else if (ARMV6_COUNTER0 == counter) @@ -432,12 +443,13 @@ armv6pmu_write_counter(int counter, WARN_ONCE(1, "invalid counter number (%d)\n", counter); } -static void -armv6pmu_enable_event(struct hw_perf_event *hwc, - int idx) +static void armv6pmu_enable_event(struct perf_event *event) { unsigned long val, mask, evt, flags; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; if (ARMV6_CYCLE_COUNTER == idx) { mask = 0; @@ -473,7 +485,8 @@ armv6pmu_handle_irq(int irq_num, { unsigned long pmcr = armv6_pmcr_read(); struct perf_sample_data data; - struct pmu_hw_events *cpuc; + struct arm_pmu *cpu_pmu = (struct arm_pmu *)dev; + struct pmu_hw_events *cpuc = cpu_pmu->get_hw_events(); struct pt_regs *regs; int idx; @@ -489,7 +502,6 @@ armv6pmu_handle_irq(int irq_num, */ armv6_pmcr_write(pmcr); - cpuc = &__get_cpu_var(cpu_hw_events); for (idx = 0; idx < cpu_pmu->num_events; ++idx) { struct perf_event *event = cpuc->events[idx]; struct hw_perf_event *hwc; @@ -506,13 +518,13 @@ armv6pmu_handle_irq(int irq_num, continue; hwc = &event->hw; - armpmu_event_update(event, hwc, idx); + armpmu_event_update(event); perf_sample_data_init(&data, 0, hwc->last_period); - if (!armpmu_event_set_period(event, hwc, idx)) + if (!armpmu_event_set_period(event)) continue; if (perf_event_overflow(event, &data, regs)) - cpu_pmu->disable(hwc, idx); + cpu_pmu->disable(event); } /* @@ -527,8 +539,7 @@ armv6pmu_handle_irq(int irq_num, return IRQ_HANDLED; } -static void -armv6pmu_start(void) +static void armv6pmu_start(struct arm_pmu *cpu_pmu) { unsigned long flags, val; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -540,8 +551,7 @@ armv6pmu_start(void) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void -armv6pmu_stop(void) +static void armv6pmu_stop(struct arm_pmu *cpu_pmu) { unsigned long flags, val; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -555,10 +565,11 @@ armv6pmu_stop(void) static int armv6pmu_get_event_idx(struct pmu_hw_events *cpuc, - struct hw_perf_event *event) + struct perf_event *event) { + struct hw_perf_event *hwc = &event->hw; /* Always place a cycle counter into the cycle counter. */ - if (ARMV6_PERFCTR_CPU_CYCLES == event->config_base) { + if (ARMV6_PERFCTR_CPU_CYCLES == hwc->config_base) { if (test_and_set_bit(ARMV6_CYCLE_COUNTER, cpuc->used_mask)) return -EAGAIN; @@ -579,12 +590,13 @@ armv6pmu_get_event_idx(struct pmu_hw_events *cpuc, } } -static void -armv6pmu_disable_event(struct hw_perf_event *hwc, - int idx) +static void armv6pmu_disable_event(struct perf_event *event) { unsigned long val, mask, evt, flags; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; if (ARMV6_CYCLE_COUNTER == idx) { mask = ARMV6_PMCR_CCOUNT_IEN; @@ -613,12 +625,13 @@ armv6pmu_disable_event(struct hw_perf_event *hwc, raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void -armv6mpcore_pmu_disable_event(struct hw_perf_event *hwc, - int idx) +static void armv6mpcore_pmu_disable_event(struct perf_event *event) { unsigned long val, mask, flags, evt = 0; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; if (ARMV6_CYCLE_COUNTER == idx) { mask = ARMV6_PMCR_CCOUNT_IEN; @@ -645,28 +658,25 @@ armv6mpcore_pmu_disable_event(struct hw_perf_event *hwc, static int armv6_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv6_perf_map, + return armpmu_map_event(event, &armv6_perf_map, &armv6_perf_cache_map, 0xFF); } -static struct arm_pmu armv6pmu = { - .name = "v6", - .handle_irq = armv6pmu_handle_irq, - .enable = armv6pmu_enable_event, - .disable = armv6pmu_disable_event, - .read_counter = armv6pmu_read_counter, - .write_counter = armv6pmu_write_counter, - .get_event_idx = armv6pmu_get_event_idx, - .start = armv6pmu_start, - .stop = armv6pmu_stop, - .map_event = armv6_map_event, - .num_events = 3, - .max_period = (1LLU << 32) - 1, -}; - -static struct arm_pmu *__init armv6pmu_init(void) +static int __devinit armv6pmu_init(struct arm_pmu *cpu_pmu) { - return &armv6pmu; + cpu_pmu->name = "v6"; + cpu_pmu->handle_irq = armv6pmu_handle_irq; + cpu_pmu->enable = armv6pmu_enable_event; + cpu_pmu->disable = armv6pmu_disable_event; + cpu_pmu->read_counter = armv6pmu_read_counter; + cpu_pmu->write_counter = armv6pmu_write_counter; + cpu_pmu->get_event_idx = armv6pmu_get_event_idx; + cpu_pmu->start = armv6pmu_start; + cpu_pmu->stop = armv6pmu_stop; + cpu_pmu->map_event = armv6_map_event; + cpu_pmu->num_events = 3; + cpu_pmu->max_period = (1LLU << 32) - 1; + return 0; } /* @@ -679,37 +689,43 @@ static struct arm_pmu *__init armv6pmu_init(void) static int armv6mpcore_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv6mpcore_perf_map, + return armpmu_map_event(event, &armv6mpcore_perf_map, &armv6mpcore_perf_cache_map, 0xFF); } -static struct arm_pmu armv6mpcore_pmu = { - .name = "v6mpcore", - .handle_irq = armv6pmu_handle_irq, - .enable = armv6pmu_enable_event, - .disable = armv6mpcore_pmu_disable_event, - .read_counter = armv6pmu_read_counter, - .write_counter = armv6pmu_write_counter, - .get_event_idx = armv6pmu_get_event_idx, - .start = armv6pmu_start, - .stop = armv6pmu_stop, - .map_event = armv6mpcore_map_event, - .num_events = 3, - .max_period = (1LLU << 32) - 1, -}; - -static struct arm_pmu *__init armv6mpcore_pmu_init(void) -{ - return &armv6mpcore_pmu; -} -#else -static struct arm_pmu *__init armv6pmu_init(void) +static int __devinit armv6mpcore_pmu_init(struct arm_pmu *cpu_pmu) { - return NULL; + cpu_pmu->name = "v6mpcore"; + cpu_pmu->handle_irq = armv6pmu_handle_irq; + cpu_pmu->enable = armv6pmu_enable_event; + cpu_pmu->disable = armv6mpcore_pmu_disable_event; + cpu_pmu->read_counter = armv6pmu_read_counter; + cpu_pmu->write_counter = armv6pmu_write_counter; + cpu_pmu->get_event_idx = armv6pmu_get_event_idx; + cpu_pmu->start = armv6pmu_start; + cpu_pmu->stop = armv6pmu_stop; + cpu_pmu->map_event = armv6mpcore_map_event; + cpu_pmu->num_events = 3; + cpu_pmu->max_period = (1LLU << 32) - 1; + return 0; } -static struct arm_pmu *__init armv6mpcore_pmu_init(void) +static struct __devinit cpu_pmu_info armv6_pmu_info[] = { + CPUPMU_INFO_ENTRY(__stringify(arm,arm11mpcore-pmu), 0x41, 0xB020, + armv6mpcore_pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,arm1176-pmu), 0x41, 0xB760, + armv6pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,arm1136-pmu), 0x41, 0xB360, + armv6pmu_init), +}; + +static int __init register_armv6pmu_driver(void) { - return NULL; + int i; + for (i = 0; i < ARRAY_SIZE(armv6_pmu_info); i++) + WARN_ON(cpu_pmu_register(&armv6_pmu_info[i])); + return 0; } + +early_initcall(register_armv6pmu_driver); #endif /* CONFIG_CPU_V6 || CONFIG_CPU_V6K */ diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c index f04070bd218..9314fa986f0 100644 --- a/arch/arm/kernel/perf_event_v7.c +++ b/arch/arm/kernel/perf_event_v7.c @@ -18,7 +18,14 @@ #ifdef CONFIG_CPU_V7 -static struct arm_pmu armv7pmu; +#include <linux/interrupt.h> +#include <linux/kernel.h> +#include <linux/perf_event.h> +#include <linux/platform_device.h> +#include <linux/spinlock.h> + +#include <asm/irq_regs.h> +#include <asm/pmu.h> /* * Common ARMv7 event types @@ -738,7 +745,8 @@ static const unsigned armv7_a7_perf_cache_map[PERF_COUNT_HW_CACHE_MAX] */ #define ARMV7_IDX_CYCLE_COUNTER 0 #define ARMV7_IDX_COUNTER0 1 -#define ARMV7_IDX_COUNTER_LAST (ARMV7_IDX_CYCLE_COUNTER + cpu_pmu->num_events - 1) +#define ARMV7_IDX_COUNTER_LAST(cpu_pmu) \ + (ARMV7_IDX_CYCLE_COUNTER + cpu_pmu->num_events - 1) #define ARMV7_MAX_COUNTERS 32 #define ARMV7_COUNTER_MASK (ARMV7_MAX_COUNTERS - 1) @@ -804,49 +812,34 @@ static inline int armv7_pmnc_has_overflowed(u32 pmnc) return pmnc & ARMV7_OVERFLOWED_MASK; } -static inline int armv7_pmnc_counter_valid(int idx) +static inline int armv7_pmnc_counter_valid(struct arm_pmu *cpu_pmu, int idx) { - return idx >= ARMV7_IDX_CYCLE_COUNTER && idx <= ARMV7_IDX_COUNTER_LAST; + return idx >= ARMV7_IDX_CYCLE_COUNTER && + idx <= ARMV7_IDX_COUNTER_LAST(cpu_pmu); } static inline int armv7_pmnc_counter_has_overflowed(u32 pmnc, int idx) { - int ret = 0; - u32 counter; - - if (!armv7_pmnc_counter_valid(idx)) { - pr_err("CPU%u checking wrong counter %d overflow status\n", - smp_processor_id(), idx); - } else { - counter = ARMV7_IDX_TO_COUNTER(idx); - ret = pmnc & BIT(counter); - } - - return ret; + return pmnc & BIT(ARMV7_IDX_TO_COUNTER(idx)); } static inline int armv7_pmnc_select_counter(int idx) { - u32 counter; - - if (!armv7_pmnc_counter_valid(idx)) { - pr_err("CPU%u selecting wrong PMNC counter %d\n", - smp_processor_id(), idx); - return -EINVAL; - } - - counter = ARMV7_IDX_TO_COUNTER(idx); + u32 counter = ARMV7_IDX_TO_COUNTER(idx); asm volatile("mcr p15, 0, %0, c9, c12, 5" : : "r" (counter)); isb(); return idx; } -static inline u32 armv7pmu_read_counter(int idx) +static inline u32 armv7pmu_read_counter(struct perf_event *event) { + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; u32 value = 0; - if (!armv7_pmnc_counter_valid(idx)) + if (!armv7_pmnc_counter_valid(cpu_pmu, idx)) pr_err("CPU%u reading wrong counter %d\n", smp_processor_id(), idx); else if (idx == ARMV7_IDX_CYCLE_COUNTER) @@ -857,9 +850,13 @@ static inline u32 armv7pmu_read_counter(int idx) return value; } -static inline void armv7pmu_write_counter(int idx, u32 value) +static inline void armv7pmu_write_counter(struct perf_event *event, u32 value) { - if (!armv7_pmnc_counter_valid(idx)) + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + + if (!armv7_pmnc_counter_valid(cpu_pmu, idx)) pr_err("CPU%u writing wrong counter %d\n", smp_processor_id(), idx); else if (idx == ARMV7_IDX_CYCLE_COUNTER) @@ -878,60 +875,28 @@ static inline void armv7_pmnc_write_evtsel(int idx, u32 val) static inline int armv7_pmnc_enable_counter(int idx) { - u32 counter; - - if (!armv7_pmnc_counter_valid(idx)) { - pr_err("CPU%u enabling wrong PMNC counter %d\n", - smp_processor_id(), idx); - return -EINVAL; - } - - counter = ARMV7_IDX_TO_COUNTER(idx); + u32 counter = ARMV7_IDX_TO_COUNTER(idx); asm volatile("mcr p15, 0, %0, c9, c12, 1" : : "r" (BIT(counter))); return idx; } static inline int armv7_pmnc_disable_counter(int idx) { - u32 counter; - - if (!armv7_pmnc_counter_valid(idx)) { - pr_err("CPU%u disabling wrong PMNC counter %d\n", - smp_processor_id(), idx); - return -EINVAL; - } - - counter = ARMV7_IDX_TO_COUNTER(idx); + u32 counter = ARMV7_IDX_TO_COUNTER(idx); asm volatile("mcr p15, 0, %0, c9, c12, 2" : : "r" (BIT(counter))); return idx; } static inline int armv7_pmnc_enable_intens(int idx) { - u32 counter; - - if (!armv7_pmnc_counter_valid(idx)) { - pr_err("CPU%u enabling wrong PMNC counter IRQ enable %d\n", - smp_processor_id(), idx); - return -EINVAL; - } - - counter = ARMV7_IDX_TO_COUNTER(idx); + u32 counter = ARMV7_IDX_TO_COUNTER(idx); asm volatile("mcr p15, 0, %0, c9, c14, 1" : : "r" (BIT(counter))); return idx; } static inline int armv7_pmnc_disable_intens(int idx) { - u32 counter; - - if (!armv7_pmnc_counter_valid(idx)) { - pr_err("CPU%u disabling wrong PMNC counter IRQ enable %d\n", - smp_processor_id(), idx); - return -EINVAL; - } - - counter = ARMV7_IDX_TO_COUNTER(idx); + u32 counter = ARMV7_IDX_TO_COUNTER(idx); asm volatile("mcr p15, 0, %0, c9, c14, 2" : : "r" (BIT(counter))); isb(); /* Clear the overflow flag in case an interrupt is pending. */ @@ -956,7 +921,7 @@ static inline u32 armv7_pmnc_getreset_flags(void) } #ifdef DEBUG -static void armv7_pmnc_dump_regs(void) +static void armv7_pmnc_dump_regs(struct arm_pmu *cpu_pmu) { u32 val; unsigned int cnt; @@ -981,7 +946,8 @@ static void armv7_pmnc_dump_regs(void) asm volatile("mrc p15, 0, %0, c9, c13, 0" : "=r" (val)); printk(KERN_INFO "CCNT =0x%08x\n", val); - for (cnt = ARMV7_IDX_COUNTER0; cnt <= ARMV7_IDX_COUNTER_LAST; cnt++) { + for (cnt = ARMV7_IDX_COUNTER0; + cnt <= ARMV7_IDX_COUNTER_LAST(cpu_pmu); cnt++) { armv7_pmnc_select_counter(cnt); asm volatile("mrc p15, 0, %0, c9, c13, 2" : "=r" (val)); printk(KERN_INFO "CNT[%d] count =0x%08x\n", @@ -993,10 +959,64 @@ static void armv7_pmnc_dump_regs(void) } #endif -static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx) +static void armv7_save_pmu_regs(struct arm_pmu *cpu_pmu, + struct cpupmu_regs *regs) +{ + unsigned int cnt; + asm volatile("mrc p15, 0, %0, c9, c12, 0" : "=r" (regs->pmc)); + if (!(regs->pmc & ARMV7_PMNC_E)) + return; + + asm volatile("mrc p15, 0, %0, c9, c12, 1" : "=r" (regs->pmcntenset)); + asm volatile("mrc p15, 0, %0, c9, c14, 0" : "=r" (regs->pmuseren)); + asm volatile("mrc p15, 0, %0, c9, c14, 1" : "=r" (regs->pmintenset)); + asm volatile("mrc p15, 0, %0, c9, c13, 0" : "=r" (regs->pmxevtcnt[0])); + for (cnt = ARMV7_IDX_COUNTER0; + cnt <= ARMV7_IDX_COUNTER_LAST(cpu_pmu); cnt++) { + armv7_pmnc_select_counter(cnt); + asm volatile("mrc p15, 0, %0, c9, c13, 1" + : "=r"(regs->pmxevttype[cnt])); + asm volatile("mrc p15, 0, %0, c9, c13, 2" + : "=r"(regs->pmxevtcnt[cnt])); + } + return; +} + +static void armv7_restore_pmu_regs(struct arm_pmu *cpu_pmu, + struct cpupmu_regs *regs) +{ + unsigned int cnt; + if (!(regs->pmc & ARMV7_PMNC_E)) + return; + + asm volatile("mcr p15, 0, %0, c9, c12, 1" : : "r" (regs->pmcntenset)); + asm volatile("mcr p15, 0, %0, c9, c14, 0" : : "r" (regs->pmuseren)); + asm volatile("mcr p15, 0, %0, c9, c14, 1" : : "r" (regs->pmintenset)); + asm volatile("mcr p15, 0, %0, c9, c13, 0" : : "r" (regs->pmxevtcnt[0])); + for (cnt = ARMV7_IDX_COUNTER0; + cnt <= ARMV7_IDX_COUNTER_LAST(cpu_pmu); cnt++) { + armv7_pmnc_select_counter(cnt); + asm volatile("mcr p15, 0, %0, c9, c13, 1" + : : "r"(regs->pmxevttype[cnt])); + asm volatile("mcr p15, 0, %0, c9, c13, 2" + : : "r"(regs->pmxevtcnt[cnt])); + } + asm volatile("mcr p15, 0, %0, c9, c12, 0" : : "r" (regs->pmc)); +} + +static void armv7pmu_enable_event(struct perf_event *event) { unsigned long flags; + struct hw_perf_event *hwc = &event->hw; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; + + if (!armv7_pmnc_counter_valid(cpu_pmu, idx)) { + pr_err("CPU%u operation on invalid PMNC counter %d\n", + smp_processor_id(), idx); + return; + } /* * Enable counter and interrupt, and set the counter to count @@ -1014,7 +1034,7 @@ static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx) * We only need to set the event for the cycle counter if we * have the ability to perform event filtering. */ - if (armv7pmu.set_event_filter || idx != ARMV7_IDX_CYCLE_COUNTER) + if (cpu_pmu->set_event_filter || idx != ARMV7_IDX_CYCLE_COUNTER) armv7_pmnc_write_evtsel(idx, hwc->config_base); /* @@ -1030,10 +1050,19 @@ static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void armv7pmu_disable_event(struct hw_perf_event *hwc, int idx) +static void armv7pmu_disable_event(struct perf_event *event) { unsigned long flags; + struct hw_perf_event *hwc = &event->hw; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; + + if (!armv7_pmnc_counter_valid(cpu_pmu, idx)) { + pr_err("CPU%u operation on invalid PMNC counter %d\n", + smp_processor_id(), idx); + return; + } /* * Disable counter and interrupt @@ -1057,7 +1086,8 @@ static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev) { u32 pmnc; struct perf_sample_data data; - struct pmu_hw_events *cpuc; + struct arm_pmu *cpu_pmu = (struct arm_pmu *)dev; + struct pmu_hw_events *cpuc = cpu_pmu->get_hw_events(); struct pt_regs *regs; int idx; @@ -1077,7 +1107,6 @@ static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev) */ regs = get_irq_regs(); - cpuc = &__get_cpu_var(cpu_hw_events); for (idx = 0; idx < cpu_pmu->num_events; ++idx) { struct perf_event *event = cpuc->events[idx]; struct hw_perf_event *hwc; @@ -1094,13 +1123,13 @@ static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev) continue; hwc = &event->hw; - armpmu_event_update(event, hwc, idx); + armpmu_event_update(event); perf_sample_data_init(&data, 0, hwc->last_period); - if (!armpmu_event_set_period(event, hwc, idx)) + if (!armpmu_event_set_period(event)) continue; if (perf_event_overflow(event, &data, regs)) - cpu_pmu->disable(hwc, idx); + cpu_pmu->disable(event); } /* @@ -1115,7 +1144,7 @@ static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev) return IRQ_HANDLED; } -static void armv7pmu_start(void) +static void armv7pmu_start(struct arm_pmu *cpu_pmu) { unsigned long flags; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -1126,7 +1155,7 @@ static void armv7pmu_start(void) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void armv7pmu_stop(void) +static void armv7pmu_stop(struct arm_pmu *cpu_pmu) { unsigned long flags; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -1138,10 +1167,12 @@ static void armv7pmu_stop(void) } static int armv7pmu_get_event_idx(struct pmu_hw_events *cpuc, - struct hw_perf_event *event) + struct perf_event *event) { int idx; - unsigned long evtype = event->config_base & ARMV7_EVTYPE_EVENT; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + unsigned long evtype = hwc->config_base & ARMV7_EVTYPE_EVENT; /* Always place a cycle counter into the cycle counter. */ if (evtype == ARMV7_PERFCTR_CPU_CYCLES) { @@ -1192,11 +1223,14 @@ static int armv7pmu_set_event_filter(struct hw_perf_event *event, static void armv7pmu_reset(void *info) { + struct arm_pmu *cpu_pmu = (struct arm_pmu *)info; u32 idx, nb_cnt = cpu_pmu->num_events; /* The counter and interrupt enable registers are unknown at reset. */ - for (idx = ARMV7_IDX_CYCLE_COUNTER; idx < nb_cnt; ++idx) - armv7pmu_disable_event(NULL, idx); + for (idx = ARMV7_IDX_CYCLE_COUNTER; idx < nb_cnt; ++idx) { + armv7_pmnc_disable_counter(idx); + armv7_pmnc_disable_intens(idx); + } /* Initialize & Reset PMNC: C and P bits */ armv7_pmnc_write(ARMV7_PMNC_P | ARMV7_PMNC_C); @@ -1204,48 +1238,51 @@ static void armv7pmu_reset(void *info) static int armv7_a8_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv7_a8_perf_map, + return armpmu_map_event(event, &armv7_a8_perf_map, &armv7_a8_perf_cache_map, 0xFF); } static int armv7_a9_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv7_a9_perf_map, + return armpmu_map_event(event, &armv7_a9_perf_map, &armv7_a9_perf_cache_map, 0xFF); } static int armv7_a5_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv7_a5_perf_map, + return armpmu_map_event(event, &armv7_a5_perf_map, &armv7_a5_perf_cache_map, 0xFF); } static int armv7_a15_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv7_a15_perf_map, + return armpmu_map_event(event, &armv7_a15_perf_map, &armv7_a15_perf_cache_map, 0xFF); } static int armv7_a7_map_event(struct perf_event *event) { - return map_cpu_event(event, &armv7_a7_perf_map, + return armpmu_map_event(event, &armv7_a7_perf_map, &armv7_a7_perf_cache_map, 0xFF); } -static struct arm_pmu armv7pmu = { - .handle_irq = armv7pmu_handle_irq, - .enable = armv7pmu_enable_event, - .disable = armv7pmu_disable_event, - .read_counter = armv7pmu_read_counter, - .write_counter = armv7pmu_write_counter, - .get_event_idx = armv7pmu_get_event_idx, - .start = armv7pmu_start, - .stop = armv7pmu_stop, - .reset = armv7pmu_reset, - .max_period = (1LLU << 32) - 1, +static void armv7pmu_init(struct arm_pmu *cpu_pmu) +{ + cpu_pmu->handle_irq = armv7pmu_handle_irq; + cpu_pmu->enable = armv7pmu_enable_event; + cpu_pmu->disable = armv7pmu_disable_event; + cpu_pmu->read_counter = armv7pmu_read_counter; + cpu_pmu->write_counter = armv7pmu_write_counter; + cpu_pmu->get_event_idx = armv7pmu_get_event_idx; + cpu_pmu->start = armv7pmu_start; + cpu_pmu->stop = armv7pmu_stop; + cpu_pmu->reset = armv7pmu_reset; + cpu_pmu->save_regs = armv7_save_pmu_regs; + cpu_pmu->restore_regs = armv7_restore_pmu_regs; + cpu_pmu->max_period = (1LLU << 32) - 1; }; -static u32 __init armv7_read_num_pmnc_events(void) +static u32 __devinit armv7_read_num_pmnc_events(void) { u32 nb_cnt; @@ -1256,70 +1293,73 @@ static u32 __init armv7_read_num_pmnc_events(void) return nb_cnt + 1; } -static struct arm_pmu *__init armv7_a8_pmu_init(void) -{ - armv7pmu.name = "ARMv7 Cortex-A8"; - armv7pmu.map_event = armv7_a8_map_event; - armv7pmu.num_events = armv7_read_num_pmnc_events(); - return &armv7pmu; -} - -static struct arm_pmu *__init armv7_a9_pmu_init(void) +static int __devinit armv7_a8_pmu_init(struct arm_pmu *cpu_pmu) { - armv7pmu.name = "ARMv7 Cortex-A9"; - armv7pmu.map_event = armv7_a9_map_event; - armv7pmu.num_events = armv7_read_num_pmnc_events(); - return &armv7pmu; + armv7pmu_init(cpu_pmu); + cpu_pmu->name = "ARMv7 Cortex-A8"; + cpu_pmu->map_event = armv7_a8_map_event; + cpu_pmu->num_events = armv7_read_num_pmnc_events(); + return 0; } -static struct arm_pmu *__init armv7_a5_pmu_init(void) +static int __devinit armv7_a9_pmu_init(struct arm_pmu *cpu_pmu) { - armv7pmu.name = "ARMv7 Cortex-A5"; - armv7pmu.map_event = armv7_a5_map_event; - armv7pmu.num_events = armv7_read_num_pmnc_events(); - return &armv7pmu; + armv7pmu_init(cpu_pmu); + cpu_pmu->name = "ARMv7 Cortex-A9"; + cpu_pmu->map_event = armv7_a9_map_event; + cpu_pmu->num_events = armv7_read_num_pmnc_events(); + return 0; } -static struct arm_pmu *__init armv7_a15_pmu_init(void) +static int __devinit armv7_a5_pmu_init(struct arm_pmu *cpu_pmu) { - armv7pmu.name = "ARMv7 Cortex-A15"; - armv7pmu.map_event = armv7_a15_map_event; - armv7pmu.num_events = armv7_read_num_pmnc_events(); - armv7pmu.set_event_filter = armv7pmu_set_event_filter; - return &armv7pmu; + armv7pmu_init(cpu_pmu); + cpu_pmu->name = "ARMv7 Cortex-A5"; + cpu_pmu->map_event = armv7_a5_map_event; + cpu_pmu->num_events = armv7_read_num_pmnc_events(); + return 0; } -static struct arm_pmu *__init armv7_a7_pmu_init(void) -{ - armv7pmu.name = "ARMv7 Cortex-A7"; - armv7pmu.map_event = armv7_a7_map_event; - armv7pmu.num_events = armv7_read_num_pmnc_events(); - armv7pmu.set_event_filter = armv7pmu_set_event_filter; - return &armv7pmu; -} -#else -static struct arm_pmu *__init armv7_a8_pmu_init(void) +static int __devinit armv7_a15_pmu_init(struct arm_pmu *cpu_pmu) { - return NULL; + armv7pmu_init(cpu_pmu); + cpu_pmu->name = "ARMv7 Cortex-A15"; + cpu_pmu->map_event = armv7_a15_map_event; + cpu_pmu->num_events = armv7_read_num_pmnc_events(); + cpu_pmu->set_event_filter = armv7pmu_set_event_filter; + return 0; } -static struct arm_pmu *__init armv7_a9_pmu_init(void) +static int __devinit armv7_a7_pmu_init(struct arm_pmu *cpu_pmu) { - return NULL; + armv7pmu_init(cpu_pmu); + cpu_pmu->name = "ARMv7 Cortex-A7"; + cpu_pmu->map_event = armv7_a7_map_event; + cpu_pmu->num_events = armv7_read_num_pmnc_events(); + cpu_pmu->set_event_filter = armv7pmu_set_event_filter; + return 0; } -static struct arm_pmu *__init armv7_a5_pmu_init(void) -{ - return NULL; -} +static struct cpu_pmu_info armv7_pmu_info[] __devinitdata = { + CPUPMU_INFO_ENTRY(__stringify(arm,cortex-a15-pmu), 0x41, 0xC0F0, + armv7_a15_pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,cortex-a9-pmu), 0x41, 0xC090, + armv7_a9_pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,cortex-a8-pmu), 0x41, 0xC080, + armv7_a8_pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,cortex-a7-pmu), 0x41, 0xC070, + armv7_a7_pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,cortex-a5-pmu), 0x41, 0xC050, + armv7_a5_pmu_init), +}; -static struct arm_pmu *__init armv7_a15_pmu_init(void) +static int __init register_armv7pmu_driver(void) { - return NULL; + int i; + for (i = 0; i < ARRAY_SIZE(armv7_pmu_info); i++) + WARN_ON(cpu_pmu_register(&armv7_pmu_info[i])); + return 0; } -static struct arm_pmu *__init armv7_a7_pmu_init(void) -{ - return NULL; -} +early_initcall(register_armv7pmu_driver); #endif /* CONFIG_CPU_V7 */ diff --git a/arch/arm/kernel/perf_event_xscale.c b/arch/arm/kernel/perf_event_xscale.c index f759fe0bab6..95e741441c4 100644 --- a/arch/arm/kernel/perf_event_xscale.c +++ b/arch/arm/kernel/perf_event_xscale.c @@ -13,6 +13,15 @@ */ #ifdef CONFIG_CPU_XSCALE +#include <linux/interrupt.h> +#include <linux/kernel.h> +#include <linux/perf_event.h> +#include <linux/platform_device.h> +#include <linux/spinlock.h> + +#include <asm/irq_regs.h> +#include <asm/pmu.h> + enum