Commit 6ce4ef94 authored by Shaokun Zhang's avatar Shaokun Zhang Committed by Will Deacon

perf: hisi: Add support for HiSilicon SoC uncore PMU driver

This patch adds support HiSilicon SoC uncore PMU driver framework and
interfaces.
Acked-by: default avatarMark Rutland <mark.rutland@arm.com>
Reviewed-by: default avatarJonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: default avatarShaokun Zhang <zhangshaokun@hisilicon.com>
Signed-off-by: default avatarAnurup M <anurup.m@huawei.com>
[will: Fix leader accounting in uncore group validation]
Signed-off-by: default avatarWill Deacon <will.deacon@arm.com>
parent 3125b5b2
...@@ -17,6 +17,13 @@ config ARM_PMU_ACPI ...@@ -17,6 +17,13 @@ config ARM_PMU_ACPI
depends on ARM_PMU && ACPI depends on ARM_PMU && ACPI
def_bool y def_bool y
config HISI_PMU
bool "HiSilicon SoC PMU"
depends on ARM64 && ACPI
help
Support for HiSilicon SoC uncore performance monitoring
unit (PMU), such as: L3C, HHA and DDRC.
config QCOM_L2_PMU config QCOM_L2_PMU
bool "Qualcomm Technologies L2-cache PMU" bool "Qualcomm Technologies L2-cache PMU"
depends on ARCH_QCOM && ARM64 && ACPI depends on ARCH_QCOM && ARM64 && ACPI
......
obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o obj-$(CONFIG_ARM_PMU) += arm_pmu.o arm_pmu_platform.o
obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o obj-$(CONFIG_ARM_PMU_ACPI) += arm_pmu_acpi.o
obj-$(CONFIG_HISI_PMU) += hisilicon/
obj-$(CONFIG_QCOM_L2_PMU) += qcom_l2_pmu.o obj-$(CONFIG_QCOM_L2_PMU) += qcom_l2_pmu.o
obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o obj-$(CONFIG_QCOM_L3_PMU) += qcom_l3_pmu.o
obj-$(CONFIG_XGENE_PMU) += xgene_pmu.o obj-$(CONFIG_XGENE_PMU) += xgene_pmu.o
......
obj-$(CONFIG_HISI_PMU) += hisi_uncore_pmu.o
/*
* HiSilicon SoC Hardware event counters support
*
* Copyright (C) 2017 Hisilicon Limited
* Author: Anurup M <anurup.m@huawei.com>
* Shaokun Zhang <zhangshaokun@hisilicon.com>
*
* This code is based on the uncore PMUs like arm-cci and arm-ccn.
*
* 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/bitmap.h>
#include <linux/bitops.h>
#include <linux/bug.h>
#include <linux/err.h>
#include <linux/errno.h>
#include <linux/interrupt.h>
#include <asm/local64.h>
#include "hisi_uncore_pmu.h"
#define HISI_GET_EVENTID(ev) (ev->hw.config_base & 0xff)
#define HISI_MAX_PERIOD(nr) (BIT_ULL(nr) - 1)
/*
* PMU format attributes
*/
ssize_t hisi_format_sysfs_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct dev_ext_attribute *eattr;
eattr = container_of(attr, struct dev_ext_attribute, attr);
return sprintf(buf, "%s\n", (char *)eattr->var);
}
/*
* PMU event attributes
*/
ssize_t hisi_event_sysfs_show(struct device *dev,
struct device_attribute *attr, char *page)
{
struct dev_ext_attribute *eattr;
eattr = container_of(attr, struct dev_ext_attribute, attr);
return sprintf(page, "config=0x%lx\n", (unsigned long)eattr->var);
}
/*
* sysfs cpumask attributes. For uncore PMU, we only have a single CPU to show
*/
ssize_t hisi_cpumask_sysfs_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(dev_get_drvdata(dev));
return sprintf(buf, "%d\n", hisi_pmu->on_cpu);
}
static bool hisi_validate_event_group(struct perf_event *event)
{
struct perf_event *sibling, *leader = event->group_leader;
