Linux内核 eBPF基础：perf（2）：perf性能管理单元PMU的注册

本文主要是介绍Linux内核 eBPF基础：perf（2）：perf性能管理单元PMU的注册，希望对大家解决编程问题提供一定的参考价值，需要的开发者们随着小编来一起学习吧！

Linux内核 eBPF基础 perf（2）：性能管理单元PMU的注册

荣涛 2021年5月18日

• 本文相关注释代码：https://github.com/Rtoax/linux-5.10.13
• Linux内核性能架构：perf_event

1. perf类型

在include\uapi\linux\perf_event.h中有：

/** attr.type*/
enum perf_type_id { /* perf 类型 */PERF_TYPE_HARDWARE			= 0,    /* 硬件 */PERF_TYPE_SOFTWARE			= 1,    /* 软件 */PERF_TYPE_TRACEPOINT		= 2,    /* 跟踪点 */PERF_TYPE_HW_CACHE			= 3,    /* 硬件cache */PERF_TYPE_RAW				= 4,    /* RAW */PERF_TYPE_BREAKPOINT		= 5,    /* 断点 */PERF_TYPE_MAX,				/* non-ABI */
};

他们是传入性能管理单元PMU注册函数perf_pmu_register的字段type。列出注册的PMU：

[rongtao@localhost src]$ grep -r "perf_pmu_register" | grep "\""
arch/x86/events/intel/bts.c:	return perf_pmu_register(&bts_pmu, "intel_bts", -1);
arch/x86/events/intel/pt.c:	ret = perf_pmu_register(&pt_pmu.pmu, "intel_pt", -1);
arch/x86/events/amd/power.c:	ret = perf_pmu_register(&pmu_class, "power", -1);
arch/x86/events/core.c:	err = perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
arch/x86/events/msr.c:	perf_pmu_register(&pmu_msr, "msr", -1);
arch/x86/events/rapl.c:	ret = perf_pmu_register(&rapl_pmus->pmu, "power", -1);
kernel/events/hw_breakpoint.c:	perf_pmu_register(&perf_breakpoint, "breakpoint", PERF_TYPE_BREAKPOINT);
kernel/events/core.c:	perf_pmu_register(&perf_tracepoint, "tracepoint", PERF_TYPE_TRACEPOINT);
kernel/events/core.c:	perf_pmu_register(&perf_kprobe, "kprobe", -1);
kernel/events/core.c:	perf_pmu_register(&perf_uprobe, "uprobe", -1);
kernel/events/core.c:	perf_pmu_register(&perf_swevent, "software", PERF_TYPE_SOFTWARE);kernel/events/core.c:	perf_pmu_register(&perf_cpu_clock, NULL, -1);
kernel/events/core.c:	perf_pmu_register(&perf_task_clock, NULL, -1);

2. perf_pmu_register

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

这里需要注意，函数perf_pmu_register是非常重要的注册函数，注册的pmu将加入全局链表pmus中：

static LIST_HEAD(pmus);

函数perf_pmu_register首先申请per-cpu变量：

pmu->pmu_disable_count = alloc_percpu(int);

接着，如果类型不是PERF_TYPE_SOFTWARE，将分配一个ID（前提是name没有设定，如perf_cpu_clock）

if (type != PERF_TYPE_SOFTWARE) {if (type >= 0)max = type;/* 分配一个ID */ret = idr_alloc(&pmu_idr, pmu, max, 0, GFP_KERNEL);if (ret < 0)goto free_pdc;WARN_ON(type >= 0 && ret != type);type = ret;
}

然后，申请一个设备：

	if (pmu_bus_running/* perf_event_sysfs_init() 中被设置 为 1 */) {ret = pmu_dev_alloc(pmu);   /* 分配一个设备 device- /sys/devices/ */if (ret)goto free_idr;}

接下来这段代码表明，每个hw只能注册一次：

	if (pmu->task_ctx_nr == perf_hw_context) {static int hw_context_taken = 0;/** Other than systems with heterogeneous CPUs, it never makes* sense for two PMUs to share perf_hw_context. PMUs which are* uncore must use perf_invalid_context.*/if (WARN_ON_ONCE(hw_context_taken &&!(pmu->capabilities & PERF_PMU_CAP_HETEROGENEOUS_CPUS)))pmu->task_ctx_nr = perf_invalid_context;hw_context_taken = 1;}

