作者簡(jiǎn)介：

榮濤，csdn博主。

組件概述

Linux性能子系統(tǒng)在性能分析中非常有用。以下顯示了這篇文章中的perf子系統(tǒng)componenet 。

“ perf”是可用于執(zhí)行性能分析的用戶程序。

僅暴露給用戶空間的系統(tǒng)調(diào)用perfeventopen返回一個(gè)perf事件fd。該系統(tǒng)調(diào)用沒有g(shù)libc包裝器。更多信息可以在手冊(cè)頁中閱讀。此功能是最復(fù)雜的功能之一。

“ perf_event”是內(nèi)核中的核心結(jié)構(gòu)。性能事件有幾種類型，例如跟蹤點(diǎn)，軟件，硬件。

我們還可以通過perf event fd將eBPF程序附加到trae事件。

抽象層

以下顯示了perf的抽象層。

每個(gè)類型的性能事件都有一個(gè)對(duì)應(yīng)的PMU（性能監(jiān)視單元）。例如，跟蹤點(diǎn)pmu具有以下pmu。

?

?

static struct pmu perf_tracepoint = {

.task_ctx_nr = perf_sw_context,

.event_init = perf_tp_event_init,

.add = perf_trace_add,

.del = perf_trace_del,

.start = perf_swevent_start,

.stop = perf_swevent_stop,

.read = perf_swevent_read,

};

與硬件相關(guān)的PMU具有與arch-spec有關(guān)的抽象結(jié)構(gòu)，例如'struct x86_pmu'。與硬件相關(guān)的結(jié)構(gòu)將讀取/寫入性能監(jiān)視器MSR。

每個(gè)PMU都通過調(diào)用“ perf_pmu_register”進(jìn)行注冊(cè)。

性能事件上下文

性能可以監(jiān)視cpu相關(guān)事件和任務(wù)相關(guān)事件。他們兩個(gè)都可以有幾個(gè)受監(jiān)視的事件。因此，我們需要一個(gè)上下文來連接事件。這是“ perf_event_context”。

有兩種上下文，軟件和硬件，定義如下：

?

?

enum perf_event_task_context {

perf_invalid_context = -1,

perf_hw_context = 0,

perf_sw_context,

perf_nr_task_contexts,

};

對(duì)于CPU級(jí)別，上下文定義為“?perf_cpu_context”，并在“ struct pmu”中定義為percpu變量。

?

?

struct pmu {

...

struct perf_cpu_context __percpu *pmu_cpu_context;

};

如果PMU是相同類型，則它們將共享一個(gè)“ struct perf_cpu_context”。

?

?

int perf_pmu_register(struct pmu *pmu, const char *name, int type)

{

int cpu, ret, max = PERF_TYPE_MAX;

mutex_lock(&pmus_lock);

...

pmu->pmu_cpu_context = find_pmu_context(pmu->task_ctx_nr);

if (pmu->pmu_cpu_context)

goto got_cpu_context;

ret = -ENOMEM;

pmu->pmu_cpu_context = alloc_percpu(struct perf_cpu_context);

if (!pmu->pmu_cpu_context)

goto free_dev;

for_each_possible_cpu(cpu) {

struct perf_cpu_context *cpuctx;

cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu);

__perf_event_init_context(&cpuctx->ctx);

lockdep_set_class(&cpuctx->ctx.mutex, &cpuctx_mutex);

lockdep_set_class(&cpuctx->ctx.lock, &cpuctx_lock);

cpuctx->ctx.pmu = pmu;

cpuctx->online = cpumask_test_cpu(cpu, perf_online_mask);

__perf_mux_hrtimer_init(cpuctx, cpu);

cpuctx->heap_size = ARRAY_SIZE(cpuctx->heap_default);

cpuctx->heap = cpuctx->heap_default;

}

...

}

下圖顯示了此帖子中的相關(guān)結(jié)構(gòu)。

對(duì)于任務(wù)級(jí)別，“ task_struct”具有如下定義的指針數(shù)組：

?