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
/* Include count for the event */
int counters = 1;
if (!is_software_event(leader)) {
/*
* We must NOT create groups containing mixed PMUs, although
* software events are acceptable
*/
if (leader->pmu != event->pmu)
return false;
/* Increment counter for the leader */
if (leader != event)
counters++;
}
list_for_each_entry(sibling, &event->group_leader->sibling_list,
group_entry) {
if (is_software_event(sibling))
continue;
if (sibling->pmu != event->pmu)
return false;
/* Increment counter for each sibling */
counters++;
}
/* The group can not count events more than the counters in the HW */
return counters <= hisi_pmu->num_counters;
}
int hisi_uncore_pmu_counter_valid(struct hisi_pmu *hisi_pmu, int idx)
{
return idx >= 0 && idx < hisi_pmu->num_counters;
}
int hisi_uncore_pmu_get_event_idx(struct perf_event *event)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
unsigned long *used_mask = hisi_pmu->pmu_events.used_mask;
u32 num_counters = hisi_pmu->num_counters;
int idx;
idx = find_first_zero_bit(used_mask, num_counters);
if (idx == num_counters)
return -EAGAIN;
set_bit(idx, used_mask);
return idx;
}
static void hisi_uncore_pmu_clear_event_idx(struct hisi_pmu *hisi_pmu, int idx)
{
if (!hisi_uncore_pmu_counter_valid(hisi_pmu, idx)) {
dev_err(hisi_pmu->dev, "Unsupported event index:%d!\n", idx);
return;
}
clear_bit(idx, hisi_pmu->pmu_events.used_mask);
}
int hisi_uncore_pmu_event_init(struct perf_event *event)
{
struct hw_perf_event *hwc = &event->hw;
struct hisi_pmu *hisi_pmu;
if (event->attr.type != event->pmu->type)
return -ENOENT;
/*
* We do not support sampling as the counters are all
* shared by all CPU cores in a CPU die(SCCL). Also we
* do not support attach to a task(per-process mode)
*/
if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK)
return -EOPNOTSUPP;
/* counters do not have these bits */
if (event->attr.exclude_user ||
event->attr.exclude_kernel ||
event->attr.exclude_host ||
event->attr.exclude_guest ||
event->attr.exclude_hv ||
event->attr.exclude_idle)
return -EINVAL;
/*
* The uncore counters not specific to any CPU, so cannot
* support per-task
*/
if (event->cpu < 0)
return -EINVAL;
/*
* Validate if the events in group does not exceed the
* available counters in hardware.
*/
if (!hisi_validate_event_group(event))
return -EINVAL;
hisi_pmu = to_hisi_pmu(event->pmu);
if (event->attr.config > hisi_pmu->check_event)
return -EINVAL;
if (hisi_pmu->on_cpu == -1)
return -EINVAL;
/*
* We don't assign an index until we actually place the event onto
* hardware. Use -1 to signify that we haven't decided where to put it
* yet.
*/
hwc->idx = -1;
hwc->config_base = event->attr.config;
/* Enforce to use the same CPU for all events in this PMU */
event->cpu = hisi_pmu->on_cpu;
return 0;
}
/*
* Set the counter to count the event that we're interested in,
* and enable interrupt and counter.
*/
static void hisi_uncore_pmu_enable_event(struct perf_event *event)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
hisi_pmu->ops->write_evtype(hisi_pmu, hwc->idx,
HISI_GET_EVENTID(event));
hisi_pmu->ops->enable_counter_int(hisi_pmu, hwc);
hisi_pmu->ops->enable_counter(hisi_pmu, hwc);
}
/*
* Disable counter and interrupt.