否则，其将被设置为perf_invalid_context。然后为每个CPU分配上下文：

    pmu->pmu_cpu_context = alloc_percpu(struct perf_cpu_context);

紧接着，进行初始化：

	for_each_possible_cpu(cpu) {    /* 遍历 CPU */struct perf_cpu_context *cpuctx;cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu);    /* 1.获取 CPU 的ctx */__perf_event_init_context(&cpuctx->ctx);            /* 2.初始化这个ctx */lockdep_set_class(&cpuctx->ctx.mutex, &cpuctx_mutex);/*3.初始化lockdep  */lockdep_set_class(&cpuctx->ctx.lock, &cpuctx_lock);cpuctx->ctx.pmu = pmu;                              /* 4.指向这个PMU */cpuctx->online = cpumask_test_cpu(cpu, perf_online_mask);/* 5.是否在线标记 */__perf_mux_hrtimer_init(cpuctx, cpu);               /* 6.高精度定时器，function=perf_mux_hrtimer_handler */cpuctx->heap_size = ARRAY_SIZE(cpuctx->heap_default);/*  */cpuctx->heap = cpuctx->heap_default;    /* 默认使用2个 */}

其中__perf_event_init_context初始化struct perf_event_context结构：

/** Initialize the perf_event context in a task_struct:*/
static void __perf_event_init_context(struct perf_event_context *ctx)   /* 初始化CPU ctx */
{raw_spin_lock_init(&ctx->lock);mutex_init(&ctx->mutex);INIT_LIST_HEAD(&ctx->active_ctx_list);perf_event_groups_init(&ctx->pinned_groups);perf_event_groups_init(&ctx->flexible_groups);INIT_LIST_HEAD(&ctx->event_list);INIT_LIST_HEAD(&ctx->pinned_active);INIT_LIST_HEAD(&ctx->flexible_active);refcount_set(&ctx->refcount, 1);
}

__perf_mux_hrtimer_init初始化一个高精度定时器，

static void __perf_mux_hrtimer_init(struct perf_cpu_context *cpuctx, int cpu)   /* 高精度定时器 */
{struct hrtimer *timer = &cpuctx->hrtimer;struct pmu *pmu = cpuctx->ctx.pmu;u64 interval;/* no multiplexing needed for SW PMU */if (pmu->task_ctx_nr == perf_sw_context)return;/** check default is sane, if not set then force to* default interval (1/tick)*/interval = pmu->hrtimer_interval_ms;if (interval < 1)interval = pmu->hrtimer_interval_ms = PERF_CPU_HRTIMER; /* 小于1ms，就让他是 1ms */cpuctx->hrtimer_interval = ns_to_ktime(NSEC_PER_MSEC * interval);raw_spin_lock_init(&cpuctx->hrtimer_lock);hrtimer_init(timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);timer->function = perf_mux_hrtimer_handler; /* 处理函数 */
}

需要注意一下几点：

• 如果是软件上下文perf_sw_context，不创建定时器；
• 如果ioctl设置的到期时间小于1ms，将其设置为1ms；
• 会调函数为perf_mux_hrtimer_handler；

在获取到CPU上下文后，给没有初始化的PMU函数指针赋值：

    /*  */if (!pmu->start_txn) {if (pmu->pmu_enable) {/** If we have pmu_enable/pmu_disable calls, install* transaction stubs that use that to try and batch* hardware accesses.*/pmu->start_txn  = perf_pmu_start_txn;pmu->commit_txn = perf_pmu_commit_txn;pmu->cancel_txn = perf_pmu_cancel_txn;} else {pmu->start_txn  = perf_pmu_nop_txn;pmu->commit_txn = perf_pmu_nop_int;pmu->cancel_txn = perf_pmu_nop_void;}}/* 使能 */if (!pmu->pmu_enable) {pmu->pmu_enable  = perf_pmu_nop_void;pmu->pmu_disable = perf_pmu_nop_void;}/* 检测周期 ioctl(PERF_EVENT_IOC_PERIOD) */if (!pmu->check_period)pmu->check_period = perf_event_nop_int;/*  */if (!pmu->event_idx)pmu->event_idx = perf_event_idx_default;