?

struct task_struct {

struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];

};

下圖顯示了相關(guān)結(jié)構(gòu)，也來自于該帖子。

CPU在線時(shí)將觸發(fā)CPU級(jí)性能事件。但是對(duì)于任務(wù)級(jí)別的perf事件，只能通過運(yùn)行任務(wù)來觸發(fā)它。“ perf_cpu_context”的task_ctx包含當(dāng)前正在運(yùn)行的任務(wù)的perf上下文。

性能事件上下文時(shí)間表

性能的一項(xiàng)工作是安排任務(wù)的perf_event_context的進(jìn)出時(shí)間。

下圖顯示了與性能相關(guān)的任務(wù)計(jì)劃輸入和輸出功能。

最后，將調(diào)用PMU的add和del回調(diào)。讓我們以跟蹤點(diǎn)為例。add回調(diào)是“ perf_trace_add”，而del回調(diào)是“ perf_trace_add”。

?

?

????int perf_trace_add(struct perf_event *p_event, int flags)

{

struct trace_event_call *tp_event = p_event->tp_event;

if (!(flags & PERF_EF_START))

p_event->hw.state = PERF_HES_STOPPED;

/*

* If TRACE_REG_PERF_ADD returns false; no custom action was performed

* and we need to take the default action of enqueueing our event on

* the right per-cpu hlist.

*/

if (!tp_event->class->reg(tp_event, TRACE_REG_PERF_ADD, p_event)) {

struct hlist_head __percpu *pcpu_list;

struct hlist_head *list;

pcpu_list = tp_event->perf_events;

if (WARN_ON_ONCE(!pcpu_list))

return -EINVAL;

list = this_cpu_ptr(pcpu_list);

hlist_add_head_rcu(&p_event->hlist_entry, list);

}

return 0;

}

void perf_trace_del(struct perf_event *p_event, int flags)

{

struct trace_event_call *tp_event = p_event->tp_event;

/*

* If TRACE_REG_PERF_DEL returns false; no custom action was performed

* and we need to take the default action of dequeueing our event from

* the right per-cpu hlist.

*/

if (!tp_event->class->reg(tp_event, TRACE_REG_PERF_DEL, p_event))

hlist_del_rcu(&p_event->hlist_entry);

}

“ perf_event”將被添加或刪除到“ tp_event-> perf_events”列表中。

perf_event_open流

?

?

perf_event_open

->perf_copy_attr

->get_unused_fd_flags(fd)

->perf_event_alloc

->perf_init_event

->perf_try_init_event

->pmu->event_init()

->find_get_context

->perf_install_in_context

->__perf_install_in_context

->add_event_to_ctx

->list_add_event

->perf_group_attach

->add_event_to_ctx

->fd_install

perf_event_open將調(diào)用'pmu-> event_init'來初始化事件。并將perf_event添加到perf_event_context中。

性能跟蹤事件

回顧跟蹤點(diǎn)PMU的定義。

?

?

static struct pmu perf_tracepoint = {

.task_ctx_nr = perf_sw_context,

.event_init = perf_tp_event_init,

.add = perf_trace_add,

.del = perf_trace_del,

.start = perf_swevent_start,

.stop = perf_swevent_stop,

.read = perf_swevent_read,

};

讓我們嘗試看一下perf子系統(tǒng)如何監(jiān)視跟蹤點(diǎn)事件。

性能事件初始化

稱為“ perf_tp_event_init”。

?

?

perf_tp_event_init

->perf_trace_init

->perf_trace_event_init

->perf_trace_event_reg

->tp_event->class->reg(TRACE_REG_PERF_REGISTER)

'perf_trace_init'將找到指定的跟蹤點(diǎn)。

“ perf_trace_event_reg”將分配并初始化“ tp_event_perf_events”列表。并使用TRACE_REG_PERF_REGISTER調(diào)用“ tp_event-> class-> reg”。

?

?

static int perf_trace_event_reg(struct trace_event_call *tp_event,

struct perf_event *p_event)

{

struct hlist_head __percpu *list;

int ret = -ENOMEM;

int cpu;

p_event->tp_event = tp_event;

if (tp_event->perf_refcount++ > 0)

return 0;

list = alloc_percpu(struct hlist_head);

if (!list)

goto fail;

for_each_possible_cpu(cpu)

INIT_HLIST_HEAD(per_cpu_ptr(list, cpu));

tp_event->perf_events = list;

...

ret = tp_event->class->reg(tp_event, TRACE_REG_PERF_REGISTER, NULL);

if (ret)

goto fail;

total_ref_count++;

return 0;

...