*/
static void hisi_uncore_pmu_disable_event(struct perf_event *event)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
hisi_pmu->ops->disable_counter(hisi_pmu, hwc);
hisi_pmu->ops->disable_counter_int(hisi_pmu, hwc);
}
void hisi_uncore_pmu_set_event_period(struct perf_event *event)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
/*
* The HiSilicon PMU counters support 32 bits or 48 bits, depending on
* the PMU. We reduce it to 2^(counter_bits - 1) to account for the
* extreme interrupt latency. So we could hopefully handle the overflow
* interrupt before another 2^(counter_bits - 1) events occur and the
* counter overtakes its previous value.
*/
u64 val = BIT_ULL(hisi_pmu->counter_bits - 1);
local64_set(&hwc->prev_count, val);
/* Write start value to the hardware event counter */
hisi_pmu->ops->write_counter(hisi_pmu, hwc, val);
}
void hisi_uncore_pmu_event_update(struct perf_event *event)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
u64 delta, prev_raw_count, new_raw_count;
do {
/* Read the count from the counter register */
new_raw_count = hisi_pmu->ops->read_counter(hisi_pmu, hwc);
prev_raw_count = local64_read(&hwc->prev_count);
} while (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
new_raw_count) != prev_raw_count);
/*
* compute the delta
*/
delta = (new_raw_count - prev_raw_count) &
HISI_MAX_PERIOD(hisi_pmu->counter_bits);
local64_add(delta, &event->count);
}
void hisi_uncore_pmu_start(struct perf_event *event, int flags)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
return;
WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
hwc->state = 0;
hisi_uncore_pmu_set_event_period(event);
if (flags & PERF_EF_RELOAD) {
u64 prev_raw_count = local64_read(&hwc->prev_count);
hisi_pmu->ops->write_counter(hisi_pmu, hwc, prev_raw_count);
}
hisi_uncore_pmu_enable_event(event);
perf_event_update_userpage(event);
}
void hisi_uncore_pmu_stop(struct perf_event *event, int flags)
{
struct hw_perf_event *hwc = &event->hw;
hisi_uncore_pmu_disable_event(event);
WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
hwc->state |= PERF_HES_STOPPED;
if (hwc->state & PERF_HES_UPTODATE)
return;
/* Read hardware counter and update the perf counter statistics */
hisi_uncore_pmu_event_update(event);
hwc->state |= PERF_HES_UPTODATE;
}
int hisi_uncore_pmu_add(struct perf_event *event, int flags)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
int idx;
hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
/* Get an available counter index for counting */
idx = hisi_pmu->ops->get_event_idx(event);
if (idx < 0)
return idx;
event->hw.idx = idx;
hisi_pmu->pmu_events.hw_events[idx] = event;
if (flags & PERF_EF_START)
hisi_uncore_pmu_start(event, PERF_EF_RELOAD);
return 0;
}
void hisi_uncore_pmu_del(struct perf_event *event, int flags)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
hisi_uncore_pmu_stop(event, PERF_EF_UPDATE);
hisi_uncore_pmu_clear_event_idx(hisi_pmu, hwc->idx);
perf_event_update_userpage(event);
hisi_pmu->pmu_events.hw_events[hwc->idx] = NULL;
}
void hisi_uncore_pmu_read(struct perf_event *event)
{
/* Read hardware counter and update the perf counter statistics */
hisi_uncore_pmu_event_update(event);
}
void hisi_uncore_pmu_enable(struct pmu *pmu)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(pmu);
int enabled = bitmap_weight(hisi_pmu->pmu_events.used_mask,
hisi_pmu->num_counters);
if (!enabled)
return;
hisi_pmu->ops->start_counters(hisi_pmu);
}
void hisi_uncore_pmu_disable(struct pmu *pmu)
{
struct hisi_pmu *hisi_pmu = to_hisi_pmu(pmu);
hisi_pmu->ops->stop_counters(hisi_pmu);
}
/*
* Read Super CPU cluster and CPU cluster ID from MPIDR_EL1.