xscale_perf_types { XSCALE_PERFCTR_ICACHE_MISS = 0x00, XSCALE_PERFCTR_ICACHE_NO_DELIVER = 0x01, @@ -224,7 +233,8 @@ xscale1pmu_handle_irq(int irq_num, void *dev) { unsigned long pmnc; struct perf_sample_data data; - struct pmu_hw_events *cpuc; + struct arm_pmu *cpu_pmu = (struct arm_pmu *)dev; + struct pmu_hw_events *cpuc = cpu_pmu->get_hw_events(); struct pt_regs *regs; int idx; @@ -248,7 +258,6 @@ xscale1pmu_handle_irq(int irq_num, void *dev) regs = get_irq_regs(); - cpuc = &__get_cpu_var(cpu_hw_events); for (idx = 0; idx < cpu_pmu->num_events; ++idx) { struct perf_event *event = cpuc->events[idx]; struct hw_perf_event *hwc; @@ -260,13 +269,13 @@ xscale1pmu_handle_irq(int irq_num, void *dev) continue; hwc = &event->hw; - armpmu_event_update(event, hwc, idx); + armpmu_event_update(event); perf_sample_data_init(&data, 0, hwc->last_period); - if (!armpmu_event_set_period(event, hwc, idx)) + if (!armpmu_event_set_period(event)) continue; if (perf_event_overflow(event, &data, regs)) - cpu_pmu->disable(hwc, idx); + cpu_pmu->disable(event); } irq_work_run(); @@ -280,11 +289,13 @@ xscale1pmu_handle_irq(int irq_num, void *dev) return IRQ_HANDLED; } -static void -xscale1pmu_enable_event(struct hw_perf_event *hwc, int idx) +static void xscale1pmu_enable_event(struct perf_event *event) { unsigned long val, mask, evt, flags; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; switch (idx) { case XSCALE_CYCLE_COUNTER: @@ -314,11 +325,13 @@ xscale1pmu_enable_event(struct hw_perf_event *hwc, int idx) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void -xscale1pmu_disable_event(struct hw_perf_event *hwc, int idx) +static void xscale1pmu_disable_event(struct perf_event *event) { unsigned long val, mask, evt, flags; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; switch (idx) { case XSCALE_CYCLE_COUNTER: @@ -348,9 +361,10 @@ xscale1pmu_disable_event(struct hw_perf_event *hwc, int idx) static int xscale1pmu_get_event_idx(struct pmu_hw_events *cpuc, - struct hw_perf_event *event) + struct perf_event *event) { - if (XSCALE_PERFCTR_CCNT == event->config_base) { + struct hw_perf_event *hwc = &event->hw; + if (XSCALE_PERFCTR_CCNT == hwc->config_base) { if (test_and_set_bit(XSCALE_CYCLE_COUNTER, cpuc->used_mask)) return -EAGAIN; @@ -366,8 +380,7 @@ xscale1pmu_get_event_idx(struct pmu_hw_events *cpuc, } } -static void -xscale1pmu_start(void) +static void xscale1pmu_start(struct arm_pmu *cpu_pmu) { unsigned long flags, val; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -379,8 +392,7 @@ xscale1pmu_start(void) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void -xscale1pmu_stop(void) +static void xscale1pmu_stop(struct arm_pmu *cpu_pmu) { unsigned long flags, val; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -392,9 +404,10 @@ xscale1pmu_stop(void) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static inline u32 -xscale1pmu_read_counter(int counter) +static inline u32 xscale1pmu_read_counter(struct perf_event *event) { + struct hw_perf_event *hwc = &event->hw; + int counter = hwc->idx; u32 val = 0; switch (counter) { @@ -412,9 +425,11 @@ xscale1pmu_read_counter(int counter) return val; } -static inline void -xscale1pmu_write_counter(int counter, u32 val) +static inline void xscale1pmu_write_counter(struct perf_event *event, u32 val) { + struct hw_perf_event *hwc = &event->hw; + int counter = hwc->idx; + switch (counter) { case XSCALE_CYCLE_COUNTER: asm volatile("mcr p14, 0, %0, c1, c0, 0" : : "r" (val)); @@ -430,28 +445,25 @@ xscale1pmu_write_counter(int counter, u32 val) static int xscale_map_event(struct perf_event *event) { - return map_cpu_event(event, &xscale_perf_map, + return armpmu_map_event(event, &xscale_perf_map, &xscale_perf_cache_map, 0xFF); } -static struct arm_pmu xscale1pmu = { - .name = "xscale1", - .handle_irq = xscale1pmu_handle_irq, - .enable = xscale1pmu_enable_event, - .disable = xscale1pmu_disable_event, - .read_counter = xscale1pmu_read_counter, - .write_counter = xscale1pmu_write_counter, - .get_event_idx = xscale1pmu_get_event_idx, - .start = xscale1pmu_start, - .stop = xscale1pmu_stop, - .map_event = xscale_map_event, - .num_events = 3, - .max_period = (1LLU << 32) - 1, -}; - -static struct arm_pmu *__init xscale1pmu_init(void) +static int __devinit xscale1pmu_init(struct arm_pmu *cpu_pmu) { - return &xscale1pmu; + cpu_pmu->name = "xscale1"; + cpu_pmu->handle_irq = xscale1pmu_handle_irq; + cpu_pmu->enable = xscale1pmu_enable_event; + cpu_pmu->disable = xscale1pmu_disable_event; + cpu_pmu->read_counter = xscale1pmu_read_counter; + cpu_pmu->write_counter = xscale1pmu_write_counter; + cpu_pmu->get_event_idx = xscale1pmu_get_event_idx; + cpu_pmu->start = xscale1pmu_start; + cpu_pmu->stop = xscale1pmu_stop; + cpu_pmu->map_event = xscale_map_event; + cpu_pmu->num_events = 3; + cpu_pmu->max_period = (1LLU << 32) - 1; + return 0; } #define XSCALE2_OVERFLOWED_MASK 0x01f @@ -567,7 +579,8 @@ xscale2pmu_handle_irq(int irq_num, void *dev) { unsigned long pmnc, of_flags; struct perf_sample_data data; - struct pmu_hw_events *cpuc; + struct arm_pmu *cpu_pmu = (struct arm_pmu *)dev; + struct pmu_hw_events *cpuc = cpu_pmu->get_hw_events(); struct pt_regs *regs; int idx; @@ -585,7 +598,6 @@ xscale2pmu_handle_irq(int irq_num, void *dev) regs = get_irq_regs(); - cpuc = &__get_cpu_var(cpu_hw_events); for (idx = 0; idx < cpu_pmu->num_events; ++idx) { struct perf_event *event = cpuc->events[idx]; struct hw_perf_event *hwc; @@ -597,13 +609,13 @@ xscale2pmu_handle_irq(int irq_num, void *dev) continue; hwc = &event->hw; - armpmu_event_update(event, hwc, idx); + armpmu_event_update(event); perf_sample_data_init(&data, 0, hwc->last_period); - if (!armpmu_event_set_period(event, hwc, idx)) + if (!armpmu_event_set_period(event)) continue; if (perf_event_overflow(event, &data, regs)) - cpu_pmu->disable(hwc, idx); + cpu_pmu->disable(event); } irq_work_run(); @@ -617,11 +629,13 @@ xscale2pmu_handle_irq(int irq_num, void *dev) return IRQ_HANDLED; } -static void -xscale2pmu_enable_event(struct hw_perf_event *hwc, int idx) +static void xscale2pmu_enable_event(struct perf_event *event) { unsigned long flags, ien, evtsel; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; ien = xscale2pmu_read_int_enable(); evtsel = xscale2pmu_read_event_select(); @@ -661,11 +675,13 @@ xscale2pmu_enable_event(struct hw_perf_event *hwc, int idx) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void -xscale2pmu_disable_event(struct hw_perf_event *hwc, int idx) +static void xscale2pmu_disable_event(struct perf_event *event) { unsigned long flags, ien, evtsel, of_flags; + struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); + int idx = hwc->idx; ien = xscale2pmu_read_int_enable(); evtsel = xscale2pmu_read_event_select(); @@ -713,7 +729,7 @@ xscale2pmu_disable_event(struct hw_perf_event *hwc, int idx) static int xscale2pmu_get_event_idx(struct pmu_hw_events *cpuc, - struct hw_perf_event *event) + struct perf_event *event) { int idx = xscale1pmu_get_event_idx(cpuc, event); if (idx >= 0) @@ -727,8 +743,7 @@ out: return idx; } -static void -xscale2pmu_start(void) +static void xscale2pmu_start(struct arm_pmu *cpu_pmu) { unsigned long flags, val; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -740,8 +755,7 @@ xscale2pmu_start(void) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static void -xscale2pmu_stop(void) +static void xscale2pmu_stop(struct arm_pmu *cpu_pmu) { unsigned long flags, val; struct pmu_hw_events *events = cpu_pmu->get_hw_events(); @@ -753,9 +767,10 @@ xscale2pmu_stop(void) raw_spin_unlock_irqrestore(&events->pmu_lock, flags); } -static inline u32 -xscale2pmu_read_counter(int counter) +static inline u32 xscale2pmu_read_counter(struct perf_event *event) { + struct hw_perf_event *hwc = &event->hw; + int counter = hwc->idx; u32 val = 0; switch (counter) { @@ -779,9 +794,11 @@ xscale2pmu_read_counter(int counter) return val; } -static inline void -xscale2pmu_write_counter(int counter, u32 val) +static inline void xscale2pmu_write_counter(struct perf_event *event, u32 val) { + struct hw_perf_event *hwc = &event->hw; + int counter = hwc->idx; + switch (counter) { case XSCALE_CYCLE_COUNTER: asm volatile("mcr p14, 0, %0, c1, c1, 0" : : "r" (val)); @@ -801,33 +818,37 @@ xscale2pmu_write_counter(int counter, u32 val) } } -static struct arm_pmu xscale2pmu = { - .name = "xscale2", - .handle_irq = xscale2pmu_handle_irq, - .enable = xscale2pmu_enable_event, - .disable = xscale2pmu_disable_event, - .read_counter = xscale2pmu_read_counter, - .write_counter = xscale2pmu_write_counter, - .get_event_idx = xscale2pmu_get_event_idx, - .start = xscale2pmu_start, - .stop = xscale2pmu_stop, - .map_event = xscale_map_event, - .num_events = 5, - .max_period = (1LLU << 32) - 1, +static int __devinit xscale2pmu_init(struct arm_pmu *cpu_pmu) +{ + cpu_pmu->name = "xscale2"; + cpu_pmu->handle_irq = xscale2pmu_handle_irq; + cpu_pmu->enable = xscale2pmu_enable_event; + cpu_pmu->disable = xscale2pmu_disable_event; + cpu_pmu->read_counter = xscale2pmu_read_counter; + cpu_pmu->write_counter = xscale2pmu_write_counter; + cpu_pmu->get_event_idx = xscale2pmu_get_event_idx; + cpu_pmu->start = xscale2pmu_start; + cpu_pmu->stop = xscale2pmu_stop; + cpu_pmu->map_event = xscale_map_event; + cpu_pmu->num_events = 5; + cpu_pmu->max_period = (1LLU << 32) - 1; + return 0; +} + +static struct __devinit cpu_pmu_info xscale_pmu_info[] = { + CPUPMU_INFO_ENTRY(__stringify(arm,xscale1-pmu), 0x69, 0x1, + xscale1pmu_init), + CPUPMU_INFO_ENTRY(__stringify(arm,xscale2-pmu), 0x69, 0x2, + xscale2pmu_init), }; -static struct arm_pmu *__init xscale2pmu_init(void) -{ - return &xscale2pmu; -} -#else -static struct arm_pmu *__init xscale1pmu_init(void) +static int __init register_xscalepmu_driver(void) { - return NULL; + int i; + for (i = 0; i < ARRAY_SIZE(xscale_pmu_info); i++) + WARN_ON(cpu_pmu_register(&xscale_pmu_info[i])); + return 0; } -static struct arm_pmu *__init xscale2pmu_init(void) -{ - return NULL; -} +early_initcall(register_xscalepmu_driver); #endif /* CONFIG_CPU_XSCALE */ diff --git a/arch/arm/kernel/pmu.c b/arch/arm/kernel/pmu.c deleted file mode 100644 index 2334bf8a650..00000000000 --- a/arch/arm/kernel/pmu.c +++ /dev/null @@ -1,36 +0,0 @@ -/* - * linux/arch/arm/kernel/pmu.c - * - * Copyright (C) 2009 picoChip Designs Ltd, Jamie Iles - * Copyright (C) 2010 ARM Ltd, Will Deacon - * - * 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/err.h> -#include <linux/kernel.h> -#include <linux/module.h> - -#include <asm/pmu.h> - -/* - * PMU locking to ensure mutual exclusion between different subsystems. - */ -static unsigned long pmu_lock[BITS_TO_LONGS(ARM_NUM_PMU_DEVICES)]; - -int -reserve_pmu(enum arm_pmu_type type) -{ - return test_and_set_bit_lock(type, pmu_lock) ? -EBUSY : 0; -} -EXPORT_SYMBOL_GPL(reserve_pmu); - -void -release_pmu(enum arm_pmu_type type) -{ - clear_bit_unlock(type, pmu_lock); -} -EXPORT_SYMBOL_GPL(release_pmu); diff --git a/arch/arm/kernel/topology.c b/arch/arm/kernel/topology.c index 26c12c6440f..58dac7a80e5 100644 --- a/arch/arm/kernel/topology.c +++ b/arch/arm/kernel/topology.c @@ -317,6 +317,124 @@ void store_cpu_topology(unsigned int cpuid) cpu_topology[cpuid].socket_id, mpidr); } + +#ifdef CONFIG_SCHED_HMP + +static const char * const little_cores[] = { + "arm,cortex-a7", + NULL, +}; + +static bool is_little_cpu(struct device_node *cn) +{ + const char * const *lc; + for (lc = little_cores; *lc; lc++) + if (of_device_is_compatible(cn, *lc)) + return true; + return false; +} + +void __init arch_get_fast_and_slow_cpus(struct cpumask *fast, + struct cpumask *slow) +{ + struct device_node *cn = NULL; + int cpu = 0; + + cpumask_clear(fast); + cpumask_clear(slow); + + /* + * Use the config options if they are given. This helps testing + * HMP scheduling on systems without a big.LITTLE architecture. + */ + if (strlen(CONFIG_HMP_FAST_CPU_MASK) && strlen(CONFIG_HMP_SLOW_CPU_MASK)) { + if (cpulist_parse(CONFIG_HMP_FAST_CPU_MASK, fast)) + WARN(1, "Failed to parse HMP fast cpu mask!\n"); + if (cpulist_parse(CONFIG_HMP_SLOW_CPU_MASK, slow)) + WARN(1, "Failed to parse HMP slow cpu mask!\n"); + return; + } + + /* + * Else, parse device tree for little cores. + */ + while ((cn = of_find_node_by_type(cn, "cpu"))) { + + if (cpu >= num_possible_cpus()) + break; + + if (is_little_cpu(cn)) + cpumask_set_cpu(cpu, slow); + else + cpumask_set_cpu(cpu, fast); + + cpu++; + } + + if (!cpumask_empty(fast) && !cpumask_empty(slow)) + return; + + /* + * We didn't find both big and little cores so let's call all cores + * fast as this will keep the system running, with all cores being + * treated equal. + */ + cpumask_setall(fast); + cpumask_clear(slow); +} + +void __init arch_get_hmp_domains(struct list_head *hmp_domains_list) +{ + struct cpumask hmp_fast_cpu_mask; + struct cpumask hmp_slow_cpu_mask; + struct hmp_domain *domain; + + arch_get_fast_and_slow_cpus(&hmp_fast_cpu_mask, &hmp_slow_cpu_mask); + + /* + * Initialize hmp_domains + * Must be ordered with respect to compute capacity. + * Fastest domain at head of list. + */ + domain = (struct hmp_domain *) + kmalloc(sizeof(struct hmp_domain), GFP_KERNEL); + cpumask_copy(&domain->cpus, &hmp_slow_cpu_mask); + list_add(&domain->hmp_domains, hmp_domains_list); + domain = (struct hmp_domain *) + kmalloc(sizeof(struct hmp_domain), GFP_KERNEL); + cpumask_copy(&domain->cpus, &hmp_fast_cpu_mask); + list_add(&domain->hmp_domains, hmp_domains_list); +} +#endif /* CONFIG_SCHED_HMP */ + + +/* + * cluster_to_logical_mask - return cpu logical mask of CPUs in a cluster + * @socket_id: cluster HW identifier + * @cluster_mask: the cpumask location to be initialized, modified by the + * function only if return value == 0 + * + * Return: + * + * 0 on success + * -EINVAL if cluster_mask is NULL or there is no record matching socket_id + */ +int cluster_to_logical_mask(unsigned int socket_id, cpumask_t *cluster_mask) +{ + int cpu; + + if (!cluster_mask) + return -EINVAL; + + for_each_online_cpu(cpu) + if (socket_id == topology_physical_package_id(cpu)) { + cpumask_copy(cluster_mask, topology_core_cpumask(cpu)); + return 0; + } + + return -EINVAL; +} + /* * init_cpu_topology is called at boot when only one cpu is running * which prevent simultaneous write access to cpu_topology array |