下面是将这个PMU添加到pmus链表中：

	/** Ensure the TYPE_SOFTWARE PMUs are at the head of the list,* since these cannot be in the IDR. This way the linear search* is fast, provided a valid software event is provided.*/if (type == PERF_TYPE_SOFTWARE || !name)list_add_rcu(&pmu->entry, &pmus);   /* 软件 或者 name=NULL */elselist_add_tail_rcu(&pmu->entry, &pmus);/*  */

需要注意的是，软件类型的PMU将放到链表开头，以提高线性查询速度。

3. 例: software

//kernel/events/core.c
static struct pmu/* 性能监控单元 */ 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,
};perf_pmu_register(&perf_swevent, "software", PERF_TYPE_SOFTWARE);

3.1. perf_swevent_init

4. 例: perf_cpu_clock

//kernel/events/core.c
static struct pmu perf_cpu_clock = {.task_ctx_nr	= perf_sw_context,.capabilities	= PERF_PMU_CAP_NO_NMI,.event_init	= cpu_clock_event_init,.add		= cpu_clock_event_add,.del		= cpu_clock_event_del,.start		= cpu_clock_event_start,.stop		= cpu_clock_event_stop,.read		= cpu_clock_event_read,
};perf_pmu_register(&perf_cpu_clock, NULL, -1);

5. 例: perf_task_clock

//kernel/events/core.c
static struct pmu perf_task_clock = {.task_ctx_nr	= perf_sw_context,.capabilities	= PERF_PMU_CAP_NO_NMI,.event_init	= task_clock_event_init,.add		= task_clock_event_add,.del		= task_clock_event_del,.start		= task_clock_event_start,.stop		= task_clock_event_stop,.read		= task_clock_event_read,
};perf_pmu_register(&perf_task_clock, NULL, -1);

6. 例: kprobe

//kernel/events/core.c
static struct pmu perf_kprobe = {.task_ctx_nr	= perf_sw_context,.event_init	= perf_kprobe_event_init,.add		= perf_trace_add,.del		= perf_trace_del,.start		= perf_swevent_start,.stop		= perf_swevent_stop,.read		= perf_swevent_read,.attr_groups	= kprobe_attr_groups,
};perf_pmu_register(&perf_kprobe, "kprobe", -1);

7. 例: tracepoint

//kernel/events/core.c
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,
};perf_pmu_register(&perf_tracepoint, "tracepoint", PERF_TYPE_TRACEPOINT);

8. pmu->event_init

perf_event_allocperf_init_eventperf_try_init_eventpmu->event_init(event);

而调用了perf_event_alloc的有：

• perf_event_open
• perf_event_create_kernel_counter
• fork|clone

kernel_clonecopy_processperf_event_init_taskperf_event_init_contextinherit_task_groupinherit_groupinherit_eventperf_event_alloc

9. pmu->add

perf_event_enable_perf_event_enable__perf_event_enablectx_sched_inctx_flexible_sched_in|ctx_pinned_sched_inmerge_sched_ingroup_sched_inevent_sched_inevent->pmu->add(event, PERF_EF_START)

10. pmu->del

perf_event_disable_perf_event_disable__perf_event_disablegroup_sched_outevent_sched_outevent->pmu->del(event, 0);

11. pmu->start

12. pmu->stop

13. pmu->read

14. 相关链接

• 注释源码：https://github.com/Rtoax/linux-5.10.13
• Linux内核 eBPF基础：perf（1）：perf_event在内核中的初始化
• Linux内核 eBPF基础：perf（2）：perf性能管理单元PMU的注册
• Linux kernel perf architecture
• Linux perf 1.1、perf_event内核框架
• Linux内核性能架构：perf_event

这篇关于Linux内核 eBPF基础：perf（2）：perf性能管理单元PMU的注册的文章就介绍到这儿，希望我们推荐的文章对编程师们有所帮助！

---