* If multi-threading is supported, SCCL_ID is in MPIDR[aff3] and CCL_ID
* is in MPIDR[aff2]; if not, SCCL_ID is in MPIDR[aff2] and CCL_ID is
* in MPIDR[aff1]. If this changes in future, this shall be updated.
*/
static void hisi_read_sccl_and_ccl_id(int *sccl_id, int *ccl_id)
{
u64 mpidr = read_cpuid_mpidr();
if (mpidr & MPIDR_MT_BITMASK) {
if (sccl_id)
*sccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 3);
if (ccl_id)
*ccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 2);
} else {
if (sccl_id)
*sccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 2);
if (ccl_id)
*ccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 1);
}
}
/*
* Check whether the CPU is associated with this uncore PMU
*/
static bool hisi_pmu_cpu_is_associated_pmu(struct hisi_pmu *hisi_pmu)
{
int sccl_id, ccl_id;
if (hisi_pmu->ccl_id == -1) {
/* If CCL_ID is -1, the PMU only shares the same SCCL */
hisi_read_sccl_and_ccl_id(&sccl_id, NULL);
return sccl_id == hisi_pmu->sccl_id;
}
hisi_read_sccl_and_ccl_id(&sccl_id, &ccl_id);
return sccl_id == hisi_pmu->sccl_id && ccl_id == hisi_pmu->ccl_id;
}
int hisi_uncore_pmu_online_cpu(unsigned int cpu, struct hlist_node *node)
{
struct hisi_pmu *hisi_pmu = hlist_entry_safe(node, struct hisi_pmu,
node);
if (!hisi_pmu_cpu_is_associated_pmu(hisi_pmu))
return 0;
cpumask_set_cpu(cpu, &hisi_pmu->associated_cpus);
/* If another CPU is already managing this PMU, simply return. */
if (hisi_pmu->on_cpu != -1)
return 0;
/* Use this CPU in cpumask for event counting */
hisi_pmu->on_cpu = cpu;
/* Overflow interrupt also should use the same CPU */
WARN_ON(irq_set_affinity(hisi_pmu->irq, cpumask_of(cpu)));
return 0;
}
int hisi_uncore_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
{
struct hisi_pmu *hisi_pmu = hlist_entry_safe(node, struct hisi_pmu,
node);
cpumask_t pmu_online_cpus;
unsigned int target;
if (!cpumask_test_and_clear_cpu(cpu, &hisi_pmu->associated_cpus))
return 0;
/* Nothing to do if this CPU doesn't own the PMU */
if (hisi_pmu->on_cpu != cpu)
return 0;
/* Give up ownership of the PMU */
hisi_pmu->on_cpu = -1;
/* Choose a new CPU to migrate ownership of the PMU to */
cpumask_and(&pmu_online_cpus, &hisi_pmu->associated_cpus,
cpu_online_mask);
target = cpumask_any_but(&pmu_online_cpus, cpu);
if (target >= nr_cpu_ids)
return 0;
perf_pmu_migrate_context(&hisi_pmu->pmu, cpu, target);
/* Use this CPU for event counting */
hisi_pmu->on_cpu = target;
WARN_ON(irq_set_affinity(hisi_pmu->irq, cpumask_of(target)));
return 0;
}
/*
* HiSilicon SoC Hardware event counters support
*
* Copyright (C) 2017 Hisilicon Limited
* Author: Anurup M <anurup.m@huawei.com>
* Shaokun Zhang <zhangshaokun@hisilicon.com>
*
* This code is based on the uncore PMUs like arm-cci and arm-ccn.
*
* 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.