}

“ tp_event_> class-> reg”回調(diào)為“ trace_event_reg”。

?

?

int trace_event_reg(struct trace_event_call *call,

enum trace_reg type, void *data)

{

struct trace_event_file *file = data;

WARN_ON(!(call->flags & TRACE_EVENT_FL_TRACEPOINT));

switch (type) {

...

#ifdef CONFIG_PERF_EVENTS

case TRACE_REG_PERF_REGISTER:

return tracepoint_probe_register(call->tp,

call->class->perf_probe,

call);

case TRACE_REG_PERF_UNREGISTER:

tracepoint_probe_unregister(call->tp,

call->class->perf_probe,

call);

return 0;

case TRACE_REG_PERF_OPEN:

case TRACE_REG_PERF_CLOSE:

case TRACE_REG_PERF_ADD:

case TRACE_REG_PERF_DEL:

return 0;

#endif

}

return 0;

}

我們可以看到'call-> class-> perf_probe'將被注冊(cè)到跟蹤點(diǎn)。從我的帖子。我們知道這個(gè)“ perf_probe”是“ perf_trace _ ## call”。

?

?

static notrace void

perf_trace_##call(void *__data, proto)

{

struct trace_event_call *event_call = __data;

struct trace_event_data_offsets_##call __maybe_unused __data_offsets;

struct trace_event_raw_##call *entry;

struct pt_regs *__regs;

u64 __count = 1;

struct task_struct *__task = NULL;

struct hlist_head *head;

int __entry_size;

int __data_size;

int rctx;

__data_size = trace_event_get_offsets_##call(&__data_offsets, args);

head = this_cpu_ptr(event_call->perf_events);

if (!bpf_prog_array_valid(event_call) &&

__builtin_constant_p(!__task) && !__task &&

hlist_empty(head))

return;

__entry_size = ALIGN(__data_size + sizeof(*entry) + sizeof(u32),

sizeof(u64));

__entry_size -= sizeof(u32);

entry = perf_trace_buf_alloc(__entry_size, &__regs, &rctx);

if (!entry)

return;

perf_fetch_caller_regs(__regs);

tstruct

{ assign; }

perf_trace_run_bpf_submit(entry, __entry_size, rctx,

event_call, __count, __regs,

head, __task);

}

如果“ event_call-> perf_events”為空，則表示沒有任何當(dāng)前的perf_event添加到該跟蹤點(diǎn)。這是'perf_event_open'初始化perf_event時(shí)的默認(rèn)狀態(tài)。

性能事件添加

在CPU中調(diào)度任務(wù)時(shí)，將調(diào)用'pmu-> add'，并將'perf_event'鏈接到'event_call-> perf_events'鏈接列表。

性能事件

從CPU調(diào)度任務(wù)后，將調(diào)用“ pmu-> del”，并且將從“ event_call-> perf_events”鏈接列表中刪除“ perf_event”。

性能事件觸發(fā)器

如果'event_call-> perf_events'不為空，則將調(diào)用'perf_trace_run_bpf_submit'。如果沒有附加eBPF程序，則將調(diào)用“ perf_tp_event”。

?

?

? ?void perf_tp_event(u16 event_type, u64 count,

void *record, int entry_size,

struct pt_regs *regs, struct hlist_head *head, int rctx,

struct task_struct *task)

{

struct perf_sample_data data;

struct perf_event *event;

struct perf_raw_record raw = {

.frag = {

.size = entry_size,

.data = record,

},

};

perf_sample_data_init(&data, 0, 0);

data.raw = &raw;

perf_trace_buf_update(record, event_type);

hlist_for_each_entry_rcu(event, head, hlist_entry) {

if (perf_tp_event_match(event, &data, regs))

perf_swevent_event(event, count, &data, regs);

}

...

perf_swevent_put_recursion_context(rctx);

}

對(duì)于“ event_call-> perf_events”列表中的每個(gè)“ perf_event”。它調(diào)用perf_swevent_event觸發(fā)性能事件。

?

?

static void perf_swevent_event(struct perf_event *event,

u64 nr,struct perf_sample_data *data,

struct pt_regs *regs)

{

struct hw_perf_event *hwc = &event->hw;

local64_add(nr, &event->count);

if (!regs)

return;

if (!is_sampling_event(event))

return;

if ((event->attr.sample_type & PERF_SAMPLE_PERIOD)

&& !event->attr.freq) {

data->period = nr;

return perf_swevent_overflow(event, 1, data, regs);

} else

data->period = event->hw.last_period;

if (nr == 1 && hwc->sample_period == 1 &&

!event->attr.freq)

return perf_swevent_overflow(event, 1, data, regs);

if (local64_add_negative(nr, &hwc->period_left))

return;

perf_swevent_overflow(event, 0, data, regs);

}

static void perf_swevent_event(struct perf_event *event,

u64 nr,struct perf_sample_data *data,

struct pt_regs *regs)

{

struct hw_perf_event *hwc = &event->hw;

local64_add(nr, &event->count);

if (!regs)

return;

if (!is_sampling_event(event))

return;

if ((event->attr.sample_type & PERF_SAMPLE_PERIOD)

&& !event->attr.freq) {

data->period = nr;

return perf_swevent_overflow(event, 1, data, regs);

} else

data->period = event->hw.last_period;

if (nr == 1 && hwc->sample_period == 1

&& !event->attr.freq)

return perf_swevent_overflow(event, 1, data, regs);

if (local64_add_negative(nr, &hwc->period_left))

return;

perf_swevent_overflow(event, 0, data, regs);

}

'perf_swevent_event'添加'event-> count'。如果事件未采樣，則僅返回。Tis是性能計(jì)數(shù)模式。如果perf_event在樣本模式下，則需要復(fù)制跟蹤點(diǎn)數(shù)據(jù)。以下是呼叫鏈。

?

?

perf_swevent_overflow

->__perf_event_overflow->event

->overflow_handler(perf_event_output).

軟件性能事件

軟件PMU定義如下：

?

?

static struct pmu perf_swevent = {

.task_ctx_nr = perf_sw_context,

.capabilities = PERF_PMU_CAP_NO_NMI,

.event_init = perf_swevent_init,

.add = perf_swevent_add,

.del = perf_swevent_del,

.start = perf_swevent_start,

.stop = perf_swevent_stop,

.read = perf_swevent_read,

};

性能事件初始化

“ perf_swevent_init”將被調(diào)用。它稱為“ swevent_hlist_get”

?

?

static int perf_swevent_init(struct perf_event *event)

{

u64 event_id = event->attr.config;

if (event->attr.type != PERF_TYPE_SOFTWARE)

return -ENOENT;

/*

* no branch sampling for software events

*/

if (has_branch_stack(event))

return -EOPNOTSUPP;

switch (event_id) {

case PERF_COUNT_SW_CPU_CLOCK:

case PERF_COUNT_SW_TASK_CLOCK:

return -ENOENT;

default:

break;

}

if (event_id >= PERF_COUNT_SW_MAX)

return -ENOENT;

if (!event->parent) {

int err;

err = swevent_hlist_get();

if (err)

return err;

static_key_slow_inc(&perf_swevent_enabled[event_id]);

event->destroy = sw_perf_event_destroy;

}

return 0;

}

這將創(chuàng)建一個(gè)percpu'swhash-> swevent_hlist'列表。還要將perf_swevent_enabled [event_id]設(shè)置為true。

性能事件添加

'perf_swevent_add'將perf_event添加到percpu哈希列表中。

?

?

static int perf_swevent_add(struct perf_event *event, int flags)

{

struct swevent_htable *swhash = this_cpu_ptr(&swevent_htable);

struct hw_perf_event *hwc = &event->hw;

struct hlist_head *head;

if (is_sampling_event(event)) {

hwc->last_period = hwc->sample_period;

perf_swevent_set_period(event);

}

hwc->state = !(flags & PERF_EF_START);

head = find_swevent_head(swhash, event);

if (WARN_ON_ONCE(!head))

return -EINVAL;

hlist_add_head_rcu(&event->hlist_entry, head);

perf_event_update_userpage(event);

return 0;

}

性能事件

'perf_swevent_del'從哈希列表中刪除。

?

?

static void perf_swevent_del(struct perf_event *event, int flags)

{

hlist_del_rcu(&event->hlist_entry);

}

性能事件觸發(fā)器

以任務(wù)開關(guān)為例。

“ perf_sw_event_sched”將被調(diào)用。

?

?

static inline void perf_event_task_sched_out(struct task_struct *prev,

struct task_struct *next)

{

perf_sw_event_sched(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 0);

if (static_branch_unlikely(&perf_sched_events))

__perf_event_task_sched_out(prev, next);

}

在perf_event_task_sched_out->?_perf_sw_event-> do_perf_sw_event調(diào)用鏈之后。

?

?

static void do_perf_sw_event(enum perf_type_id type, u32 event_id,

u64 nr,

struct perf_sample_data *data,

struct pt_regs *regs)

{

struct swevent_htable *swhash = this_cpu_ptr(&swevent_htable);

struct perf_event *event;

struct hlist_head *head;

rcu_read_lock();

head = find_swevent_head_rcu(swhash, type, event_id);

if (!head)

goto end;

hlist_for_each_entry_rcu(event, head, hlist_entry) {

if (perf_swevent_match(event, type, event_id, data, regs))

perf_swevent_event(event, nr, data, regs);

}

end:

rcu_read_unlock();

}

如我們所見，它最終會(huì)調(diào)用“ perf_swevent_event”來觸發(fā)事件。

硬件性能事件

硬件PMU之一定義如下：

?

?

static struct pmu pmu = {

.pmu_enable = x86_pmu_enable,

.pmu_disable = x86_pmu_disable,

.attr_groups = x86_pmu_attr_groups,

.event_init = x86_pmu_event_init,

.event_mapped = x86_pmu_event_mapped,

.event_unmapped = x86_pmu_event_unmapped,

.add = x86_pmu_add,

.del = x86_pmu_del,

.start = x86_pmu_start,

.stop = x86_pmu_stop,

.read = x86_pmu_read,

.start_txn = x86_pmu_start_txn,

.cancel_txn = x86_pmu_cancel_txn,

.commit_txn = x86_pmu_commit_txn,

.event_idx = x86_pmu_event_idx,

.sched_task = x86_pmu_sched_task,

.task_ctx_size = sizeof(struct x86_perf_task_context),

.swap_task_ctx = x86_pmu_swap_task_ctx,

.check_period = x86_pmu_check_period,

.aux_output_match = x86_pmu_aux_output_match,

};

硬件性能事件非常復(fù)雜，因?yàn)樗鼘⑴c硬件交互。這里不會(huì)深入介紹硬件。

性能事件初始化

?

?

x86_pmu_event_init

->__x86_pmu_event_init

->x86_reserve_hardware

->x86_pmu.hw_config()

->validate_event

此處的“ x86_pmu”是基于arch規(guī)范的PMU結(jié)構(gòu)。

性能事件添加

x86_pmu_add->收集事件->-> x86_pmu.schedule_events（）-> x86_pmu.add

'collect_events'集

?

?

cpuc->event_list[n] = leader;

性能事件

x86_pmu_del將刪除“ cpuc-> event_list”中的事件。

性能事件觸發(fā)器

觸發(fā)硬件事件時(shí)，它將觸發(fā)NMI中斷。此處理程序是“ perf_event_nmi_handler”。

?

?

static int

perf_event_nmi_handler(unsigned int cmd, struct pt_regs *regs)

{

u64 start_clock;

u64 finish_clock;

int ret;

/*

* All PMUs/events that share this PMI handler should make sure to

* increment active_events for their events.

*/

if (!atomic_read(&active_events))

return NMI_DONE;

start_clock = sched_clock();

ret = x86_pmu.handle_irq(regs);

finish_clock = sched_clock();

perf_sample_event_took(finish_clock - start_clock);

return ret;

}

以Taks'x86_pmu.handle_irq'= x86_pmu_handle_irq為例。

?

?

for (idx = 0; idx < x86_pmu.num_counters; idx++) {

if (!test_bit(idx, cpuc->active_mask))

continue;

event = cpuc->events[idx];

val = x86_perf_event_update(event);

if (val & (1ULL << (x86_pmu.cntval_bits - 1)))

continue;

/*

* event overflow

*/

handled++;

perf_sample_data_init(&data, 0, event->hw.last_period);

if (!x86_perf_event_set_period(event))

continue;

if (perf_event_overflow(event, &data, regs))

x86_pmu_stop(event, 0);

}

在這里，我們可以看到它對(duì)“ cpuc”進(jìn)行了迭代，以查找觸發(fā)該中斷的事件。

　　審核編輯：湯梓紅