*/
#ifndef __HISI_UNCORE_PMU_H__
#define __HISI_UNCORE_PMU_H__
#include <linux/cpumask.h>
#include <linux/device.h>
#include <linux/kernel.h>
#include <linux/perf_event.h>
#include <linux/types.h>
#undef pr_fmt
#define pr_fmt(fmt) "hisi_pmu: " fmt
#define HISI_MAX_COUNTERS 0x10
#define to_hisi_pmu(p) (container_of(p, struct hisi_pmu, pmu))
#define HISI_PMU_ATTR(_name, _func, _config) \
(&((struct dev_ext_attribute[]) { \
{ __ATTR(_name, 0444, _func, NULL), (void *)_config } \
})[0].attr.attr)
#define HISI_PMU_FORMAT_ATTR(_name, _config) \
HISI_PMU_ATTR(_name, hisi_format_sysfs_show, (void *)_config)
#define HISI_PMU_EVENT_ATTR(_name, _config) \
HISI_PMU_ATTR(_name, hisi_event_sysfs_show, (unsigned long)_config)
struct hisi_pmu;
struct hisi_uncore_ops {
void (*write_evtype)(struct hisi_pmu *, int, u32);
int (*get_event_idx)(struct perf_event *);
u64 (*read_counter)(struct hisi_pmu *, struct hw_perf_event *);
void (*write_counter)(struct hisi_pmu *, struct hw_perf_event *, u64);
void (*enable_counter)(struct hisi_pmu *, struct hw_perf_event *);
void (*disable_counter)(struct hisi_pmu *, struct hw_perf_event *);
void (*enable_counter_int)(struct hisi_pmu *, struct hw_perf_event *);
void (*disable_counter_int)(struct hisi_pmu *, struct hw_perf_event *);
void (*start_counters)(struct hisi_pmu *);
void (*stop_counters)(struct hisi_pmu *);
};
struct hisi_pmu_hwevents {
struct perf_event *hw_events[HISI_MAX_COUNTERS];
DECLARE_BITMAP(used_mask, HISI_MAX_COUNTERS);
};
/* Generic pmu struct for different pmu types */
struct hisi_pmu {
struct pmu pmu;
const struct hisi_uncore_ops *ops;
struct hisi_pmu_hwevents pmu_events;
/* associated_cpus: All CPUs associated with the PMU */
cpumask_t associated_cpus;
/* CPU used for counting */
int on_cpu;
int irq;
struct device *dev;
struct hlist_node node;
int sccl_id;
int ccl_id;
void __iomem *base;
/* the ID of the PMU modules */
u32 index_id;
int num_counters;
int counter_bits;
/* check event code range */
int check_event;
};
int hisi_uncore_pmu_counter_valid(struct hisi_pmu *hisi_pmu, int idx);
int hisi_uncore_pmu_get_event_idx(struct perf_event *event);
void hisi_uncore_pmu_read(struct perf_event *event);
int hisi_uncore_pmu_add(struct perf_event *event, int flags);
void hisi_uncore_pmu_del(struct perf_event *event, int flags);
void hisi_uncore_pmu_start(struct perf_event *event, int flags);
void hisi_uncore_pmu_stop(struct perf_event *event, int flags);
void hisi_uncore_pmu_set_event_period(struct perf_event *event);
void hisi_uncore_pmu_event_update(struct perf_event *event);
int hisi_uncore_pmu_event_init(struct perf_event *event);
void hisi_uncore_pmu_enable(struct pmu *pmu);
void hisi_uncore_pmu_disable(struct pmu *pmu);
ssize_t hisi_event_sysfs_show(struct device *dev,
struct device_attribute *attr, char *buf);
ssize_t hisi_format_sysfs_show(struct device *dev,
struct device_attribute *attr, char *buf);
ssize_t hisi_cpumask_sysfs_show(struct device *dev,
struct device_attribute *attr, char *buf);
int hisi_uncore_pmu_online_cpu(unsigned int cpu, struct hlist_node *node);
int hisi_uncore_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node);
#endif /* __HISI_UNCORE_PMU_H__ */
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment