perf/core: Invert perf_read_group() loops

[pandora-kernel.git] / kernel / events / core.c

/*
 * Performance events core code:
 *
 *  Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
 *  Copyright (C) 2008-2011 Red Hat, Inc., Ingo Molnar
 *  Copyright (C) 2008-2011 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
 *  Copyright  ©  2009 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com>
 *
 * For licensing details see kernel-base/COPYING
 */

#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/cpu.h>
#include <linux/smp.h>
#include <linux/idr.h>
#include <linux/file.h>
#include <linux/poll.h>
#include <linux/slab.h>
#include <linux/hash.h>
#include <linux/sysfs.h>
#include <linux/dcache.h>
#include <linux/percpu.h>
#include <linux/ptrace.h>
#include <linux/reboot.h>
#include <linux/vmstat.h>
#include <linux/device.h>
#include <linux/export.h>
#include <linux/vmalloc.h>
#include <linux/hardirq.h>
#include <linux/rculist.h>
#include <linux/uaccess.h>
#include <linux/syscalls.h>
#include <linux/anon_inodes.h>
#include <linux/kernel_stat.h>
#include <linux/perf_event.h>
#include <linux/ftrace_event.h>
#include <linux/hw_breakpoint.h>
#include <linux/compat.h>

#include "internal.h"

#include <asm/irq_regs.h>

struct remote_function_call {
        struct task_struct      *p;
        int                     (*func)(void *info);
        void                    *info;
        int                     ret;
};

static void remote_function(void *data)
{
        struct remote_function_call *tfc = data;
        struct task_struct *p = tfc->p;

        if (p) {
                tfc->ret = -EAGAIN;
                if (task_cpu(p) != smp_processor_id() || !task_curr(p))
                        return;
        }

        tfc->ret = tfc->func(tfc->info);
}

/**
 * task_function_call - call a function on the cpu on which a task runs
 * @p:          the task to evaluate
 * @func:       the function to be called
 * @info:       the function call argument
 *
 * Calls the function @func when the task is currently running. This might
 * be on the current CPU, which just calls the function directly
 *
 * returns: @func return value, or
 *          -ESRCH  - when the process isn't running
 *          -EAGAIN - when the process moved away
 */
static int
task_function_call(struct task_struct *p, int (*func) (void *info), void *info)
{
        struct remote_function_call data = {
                .p      = p,
                .func   = func,
                .info   = info,
                .ret    = -ESRCH, /* No such (running) process */
        };

        if (task_curr(p))
                smp_call_function_single(task_cpu(p), remote_function, &data, 1);

        return data.ret;
}

/**
 * cpu_function_call - call a function on the cpu
 * @func:       the function to be called
 * @info:       the function call argument
 *
 * Calls the function @func on the remote cpu.
 *
 * returns: @func return value or -ENXIO when the cpu is offline
 */
static int cpu_function_call(int cpu, int (*func) (void *info), void *info)
{
        struct remote_function_call data = {
                .p      = NULL,
                .func   = func,
                .info   = info,
                .ret    = -ENXIO, /* No such CPU */
        };

        smp_call_function_single(cpu, remote_function, &data, 1);

        return data.ret;
}

#define PERF_FLAG_ALL (PERF_FLAG_FD_NO_GROUP |\
                       PERF_FLAG_FD_OUTPUT  |\
                       PERF_FLAG_PID_CGROUP)

enum event_type_t {
        EVENT_FLEXIBLE = 0x1,
        EVENT_PINNED = 0x2,
        EVENT_ALL = EVENT_FLEXIBLE | EVENT_PINNED,
};

/*
 * perf_sched_events : >0 events exist
 * perf_cgroup_events: >0 per-cpu cgroup events exist on this cpu
 */
struct jump_label_key perf_sched_events __read_mostly;
static DEFINE_PER_CPU(atomic_t, perf_cgroup_events);

static atomic_t nr_mmap_events __read_mostly;
static atomic_t nr_comm_events __read_mostly;
static atomic_t nr_task_events __read_mostly;

static LIST_HEAD(pmus);
static DEFINE_MUTEX(pmus_lock);
static struct srcu_struct pmus_srcu;

/*
 * perf event paranoia level:
 *  -1 - not paranoid at all
 *   0 - disallow raw tracepoint access for unpriv
 *   1 - disallow cpu events for unpriv
 *   2 - disallow kernel profiling for unpriv
 */
int sysctl_perf_event_paranoid __read_mostly = 1;

/* Minimum for 512 kiB + 1 user control page */
int sysctl_perf_event_mlock __read_mostly = 512 + (PAGE_SIZE / 1024); /* 'free' kiB per user */

/*
 * max perf event sample rate
 */
#define DEFAULT_MAX_SAMPLE_RATE 100000
int sysctl_perf_event_sample_rate __read_mostly = DEFAULT_MAX_SAMPLE_RATE;
static int max_samples_per_tick __read_mostly =
        DIV_ROUND_UP(DEFAULT_MAX_SAMPLE_RATE, HZ);

int perf_proc_update_handler(struct ctl_table *table, int write,
                void __user *buffer, size_t *lenp,
                loff_t *ppos)
{
        int ret = proc_dointvec(table, write, buffer, lenp, ppos);

        if (ret || !write)
                return ret;

        max_samples_per_tick = DIV_ROUND_UP(sysctl_perf_event_sample_rate, HZ);

        return 0;
}

static atomic64_t perf_event_id;

static void cpu_ctx_sched_out(struct perf_cpu_context *cpuctx,
                              enum event_type_t event_type);

static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx,
                             enum event_type_t event_type,
                             struct task_struct *task);

static void update_context_time(struct perf_event_context *ctx);
static u64 perf_event_time(struct perf_event *event);

void __weak perf_event_print_debug(void)        { }

extern __weak const char *perf_pmu_name(void)
{
        return "pmu";
}

static inline u64 perf_clock(void)
{
        return local_clock();
}

static inline struct perf_cpu_context *
__get_cpu_context(struct perf_event_context *ctx)
{
        return this_cpu_ptr(ctx->pmu->pmu_cpu_context);
}

static void perf_ctx_lock(struct perf_cpu_context *cpuctx,
                          struct perf_event_context *ctx)
{
        raw_spin_lock(&cpuctx->ctx.lock);
        if (ctx)
                raw_spin_lock(&ctx->lock);
}

static void perf_ctx_unlock(struct perf_cpu_context *cpuctx,
                            struct perf_event_context *ctx)
{
        if (ctx)
                raw_spin_unlock(&ctx->lock);
        raw_spin_unlock(&cpuctx->ctx.lock);
}

#ifdef CONFIG_CGROUP_PERF

/*
 * Must ensure cgroup is pinned (css_get) before calling
 * this function. In other words, we cannot call this function
 * if there is no cgroup event for the current CPU context.
 */
static inline struct perf_cgroup *
perf_cgroup_from_task(struct task_struct *task)
{
        return container_of(task_subsys_state(task, perf_subsys_id),
                        struct perf_cgroup, css);
}

static inline bool
perf_cgroup_match(struct perf_event *event)
{
        struct perf_event_context *ctx = event->ctx;
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);

        return !event->cgrp || event->cgrp == cpuctx->cgrp;
}

static inline bool perf_tryget_cgroup(struct perf_event *event)
{
        return css_tryget(&event->cgrp->css);
}

static inline void perf_put_cgroup(struct perf_event *event)
{
        css_put(&event->cgrp->css);
}

static inline void perf_detach_cgroup(struct perf_event *event)
{
        perf_put_cgroup(event);
        event->cgrp = NULL;
}

static inline int is_cgroup_event(struct perf_event *event)
{
        return event->cgrp != NULL;
}

static inline u64 perf_cgroup_event_time(struct perf_event *event)
{
        struct perf_cgroup_info *t;

        t = per_cpu_ptr(event->cgrp->info, event->cpu);
        return t->time;
}

static inline void __update_cgrp_time(struct perf_cgroup *cgrp)
{
        struct perf_cgroup_info *info;
        u64 now;

        now = perf_clock();

        info = this_cpu_ptr(cgrp->info);

        info->time += now - info->timestamp;
        info->timestamp = now;
}

static inline void update_cgrp_time_from_cpuctx(struct perf_cpu_context *cpuctx)
{
        struct perf_cgroup *cgrp_out = cpuctx->cgrp;
        if (cgrp_out)
                __update_cgrp_time(cgrp_out);
}

static inline void update_cgrp_time_from_event(struct perf_event *event)
{
        struct perf_cgroup *cgrp;

        /*
         * ensure we access cgroup data only when needed and
         * when we know the cgroup is pinned (css_get)
         */
        if (!is_cgroup_event(event))
                return;

        cgrp = perf_cgroup_from_task(current);
        /*
         * Do not update time when cgroup is not active
         */
        if (cgrp == event->cgrp)
                __update_cgrp_time(event->cgrp);
}

static inline void
perf_cgroup_set_timestamp(struct task_struct *task,
                          struct perf_event_context *ctx)
{
        struct perf_cgroup *cgrp;
        struct perf_cgroup_info *info;

        /*
         * ctx->lock held by caller
         * ensure we do not access cgroup data
         * unless we have the cgroup pinned (css_get)
         */
        if (!task || !ctx->nr_cgroups)
                return;

        cgrp = perf_cgroup_from_task(task);
        info = this_cpu_ptr(cgrp->info);
        info->timestamp = ctx->timestamp;
}

#define PERF_CGROUP_SWOUT       0x1 /* cgroup switch out every event */
#define PERF_CGROUP_SWIN        0x2 /* cgroup switch in events based on task */

/*
 * reschedule events based on the cgroup constraint of task.
 *
 * mode SWOUT : schedule out everything
 * mode SWIN : schedule in based on cgroup for next
 */
void perf_cgroup_switch(struct task_struct *task, int mode)
{
        struct perf_cpu_context *cpuctx;
        struct pmu *pmu;
        unsigned long flags;

        /*
         * disable interrupts to avoid geting nr_cgroup
         * changes via __perf_event_disable(). Also
         * avoids preemption.
         */
        local_irq_save(flags);

        /*
         * we reschedule only in the presence of cgroup
         * constrained events.
         */
        rcu_read_lock();

        list_for_each_entry_rcu(pmu, &pmus, entry) {
                cpuctx = this_cpu_ptr(pmu->pmu_cpu_context);
                if (cpuctx->unique_pmu != pmu)
                        continue; /* ensure we process each cpuctx once */

                /*
                 * perf_cgroup_events says at least one
                 * context on this CPU has cgroup events.
                 *
                 * ctx->nr_cgroups reports the number of cgroup
                 * events for a context.
                 */
                if (cpuctx->ctx.nr_cgroups > 0) {
                        perf_ctx_lock(cpuctx, cpuctx->task_ctx);
                        perf_pmu_disable(cpuctx->ctx.pmu);

                        if (mode & PERF_CGROUP_SWOUT) {
                                cpu_ctx_sched_out(cpuctx, EVENT_ALL);
                                /*
                                 * must not be done before ctxswout due
                                 * to event_filter_match() in event_sched_out()
                                 */
                                cpuctx->cgrp = NULL;
                        }

                        if (mode & PERF_CGROUP_SWIN) {
                                WARN_ON_ONCE(cpuctx->cgrp);
                                /*
                                 * set cgrp before ctxsw in to allow
                                 * event_filter_match() to not have to pass
                                 * task around
                                 */
                                cpuctx->cgrp = perf_cgroup_from_task(task);
                                cpu_ctx_sched_in(cpuctx, EVENT_ALL, task);
                        }
                        perf_pmu_enable(cpuctx->ctx.pmu);
                        perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
                }
        }

        rcu_read_unlock();

        local_irq_restore(flags);
}

static inline void perf_cgroup_sched_out(struct task_struct *task,
                                         struct task_struct *next)
{
        struct perf_cgroup *cgrp1;
        struct perf_cgroup *cgrp2 = NULL;

        /*
         * we come here when we know perf_cgroup_events > 0
         */
        cgrp1 = perf_cgroup_from_task(task);

        /*
         * next is NULL when called from perf_event_enable_on_exec()
         * that will systematically cause a cgroup_switch()
         */
        if (next)
                cgrp2 = perf_cgroup_from_task(next);

        /*
         * only schedule out current cgroup events if we know
         * that we are switching to a different cgroup. Otherwise,
         * do no touch the cgroup events.
         */
        if (cgrp1 != cgrp2)
                perf_cgroup_switch(task, PERF_CGROUP_SWOUT);
}

static inline void perf_cgroup_sched_in(struct task_struct *prev,
                                        struct task_struct *task)
{
        struct perf_cgroup *cgrp1;
        struct perf_cgroup *cgrp2 = NULL;

        /*
         * we come here when we know perf_cgroup_events > 0
         */
        cgrp1 = perf_cgroup_from_task(task);

        /* prev can never be NULL */
        cgrp2 = perf_cgroup_from_task(prev);

        /*
         * only need to schedule in cgroup events if we are changing
         * cgroup during ctxsw. Cgroup events were not scheduled
         * out of ctxsw out if that was not the case.
         */
        if (cgrp1 != cgrp2)
                perf_cgroup_switch(task, PERF_CGROUP_SWIN);
}

static inline int perf_cgroup_connect(int fd, struct perf_event *event,
                                      struct perf_event_attr *attr,
                                      struct perf_event *group_leader)
{
        struct perf_cgroup *cgrp;
        struct cgroup_subsys_state *css;
        struct file *file;
        int ret = 0, fput_needed;

        file = fget_light(fd, &fput_needed);
        if (!file)
                return -EBADF;

        css = cgroup_css_from_dir(file, perf_subsys_id);
        if (IS_ERR(css)) {
                ret = PTR_ERR(css);
                goto out;
        }

        cgrp = container_of(css, struct perf_cgroup, css);
        event->cgrp = cgrp;

        /* must be done before we fput() the file */
        if (!perf_tryget_cgroup(event)) {
                event->cgrp = NULL;
                ret = -ENOENT;
                goto out;
        }

        /*
         * all events in a group must monitor
         * the same cgroup because a task belongs
         * to only one perf cgroup at a time
         */
        if (group_leader && group_leader->cgrp != cgrp) {
                perf_detach_cgroup(event);
                ret = -EINVAL;
        }
out:
        fput_light(file, fput_needed);
        return ret;
}

static inline void
perf_cgroup_set_shadow_time(struct perf_event *event, u64 now)
{
        struct perf_cgroup_info *t;
        t = per_cpu_ptr(event->cgrp->info, event->cpu);
        event->shadow_ctx_time = now - t->timestamp;
}

static inline void
perf_cgroup_defer_enabled(struct perf_event *event)
{
        /*
         * when the current task's perf cgroup does not match
         * the event's, we need to remember to call the
         * perf_mark_enable() function the first time a task with
         * a matching perf cgroup is scheduled in.
         */
        if (is_cgroup_event(event) && !perf_cgroup_match(event))
                event->cgrp_defer_enabled = 1;
}

static inline void
perf_cgroup_mark_enabled(struct perf_event *event,
                         struct perf_event_context *ctx)
{
        struct perf_event *sub;
        u64 tstamp = perf_event_time(event);

        if (!event->cgrp_defer_enabled)
                return;

        event->cgrp_defer_enabled = 0;

        event->tstamp_enabled = tstamp - event->total_time_enabled;
        list_for_each_entry(sub, &event->sibling_list, group_entry) {
                if (sub->state >= PERF_EVENT_STATE_INACTIVE) {
                        sub->tstamp_enabled = tstamp - sub->total_time_enabled;
                        sub->cgrp_defer_enabled = 0;
                }
        }
}
#else /* !CONFIG_CGROUP_PERF */

static inline bool
perf_cgroup_match(struct perf_event *event)
{
        return true;
}

static inline void perf_detach_cgroup(struct perf_event *event)
{}

static inline int is_cgroup_event(struct perf_event *event)
{
        return 0;
}

static inline u64 perf_cgroup_event_cgrp_time(struct perf_event *event)
{
        return 0;
}

static inline void update_cgrp_time_from_event(struct perf_event *event)
{
}

static inline void update_cgrp_time_from_cpuctx(struct perf_cpu_context *cpuctx)
{
}

static inline void perf_cgroup_sched_out(struct task_struct *task,
                                         struct task_struct *next)
{
}

static inline void perf_cgroup_sched_in(struct task_struct *prev,
                                        struct task_struct *task)
{
}

static inline int perf_cgroup_connect(pid_t pid, struct perf_event *event,
                                      struct perf_event_attr *attr,
                                      struct perf_event *group_leader)
{
        return -EINVAL;
}

static inline void
perf_cgroup_set_timestamp(struct task_struct *task,
                          struct perf_event_context *ctx)
{
}

void
perf_cgroup_switch(struct task_struct *task, struct task_struct *next)
{
}

static inline void
perf_cgroup_set_shadow_time(struct perf_event *event, u64 now)
{
}

static inline u64 perf_cgroup_event_time(struct perf_event *event)
{
        return 0;
}

static inline void
perf_cgroup_defer_enabled(struct perf_event *event)
{
}

static inline void
perf_cgroup_mark_enabled(struct perf_event *event,
                         struct perf_event_context *ctx)
{
}
#endif

void perf_pmu_disable(struct pmu *pmu)
{
        int *count = this_cpu_ptr(pmu->pmu_disable_count);
        if (!(*count)++)
                pmu->pmu_disable(pmu);
}

void perf_pmu_enable(struct pmu *pmu)
{
        int *count = this_cpu_ptr(pmu->pmu_disable_count);
        if (!--(*count))
                pmu->pmu_enable(pmu);
}

static DEFINE_PER_CPU(struct list_head, rotation_list);

/*
 * perf_pmu_rotate_start() and perf_rotate_context() are fully serialized
 * because they're strictly cpu affine and rotate_start is called with IRQs
 * disabled, while rotate_context is called from IRQ context.
 */
static void perf_pmu_rotate_start(struct pmu *pmu)
{
        struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context);
        struct list_head *head = &__get_cpu_var(rotation_list);

        WARN_ON(!irqs_disabled());

        if (list_empty(&cpuctx->rotation_list))
                list_add(&cpuctx->rotation_list, head);
}

static void get_ctx(struct perf_event_context *ctx)
{
        WARN_ON(!atomic_inc_not_zero(&ctx->refcount));
}

static void put_ctx(struct perf_event_context *ctx)
{
        if (atomic_dec_and_test(&ctx->refcount)) {
                if (ctx->parent_ctx)
                        put_ctx(ctx->parent_ctx);
                if (ctx->task)
                        put_task_struct(ctx->task);
                kfree_rcu(ctx, rcu_head);
        }
}

/*
 * Because of perf_event::ctx migration in sys_perf_event_open::move_group we
 * need some magic.
 *
 * Those places that change perf_event::ctx will hold both
 * perf_event_ctx::mutex of the 'old' and 'new' ctx value.
 *
 * Lock ordering is by mutex address. There is one other site where
 * perf_event_context::mutex nests and that is put_event(). But remember that
 * that is a parent<->child context relation, and migration does not affect
 * children, therefore these two orderings should not interact.
 *
 * The change in perf_event::ctx does not affect children (as claimed above)
 * because the sys_perf_event_open() case will install a new event and break
 * the ctx parent<->child relation.
 *
 * The places that change perf_event::ctx will issue:
 *
 *   perf_remove_from_context();
 *   synchronize_rcu();
 *   perf_install_in_context();
 *
 * to affect the change. The remove_from_context() + synchronize_rcu() should
 * quiesce the event, after which we can install it in the new location. This
 * means that only external vectors (perf_fops, prctl) can perturb the event
 * while in transit. Therefore all such accessors should also acquire
 * perf_event_context::mutex to serialize against this.
 *
 * However; because event->ctx can change while we're waiting to acquire
 * ctx->mutex we must be careful and use the below perf_event_ctx_lock()
 * function.
 *
 * Lock order:
 *      task_struct::perf_event_mutex
 *        perf_event_context::mutex
 *          perf_event_context::lock
 *          perf_event::child_mutex;
 *          perf_event::mmap_mutex
 *          mmap_sem
 */
static struct perf_event_context *perf_event_ctx_lock(struct perf_event *event)
{
        struct perf_event_context *ctx;

again:
        rcu_read_lock();
        ctx = ACCESS_ONCE(event->ctx);
        if (!atomic_inc_not_zero(&ctx->refcount)) {
                rcu_read_unlock();
                goto again;
        }
        rcu_read_unlock();

        mutex_lock(&ctx->mutex);
        if (event->ctx != ctx) {
                mutex_unlock(&ctx->mutex);
                put_ctx(ctx);
                goto again;
        }

        return ctx;
}

static void perf_event_ctx_unlock(struct perf_event *event,
                                  struct perf_event_context *ctx)
{
        mutex_unlock(&ctx->mutex);
        put_ctx(ctx);
}

static void unclone_ctx(struct perf_event_context *ctx)
{
        if (ctx->parent_ctx) {
                put_ctx(ctx->parent_ctx);
                ctx->parent_ctx = NULL;
        }
}

static u32 perf_event_pid(struct perf_event *event, struct task_struct *p)
{
        /*
         * only top level events have the pid namespace they were created in
         */
        if (event->parent)
                event = event->parent;

        return task_tgid_nr_ns(p, event->ns);
}

static u32 perf_event_tid(struct perf_event *event, struct task_struct *p)
{
        /*
         * only top level events have the pid namespace they were created in
         */
        if (event->parent)
                event = event->parent;

        return task_pid_nr_ns(p, event->ns);
}

/*
 * If we inherit events we want to return the parent event id
 * to userspace.
 */
static u64 primary_event_id(struct perf_event *event)
{
        u64 id = event->id;

        if (event->parent)
                id = event->parent->id;

        return id;
}

/*
 * Get the perf_event_context for a task and lock it.
 * This has to cope with with the fact that until it is locked,
 * the context could get moved to another task.
 */
static struct perf_event_context *
perf_lock_task_context(struct task_struct *task, int ctxn, unsigned long *flags)
{
        struct perf_event_context *ctx;

retry:
        /*
         * One of the few rules of preemptible RCU is that one cannot do
         * rcu_read_unlock() while holding a scheduler (or nested) lock when
         * part of the read side critical section was preemptible -- see
         * rcu_read_unlock_special().
         *
         * Since ctx->lock nests under rq->lock we must ensure the entire read
         * side critical section is non-preemptible.
         */
        preempt_disable();
        rcu_read_lock();
        ctx = rcu_dereference(task->perf_event_ctxp[ctxn]);
        if (ctx) {
                /*
                 * If this context is a clone of another, it might
                 * get swapped for another underneath us by
                 * perf_event_task_sched_out, though the
                 * rcu_read_lock() protects us from any context
                 * getting freed.  Lock the context and check if it
                 * got swapped before we could get the lock, and retry
                 * if so.  If we locked the right context, then it
                 * can't get swapped on us any more.
                 */
                raw_spin_lock_irqsave(&ctx->lock, *flags);
                if (ctx != rcu_dereference(task->perf_event_ctxp[ctxn])) {
                        raw_spin_unlock_irqrestore(&ctx->lock, *flags);
                        rcu_read_unlock();
                        preempt_enable();
                        goto retry;
                }

                if (!atomic_inc_not_zero(&ctx->refcount)) {
                        raw_spin_unlock_irqrestore(&ctx->lock, *flags);
                        ctx = NULL;
                }
        }
        rcu_read_unlock();
        preempt_enable();
        return ctx;
}

/*
 * Get the context for a task and increment its pin_count so it
 * can't get swapped to another task.  This also increments its
 * reference count so that the context can't get freed.
 */
static struct perf_event_context *
perf_pin_task_context(struct task_struct *task, int ctxn)
{
        struct perf_event_context *ctx;
        unsigned long flags;

        ctx = perf_lock_task_context(task, ctxn, &flags);
        if (ctx) {
                ++ctx->pin_count;
                raw_spin_unlock_irqrestore(&ctx->lock, flags);
        }
        return ctx;
}

static void perf_unpin_context(struct perf_event_context *ctx)
{
        unsigned long flags;

        raw_spin_lock_irqsave(&ctx->lock, flags);
        --ctx->pin_count;
        raw_spin_unlock_irqrestore(&ctx->lock, flags);
}

/*
 * Update the record of the current time in a context.
 */
static void update_context_time(struct perf_event_context *ctx)
{
        u64 now = perf_clock();

        ctx->time += now - ctx->timestamp;
        ctx->timestamp = now;
}

static u64 perf_event_time(struct perf_event *event)
{
        struct perf_event_context *ctx = event->ctx;

        if (is_cgroup_event(event))
                return perf_cgroup_event_time(event);

        return ctx ? ctx->time : 0;
}

/*
 * Update the total_time_enabled and total_time_running fields for a event.
 * The caller of this function needs to hold the ctx->lock.
 */
static void update_event_times(struct perf_event *event)
{
        struct perf_event_context *ctx = event->ctx;
        u64 run_end;

        if (event->state < PERF_EVENT_STATE_INACTIVE ||
            event->group_leader->state < PERF_EVENT_STATE_INACTIVE)
                return;
        /*
         * in cgroup mode, time_enabled represents
         * the time the event was enabled AND active
         * tasks were in the monitored cgroup. This is
         * independent of the activity of the context as
         * there may be a mix of cgroup and non-cgroup events.
         *
         * That is why we treat cgroup events differently
         * here.
         */
        if (is_cgroup_event(event))
                run_end = perf_event_time(event);
        else if (ctx->is_active)
                run_end = ctx->time;
        else
                run_end = event->tstamp_stopped;

        event->total_time_enabled = run_end - event->tstamp_enabled;

        if (event->state == PERF_EVENT_STATE_INACTIVE)
                run_end = event->tstamp_stopped;
        else
                run_end = perf_event_time(event);

        event->total_time_running = run_end - event->tstamp_running;

}

/*
 * Update total_time_enabled and total_time_running for all events in a group.
 */
static void update_group_times(struct perf_event *leader)
{
        struct perf_event *event;

        update_event_times(leader);
        list_for_each_entry(event, &leader->sibling_list, group_entry)
                update_event_times(event);
}

static struct list_head *
ctx_group_list(struct perf_event *event, struct perf_event_context *ctx)
{
        if (event->attr.pinned)
                return &ctx->pinned_groups;
        else
                return &ctx->flexible_groups;
}

/*
 * Add a event from the lists for its context.
 * Must be called with ctx->mutex and ctx->lock held.
 */
static void
list_add_event(struct perf_event *event, struct perf_event_context *ctx)
{
        WARN_ON_ONCE(event->attach_state & PERF_ATTACH_CONTEXT);
        event->attach_state |= PERF_ATTACH_CONTEXT;

        /*
         * If we're a stand alone event or group leader, we go to the context
         * list, group events are kept attached to the group so that
         * perf_group_detach can, at all times, locate all siblings.
         */
        if (event->group_leader == event) {
                struct list_head *list;

                if (is_software_event(event))
                        event->group_flags |= PERF_GROUP_SOFTWARE;

                list = ctx_group_list(event, ctx);
                list_add_tail(&event->group_entry, list);
        }

        if (is_cgroup_event(event))
                ctx->nr_cgroups++;

        list_add_rcu(&event->event_entry, &ctx->event_list);
        if (!ctx->nr_events)
                perf_pmu_rotate_start(ctx->pmu);
        ctx->nr_events++;
        if (event->attr.inherit_stat)
                ctx->nr_stat++;
}

/*
 * Initialize event state based on the perf_event_attr::disabled.
 */
static inline void perf_event__state_init(struct perf_event *event)
{
        event->state = event->attr.disabled ? PERF_EVENT_STATE_OFF :
                                              PERF_EVENT_STATE_INACTIVE;
}

/*
 * Called at perf_event creation and when events are attached/detached from a
 * group.
 */
static void perf_event__read_size(struct perf_event *event)
{
        int entry = sizeof(u64); /* value */
        int size = 0;
        int nr = 1;

        if (event->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
                size += sizeof(u64);

        if (event->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
                size += sizeof(u64);

        if (event->attr.read_format & PERF_FORMAT_ID)
                entry += sizeof(u64);

        if (event->attr.read_format & PERF_FORMAT_GROUP) {
                nr += event->group_leader->nr_siblings;
                size += sizeof(u64);
        }

        size += entry * nr;
        event->read_size = size;
}

static void perf_event__header_size(struct perf_event *event)
{
        struct perf_sample_data *data;
        u64 sample_type = event->attr.sample_type;
        u16 size = 0;

        perf_event__read_size(event);

        if (sample_type & PERF_SAMPLE_IP)
                size += sizeof(data->ip);

        if (sample_type & PERF_SAMPLE_ADDR)
                size += sizeof(data->addr);

        if (sample_type & PERF_SAMPLE_PERIOD)
                size += sizeof(data->period);

        if (sample_type & PERF_SAMPLE_READ)
                size += event->read_size;

        event->header_size = size;
}

static void perf_event__id_header_size(struct perf_event *event)
{
        struct perf_sample_data *data;
        u64 sample_type = event->attr.sample_type;
        u16 size = 0;

        if (sample_type & PERF_SAMPLE_TID)
                size += sizeof(data->tid_entry);

        if (sample_type & PERF_SAMPLE_TIME)
                size += sizeof(data->time);

        if (sample_type & PERF_SAMPLE_ID)
                size += sizeof(data->id);

        if (sample_type & PERF_SAMPLE_STREAM_ID)
                size += sizeof(data->stream_id);

        if (sample_type & PERF_SAMPLE_CPU)
                size += sizeof(data->cpu_entry);

        event->id_header_size = size;
}

static void perf_group_attach(struct perf_event *event)
{
        struct perf_event *group_leader = event->group_leader, *pos;

        /*
         * We can have double attach due to group movement in perf_event_open.
         */
        if (event->attach_state & PERF_ATTACH_GROUP)
                return;

        event->attach_state |= PERF_ATTACH_GROUP;

        if (group_leader == event)
                return;

        if (group_leader->group_flags & PERF_GROUP_SOFTWARE &&
                        !is_software_event(event))
                group_leader->group_flags &= ~PERF_GROUP_SOFTWARE;

        list_add_tail(&event->group_entry, &group_leader->sibling_list);
        group_leader->nr_siblings++;

        perf_event__header_size(group_leader);

        list_for_each_entry(pos, &group_leader->sibling_list, group_entry)
                perf_event__header_size(pos);
}

/*
 * Remove a event from the lists for its context.
 * Must be called with ctx->mutex and ctx->lock held.
 */
static void
list_del_event(struct perf_event *event, struct perf_event_context *ctx)
{
        struct perf_cpu_context *cpuctx;
        /*
         * We can have double detach due to exit/hot-unplug + close.
         */
        if (!(event->attach_state & PERF_ATTACH_CONTEXT))
                return;

        event->attach_state &= ~PERF_ATTACH_CONTEXT;

        if (is_cgroup_event(event)) {
                ctx->nr_cgroups--;
                cpuctx = __get_cpu_context(ctx);
                /*
                 * if there are no more cgroup events
                 * then cler cgrp to avoid stale pointer
                 * in update_cgrp_time_from_cpuctx()
                 */
                if (!ctx->nr_cgroups)
                        cpuctx->cgrp = NULL;
        }

        ctx->nr_events--;
        if (event->attr.inherit_stat)
                ctx->nr_stat--;

        list_del_rcu(&event->event_entry);

        if (event->group_leader == event)
                list_del_init(&event->group_entry);

        update_group_times(event);

        /*
         * If event was in error state, then keep it
         * that way, otherwise bogus counts will be
         * returned on read(). The only way to get out
         * of error state is by explicit re-enabling
         * of the event
         */
        if (event->state > PERF_EVENT_STATE_OFF)
                event->state = PERF_EVENT_STATE_OFF;
}

static void perf_group_detach(struct perf_event *event)
{
        struct perf_event *sibling, *tmp;
        struct list_head *list = NULL;

        /*
         * We can have double detach due to exit/hot-unplug + close.
         */
        if (!(event->attach_state & PERF_ATTACH_GROUP))
                return;

        event->attach_state &= ~PERF_ATTACH_GROUP;

        /*
         * If this is a sibling, remove it from its group.
         */
        if (event->group_leader != event) {
                list_del_init(&event->group_entry);
                event->group_leader->nr_siblings--;
                goto out;
        }

        if (!list_empty(&event->group_entry))
                list = &event->group_entry;

        /*
         * If this was a group event with sibling events then
         * upgrade the siblings to singleton events by adding them
         * to whatever list we are on.
         */
        list_for_each_entry_safe(sibling, tmp, &event->sibling_list, group_entry) {
                if (list)
                        list_move_tail(&sibling->group_entry, list);
                sibling->group_leader = sibling;

                /* Inherit group flags from the previous leader */
                sibling->group_flags = event->group_flags;
        }

out:
        perf_event__header_size(event->group_leader);

        list_for_each_entry(tmp, &event->group_leader->sibling_list, group_entry)
                perf_event__header_size(tmp);
}

static inline int
event_filter_match(struct perf_event *event)
{
        return (event->cpu == -1 || event->cpu == smp_processor_id())
            && perf_cgroup_match(event);
}

static void
event_sched_out(struct perf_event *event,
                  struct perf_cpu_context *cpuctx,
                  struct perf_event_context *ctx)
{
        u64 tstamp = perf_event_time(event);
        u64 delta;
        /*
         * An event which could not be activated because of
         * filter mismatch still needs to have its timings
         * maintained, otherwise bogus information is return
         * via read() for time_enabled, time_running:
         */
        if (event->state == PERF_EVENT_STATE_INACTIVE
            && !event_filter_match(event)) {
                delta = tstamp - event->tstamp_stopped;
                event->tstamp_running += delta;
                event->tstamp_stopped = tstamp;
        }

        if (event->state != PERF_EVENT_STATE_ACTIVE)
                return;

        event->state = PERF_EVENT_STATE_INACTIVE;
        if (event->pending_disable) {
                event->pending_disable = 0;
                event->state = PERF_EVENT_STATE_OFF;
        }
        event->tstamp_stopped = tstamp;
        event->pmu->del(event, 0);
        event->oncpu = -1;

        if (!is_software_event(event))
                cpuctx->active_oncpu--;
        ctx->nr_active--;
        if (event->attr.exclusive || !cpuctx->active_oncpu)
                cpuctx->exclusive = 0;
}

static void
group_sched_out(struct perf_event *group_event,
                struct perf_cpu_context *cpuctx,
                struct perf_event_context *ctx)
{
        struct perf_event *event;
        int state = group_event->state;

        event_sched_out(group_event, cpuctx, ctx);

        /*
         * Schedule out siblings (if any):
         */
        list_for_each_entry(event, &group_event->sibling_list, group_entry)
                event_sched_out(event, cpuctx, ctx);

        if (state == PERF_EVENT_STATE_ACTIVE && group_event->attr.exclusive)
                cpuctx->exclusive = 0;
}

struct remove_event {
        struct perf_event *event;
        bool detach_group;
};

/*
 * Cross CPU call to remove a performance event
 *
 * We disable the event on the hardware level first. After that we
 * remove it from the context list.
 */
static int __perf_remove_from_context(void *info)
{
        struct remove_event *re = info;
        struct perf_event *event = re->event;
        struct perf_event_context *ctx = event->ctx;
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);

        raw_spin_lock(&ctx->lock);
        event_sched_out(event, cpuctx, ctx);
        if (re->detach_group)
                perf_group_detach(event);
        list_del_event(event, ctx);
        if (!ctx->nr_events && cpuctx->task_ctx == ctx) {
                ctx->is_active = 0;
                cpuctx->task_ctx = NULL;
        }
        raw_spin_unlock(&ctx->lock);

        return 0;
}


/*
 * Remove the event from a task's (or a CPU's) list of events.
 *
 * CPU events are removed with a smp call. For task events we only
 * call when the task is on a CPU.
 *
 * If event->ctx is a cloned context, callers must make sure that
 * every task struct that event->ctx->task could possibly point to
 * remains valid.  This is OK when called from perf_release since
 * that only calls us on the top-level context, which can't be a clone.
 * When called from perf_event_exit_task, it's OK because the
 * context has been detached from its task.
 */
static void perf_remove_from_context(struct perf_event *event, bool detach_group)
{
        struct perf_event_context *ctx = event->ctx;
        struct task_struct *task = ctx->task;
        struct remove_event re = {
                .event = event,
                .detach_group = detach_group,
        };

        lockdep_assert_held(&ctx->mutex);

        if (!task) {
                /*
                 * Per cpu events are removed via an smp call and
                 * the removal is always successful.
                 */
                cpu_function_call(event->cpu, __perf_remove_from_context, &re);
                return;
        }

retry:
        if (!task_function_call(task, __perf_remove_from_context, &re))
                return;

        raw_spin_lock_irq(&ctx->lock);
        /*
         * If we failed to find a running task, but find the context active now
         * that we've acquired the ctx->lock, retry.
         */
        if (ctx->is_active) {
                raw_spin_unlock_irq(&ctx->lock);
                goto retry;
        }

        /*
         * Since the task isn't running, its safe to remove the event, us
         * holding the ctx->lock ensures the task won't get scheduled in.
         */
        if (detach_group)
                perf_group_detach(event);
        list_del_event(event, ctx);
        raw_spin_unlock_irq(&ctx->lock);
}

/*
 * Cross CPU call to disable a performance event
 */
static int __perf_event_disable(void *info)
{
        struct perf_event *event = info;
        struct perf_event_context *ctx = event->ctx;
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);

        /*
         * If this is a per-task event, need to check whether this
         * event's task is the current task on this cpu.
         *
         * Can trigger due to concurrent perf_event_context_sched_out()
         * flipping contexts around.
         */
        if (ctx->task && cpuctx->task_ctx != ctx)
                return -EINVAL;

        raw_spin_lock(&ctx->lock);

        /*
         * If the event is on, turn it off.
         * If it is in error state, leave it in error state.
         */
        if (event->state >= PERF_EVENT_STATE_INACTIVE) {
                update_context_time(ctx);
                update_cgrp_time_from_event(event);
                update_group_times(event);
                if (event == event->group_leader)
                        group_sched_out(event, cpuctx, ctx);
                else
                        event_sched_out(event, cpuctx, ctx);
                event->state = PERF_EVENT_STATE_OFF;
        }

        raw_spin_unlock(&ctx->lock);

        return 0;
}

/*
 * Disable a event.
 *
 * If event->ctx is a cloned context, callers must make sure that
 * every task struct that event->ctx->task could possibly point to
 * remains valid.  This condition is satisifed when called through
 * perf_event_for_each_child or perf_event_for_each because they
 * hold the top-level event's child_mutex, so any descendant that
 * goes to exit will block in sync_child_event.
 * When called from perf_pending_event it's OK because event->ctx
 * is the current context on this CPU and preemption is disabled,
 * hence we can't get into perf_event_task_sched_out for this context.
 */
static void _perf_event_disable(struct perf_event *event)
{
        struct perf_event_context *ctx = event->ctx;
        struct task_struct *task = ctx->task;

        if (!task) {
                /*
                 * Disable the event on the cpu that it's on
                 */
                cpu_function_call(event->cpu, __perf_event_disable, event);
                return;
        }

retry:
        if (!task_function_call(task, __perf_event_disable, event))
                return;

        raw_spin_lock_irq(&ctx->lock);
        /*
         * If the event is still active, we need to retry the cross-call.
         */
        if (event->state == PERF_EVENT_STATE_ACTIVE) {
                raw_spin_unlock_irq(&ctx->lock);
                /*
                 * Reload the task pointer, it might have been changed by
                 * a concurrent perf_event_context_sched_out().
                 */
                task = ctx->task;
                goto retry;
        }

        /*
         * Since we have the lock this context can't be scheduled
         * in, so we can change the state safely.
         */
        if (event->state == PERF_EVENT_STATE_INACTIVE) {
                update_group_times(event);
                event->state = PERF_EVENT_STATE_OFF;
        }
        raw_spin_unlock_irq(&ctx->lock);
}

/*
 * Strictly speaking kernel users cannot create groups and therefore this
 * interface does not need the perf_event_ctx_lock() magic.
 */
void perf_event_disable(struct perf_event *event)
{
        struct perf_event_context *ctx;

        ctx = perf_event_ctx_lock(event);
        _perf_event_disable(event);
        perf_event_ctx_unlock(event, ctx);
}

static void perf_set_shadow_time(struct perf_event *event,
                                 struct perf_event_context *ctx,
                                 u64 tstamp)
{
        /*
         * use the correct time source for the time snapshot
         *
         * We could get by without this by leveraging the
         * fact that to get to this function, the caller
         * has most likely already called update_context_time()
         * and update_cgrp_time_xx() and thus both timestamp
         * are identical (or very close). Given that tstamp is,
         * already adjusted for cgroup, we could say that:
         *    tstamp - ctx->timestamp
         * is equivalent to
         *    tstamp - cgrp->timestamp.
         *
         * Then, in perf_output_read(), the calculation would
         * work with no changes because:
         * - event is guaranteed scheduled in
         * - no scheduled out in between
         * - thus the timestamp would be the same
         *
         * But this is a bit hairy.
         *
         * So instead, we have an explicit cgroup call to remain
         * within the time time source all along. We believe it
         * is cleaner and simpler to understand.
         */
        if (is_cgroup_event(event))
                perf_cgroup_set_shadow_time(event, tstamp);
        else
                event->shadow_ctx_time = tstamp - ctx->timestamp;
}

#define MAX_INTERRUPTS (~0ULL)

static void perf_log_throttle(struct perf_event *event, int enable);

static int
event_sched_in(struct perf_event *event,
                 struct perf_cpu_context *cpuctx,
                 struct perf_event_context *ctx)
{
        u64 tstamp = perf_event_time(event);

        if (event->state <= PERF_EVENT_STATE_OFF)
                return 0;

        event->state = PERF_EVENT_STATE_ACTIVE;
        event->oncpu = smp_processor_id();

        /*
         * Unthrottle events, since we scheduled we might have missed several
         * ticks already, also for a heavily scheduling task there is little
         * guarantee it'll get a tick in a timely manner.
         */
        if (unlikely(event->hw.interrupts == MAX_INTERRUPTS)) {
                perf_log_throttle(event, 1);
                event->hw.interrupts = 0;
        }

        /*
         * The new state must be visible before we turn it on in the hardware:
         */
        smp_wmb();

        if (event->pmu->add(event, PERF_EF_START)) {
                event->state = PERF_EVENT_STATE_INACTIVE;
                event->oncpu = -1;
                return -EAGAIN;
        }

        event->tstamp_running += tstamp - event->tstamp_stopped;

        perf_set_shadow_time(event, ctx, tstamp);

        if (!is_software_event(event))
                cpuctx->active_oncpu++;
        ctx->nr_active++;

        if (event->attr.exclusive)
                cpuctx->exclusive = 1;

        return 0;
}

static int
group_sched_in(struct perf_event *group_event,
               struct perf_cpu_context *cpuctx,
               struct perf_event_context *ctx)
{
        struct perf_event *event, *partial_group = NULL;
        struct pmu *pmu = group_event->pmu;
        u64 now = ctx->time;
        bool simulate = false;

        if (group_event->state == PERF_EVENT_STATE_OFF)
                return 0;

        pmu->start_txn(pmu);

        if (event_sched_in(group_event, cpuctx, ctx)) {
                pmu->cancel_txn(pmu);
                return -EAGAIN;
        }

        /*
         * Schedule in siblings as one group (if any):
         */
        list_for_each_entry(event, &group_event->sibling_list, group_entry) {
                if (event_sched_in(event, cpuctx, ctx)) {
                        partial_group = event;
                        goto group_error;
                }
        }

        if (!pmu->commit_txn(pmu))
                return 0;

group_error:
        /*
         * Groups can be scheduled in as one unit only, so undo any
         * partial group before returning:
         * The events up to the failed event are scheduled out normally,
         * tstamp_stopped will be updated.
         *
         * The failed events and the remaining siblings need to have
         * their timings updated as if they had gone thru event_sched_in()
         * and event_sched_out(). This is required to get consistent timings
         * across the group. This also takes care of the case where the group
         * could never be scheduled by ensuring tstamp_stopped is set to mark
         * the time the event was actually stopped, such that time delta
         * calculation in update_event_times() is correct.
         */
        list_for_each_entry(event, &group_event->sibling_list, group_entry) {
                if (event == partial_group)
                        simulate = true;

                if (simulate) {
                        event->tstamp_running += now - event->tstamp_stopped;
                        event->tstamp_stopped = now;
                } else {
                        event_sched_out(event, cpuctx, ctx);
                }
        }
        event_sched_out(group_event, cpuctx, ctx);

        pmu->cancel_txn(pmu);

        return -EAGAIN;
}

/*
 * Work out whether we can put this event group on the CPU now.
 */
static int group_can_go_on(struct perf_event *event,
                           struct perf_cpu_context *cpuctx,
                           int can_add_hw)
{
        /*
         * Groups consisting entirely of software events can always go on.
         */
        if (event->group_flags & PERF_GROUP_SOFTWARE)
                return 1;
        /*
         * If an exclusive group is already on, no other hardware
         * events can go on.
         */
        if (cpuctx->exclusive)
                return 0;
        /*
         * If this group is exclusive and there are already
         * events on the CPU, it can't go on.
         */
        if (event->attr.exclusive && cpuctx->active_oncpu)
                return 0;
        /*
         * Otherwise, try to add it if all previous groups were able
         * to go on.
         */
        return can_add_hw;
}

static void add_event_to_ctx(struct perf_event *event,
                               struct perf_event_context *ctx)
{
        u64 tstamp = perf_event_time(event);

        list_add_event(event, ctx);
        perf_group_attach(event);
        event->tstamp_enabled = tstamp;
        event->tstamp_running = tstamp;
        event->tstamp_stopped = tstamp;
}

static void task_ctx_sched_out(struct perf_event_context *ctx);
static void
ctx_sched_in(struct perf_event_context *ctx,
             struct perf_cpu_context *cpuctx,
             enum event_type_t event_type,
             struct task_struct *task);

static void perf_event_sched_in(struct perf_cpu_context *cpuctx,
                                struct perf_event_context *ctx,
                                struct task_struct *task)
{
        cpu_ctx_sched_in(cpuctx, EVENT_PINNED, task);
        if (ctx)
                ctx_sched_in(ctx, cpuctx, EVENT_PINNED, task);
        cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE, task);
        if (ctx)
                ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE, task);
}

/*
 * Cross CPU call to install and enable a performance event
 *
 * Must be called with ctx->mutex held
 */
static int  __perf_install_in_context(void *info)
{
        struct perf_event *event = info;
        struct perf_event_context *ctx = event->ctx;
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);
        struct perf_event_context *task_ctx = cpuctx->task_ctx;
        struct task_struct *task = current;

        perf_ctx_lock(cpuctx, task_ctx);
        perf_pmu_disable(cpuctx->ctx.pmu);

        /*
         * If there was an active task_ctx schedule it out.
         */
        if (task_ctx)
                task_ctx_sched_out(task_ctx);

        /*
         * If the context we're installing events in is not the
         * active task_ctx, flip them.
         */
        if (ctx->task && task_ctx != ctx) {
                if (task_ctx)
                        raw_spin_unlock(&task_ctx->lock);
                raw_spin_lock(&ctx->lock);
                task_ctx = ctx;
        }

        if (task_ctx) {
                cpuctx->task_ctx = task_ctx;
                task = task_ctx->task;
        }

        cpu_ctx_sched_out(cpuctx, EVENT_ALL);

        update_context_time(ctx);
        /*
         * update cgrp time only if current cgrp
         * matches event->cgrp. Must be done before
         * calling add_event_to_ctx()
         */
        update_cgrp_time_from_event(event);

        add_event_to_ctx(event, ctx);

        /*
         * Schedule everything back in
         */
        perf_event_sched_in(cpuctx, task_ctx, task);

        perf_pmu_enable(cpuctx->ctx.pmu);
        perf_ctx_unlock(cpuctx, task_ctx);

        return 0;
}

/*
 * Attach a performance event to a context
 *
 * First we add the event to the list with the hardware enable bit
 * in event->hw_config cleared.
 *
 * If the event is attached to a task which is on a CPU we use a smp
 * call to enable it in the task context. The task might have been
 * scheduled away, but we check this in the smp call again.
 */
static void
perf_install_in_context(struct perf_event_context *ctx,
                        struct perf_event *event,
                        int cpu)
{
        struct task_struct *task = ctx->task;

        lockdep_assert_held(&ctx->mutex);

        event->ctx = ctx;

        if (!task) {
                /*
                 * Per cpu events are installed via an smp call and
                 * the install is always successful.
                 */
                cpu_function_call(cpu, __perf_install_in_context, event);
                return;
        }

retry:
        if (!task_function_call(task, __perf_install_in_context, event))
                return;

        raw_spin_lock_irq(&ctx->lock);
        /*
         * If we failed to find a running task, but find the context active now
         * that we've acquired the ctx->lock, retry.
         */
        if (ctx->is_active) {
                raw_spin_unlock_irq(&ctx->lock);
                /*
                 * Reload the task pointer, it might have been changed by
                 * a concurrent perf_event_context_sched_out().
                 */
                task = ctx->task;
                /*
                 * Reload the task pointer, it might have been changed by
                 * a concurrent perf_event_context_sched_out().
                 */
                task = ctx->task;
                goto retry;
        }

        /*
         * Since the task isn't running, its safe to add the event, us holding
         * the ctx->lock ensures the task won't get scheduled in.
         */
        add_event_to_ctx(event, ctx);
        raw_spin_unlock_irq(&ctx->lock);
}

/*
 * Put a event into inactive state and update time fields.
 * Enabling the leader of a group effectively enables all
 * the group members that aren't explicitly disabled, so we
 * have to update their ->tstamp_enabled also.
 * Note: this works for group members as well as group leaders
 * since the non-leader members' sibling_lists will be empty.
 */
static void __perf_event_mark_enabled(struct perf_event *event,
                                        struct perf_event_context *ctx)
{
        struct perf_event *sub;
        u64 tstamp = perf_event_time(event);

        event->state = PERF_EVENT_STATE_INACTIVE;
        event->tstamp_enabled = tstamp - event->total_time_enabled;
        list_for_each_entry(sub, &event->sibling_list, group_entry) {
                if (sub->state >= PERF_EVENT_STATE_INACTIVE)
                        sub->tstamp_enabled = tstamp - sub->total_time_enabled;
        }
}

/*
 * Cross CPU call to enable a performance event
 */
static int __perf_event_enable(void *info)
{
        struct perf_event *event = info;
        struct perf_event_context *ctx = event->ctx;
        struct perf_event *leader = event->group_leader;
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);
        int err;

        /*
         * There's a time window between 'ctx->is_active' check
         * in perf_event_enable function and this place having:
         *   - IRQs on
         *   - ctx->lock unlocked
         *
         * where the task could be killed and 'ctx' deactivated
         * by perf_event_exit_task.
         */
        if (!ctx->is_active)
                return -EINVAL;

        raw_spin_lock(&ctx->lock);
        update_context_time(ctx);

        if (event->state >= PERF_EVENT_STATE_INACTIVE)
                goto unlock;

        /*
         * set current task's cgroup time reference point
         */
        perf_cgroup_set_timestamp(current, ctx);

        __perf_event_mark_enabled(event, ctx);

        if (!event_filter_match(event)) {
                if (is_cgroup_event(event))
                        perf_cgroup_defer_enabled(event);
                goto unlock;
        }

        /*
         * If the event is in a group and isn't the group leader,
         * then don't put it on unless the group is on.
         */
        if (leader != event && leader->state != PERF_EVENT_STATE_ACTIVE)
                goto unlock;

        if (!group_can_go_on(event, cpuctx, 1)) {
                err = -EEXIST;
        } else {
                if (event == leader)
                        err = group_sched_in(event, cpuctx, ctx);
                else
                        err = event_sched_in(event, cpuctx, ctx);
        }

        if (err) {
                /*
                 * If this event can't go on and it's part of a
                 * group, then the whole group has to come off.
                 */
                if (leader != event)
                        group_sched_out(leader, cpuctx, ctx);
                if (leader->attr.pinned) {
                        update_group_times(leader);
                        leader->state = PERF_EVENT_STATE_ERROR;
                }
        }

unlock:
        raw_spin_unlock(&ctx->lock);

        return 0;
}

/*
 * Enable a event.
 *
 * If event->ctx is a cloned context, callers must make sure that
 * every task struct that event->ctx->task could possibly point to
 * remains valid.  This condition is satisfied when called through
 * perf_event_for_each_child or perf_event_for_each as described
 * for perf_event_disable.
 */
static void _perf_event_enable(struct perf_event *event)
{
        struct perf_event_context *ctx = event->ctx;
        struct task_struct *task = ctx->task;

        if (!task) {
                /*
                 * Enable the event on the cpu that it's on
                 */
                cpu_function_call(event->cpu, __perf_event_enable, event);
                return;
        }

        raw_spin_lock_irq(&ctx->lock);
        if (event->state >= PERF_EVENT_STATE_INACTIVE)
                goto out;

        /*
         * If the event is in error state, clear that first.
         * That way, if we see the event in error state below, we
         * know that it has gone back into error state, as distinct
         * from the task having been scheduled away before the
         * cross-call arrived.
         */
        if (event->state == PERF_EVENT_STATE_ERROR)
                event->state = PERF_EVENT_STATE_OFF;

retry:
        if (!ctx->is_active) {
                __perf_event_mark_enabled(event, ctx);
                goto out;
        }

        raw_spin_unlock_irq(&ctx->lock);

        if (!task_function_call(task, __perf_event_enable, event))
                return;

        raw_spin_lock_irq(&ctx->lock);

        /*
         * If the context is active and the event is still off,
         * we need to retry the cross-call.
         */
        if (ctx->is_active && event->state == PERF_EVENT_STATE_OFF) {
                /*
                 * task could have been flipped by a concurrent
                 * perf_event_context_sched_out()
                 */
                task = ctx->task;
                goto retry;
        }

out:
        raw_spin_unlock_irq(&ctx->lock);
}

/*
 * See perf_event_disable();
 */
void perf_event_enable(struct perf_event *event)
{
        struct perf_event_context *ctx;

        ctx = perf_event_ctx_lock(event);
        _perf_event_enable(event);
        perf_event_ctx_unlock(event, ctx);
}

static int _perf_event_refresh(struct perf_event *event, int refresh)
{
        /*
         * not supported on inherited events
         */
        if (event->attr.inherit || !is_sampling_event(event))
                return -EINVAL;

        atomic_add(refresh, &event->event_limit);
        _perf_event_enable(event);

        return 0;
}

/*
 * See perf_event_disable()
 */
int perf_event_refresh(struct perf_event *event, int refresh)
{
        struct perf_event_context *ctx;
        int ret;

        ctx = perf_event_ctx_lock(event);
        ret = _perf_event_refresh(event, refresh);
        perf_event_ctx_unlock(event, ctx);

        return ret;
}
EXPORT_SYMBOL_GPL(perf_event_refresh);

static void ctx_sched_out(struct perf_event_context *ctx,
                          struct perf_cpu_context *cpuctx,
                          enum event_type_t event_type)
{
        struct perf_event *event;
        int is_active = ctx->is_active;

        ctx->is_active &= ~event_type;
        if (likely(!ctx->nr_events))
                return;

        update_context_time(ctx);
        update_cgrp_time_from_cpuctx(cpuctx);
        if (!ctx->nr_active)
                return;

        perf_pmu_disable(ctx->pmu);
        if ((is_active & EVENT_PINNED) && (event_type & EVENT_PINNED)) {
                list_for_each_entry(event, &ctx->pinned_groups, group_entry)
                        group_sched_out(event, cpuctx, ctx);
        }

        if ((is_active & EVENT_FLEXIBLE) && (event_type & EVENT_FLEXIBLE)) {
                list_for_each_entry(event, &ctx->flexible_groups, group_entry)
                        group_sched_out(event, cpuctx, ctx);
        }
        perf_pmu_enable(ctx->pmu);
}

/*
 * Test whether two contexts are equivalent, i.e. whether they
 * have both been cloned from the same version of the same context
 * and they both have the same number of enabled events.
 * If the number of enabled events is the same, then the set
 * of enabled events should be the same, because these are both
 * inherited contexts, therefore we can't access individual events
 * in them directly with an fd; we can only enable/disable all
 * events via prctl, or enable/disable all events in a family
 * via ioctl, which will have the same effect on both contexts.
 */
static int context_equiv(struct perf_event_context *ctx1,
                         struct perf_event_context *ctx2)
{
        return ctx1->parent_ctx && ctx1->parent_ctx == ctx2->parent_ctx
                && ctx1->parent_gen == ctx2->parent_gen
                && !ctx1->pin_count && !ctx2->pin_count;
}

static void __perf_event_sync_stat(struct perf_event *event,
                                     struct perf_event *next_event)
{
        u64 value;

        if (!event->attr.inherit_stat)
                return;

        /*
         * Update the event value, we cannot use perf_event_read()
         * because we're in the middle of a context switch and have IRQs
         * disabled, which upsets smp_call_function_single(), however
         * we know the event must be on the current CPU, therefore we
         * don't need to use it.
         */
        switch (event->state) {
        case PERF_EVENT_STATE_ACTIVE:
                event->pmu->read(event);
                /* fall-through */

        case PERF_EVENT_STATE_INACTIVE:
                update_event_times(event);
                break;

        default:
                break;
        }

        /*
         * In order to keep per-task stats reliable we need to flip the event
         * values when we flip the contexts.
         */
        value = local64_read(&next_event->count);
        value = local64_xchg(&event->count, value);
        local64_set(&next_event->count, value);

        swap(event->total_time_enabled, next_event->total_time_enabled);
        swap(event->total_time_running, next_event->total_time_running);

        /*
         * Since we swizzled the values, update the user visible data too.
         */
        perf_event_update_userpage(event);
        perf_event_update_userpage(next_event);
}

#define list_next_entry(pos, member) \
        list_entry(pos->member.next, typeof(*pos), member)

static void perf_event_sync_stat(struct perf_event_context *ctx,
                                   struct perf_event_context *next_ctx)
{
        struct perf_event *event, *next_event;

        if (!ctx->nr_stat)
                return;

        update_context_time(ctx);

        event = list_first_entry(&ctx->event_list,
                                   struct perf_event, event_entry);

        next_event = list_first_entry(&next_ctx->event_list,
                                        struct perf_event, event_entry);

        while (&event->event_entry != &ctx->event_list &&
               &next_event->event_entry != &next_ctx->event_list) {

                __perf_event_sync_stat(event, next_event);

                event = list_next_entry(event, event_entry);
                next_event = list_next_entry(next_event, event_entry);
        }
}

static void perf_event_context_sched_out(struct task_struct *task, int ctxn,
                                         struct task_struct *next)
{
        struct perf_event_context *ctx = task->perf_event_ctxp[ctxn];
        struct perf_event_context *next_ctx;
        struct perf_event_context *parent;
        struct perf_cpu_context *cpuctx;
        int do_switch = 1;

        if (likely(!ctx))
                return;

        cpuctx = __get_cpu_context(ctx);
        if (!cpuctx->task_ctx)
                return;

        rcu_read_lock();
        parent = rcu_dereference(ctx->parent_ctx);
        next_ctx = next->perf_event_ctxp[ctxn];
        if (parent && next_ctx &&
            rcu_dereference(next_ctx->parent_ctx) == parent) {
                /*
                 * Looks like the two contexts are clones, so we might be
                 * able to optimize the context switch.  We lock both
                 * contexts and check that they are clones under the
                 * lock (including re-checking that neither has been
                 * uncloned in the meantime).  It doesn't matter which
                 * order we take the locks because no other cpu could
                 * be trying to lock both of these tasks.
                 */
                raw_spin_lock(&ctx->lock);
                raw_spin_lock_nested(&next_ctx->lock, SINGLE_DEPTH_NESTING);
                if (context_equiv(ctx, next_ctx)) {
                        /*
                         * XXX do we need a memory barrier of sorts
                         * wrt to rcu_dereference() of perf_event_ctxp
                         */
                        task->perf_event_ctxp[ctxn] = next_ctx;
                        next->perf_event_ctxp[ctxn] = ctx;
                        ctx->task = next;
                        next_ctx->task = task;
                        do_switch = 0;

                        perf_event_sync_stat(ctx, next_ctx);
                }
                raw_spin_unlock(&next_ctx->lock);
                raw_spin_unlock(&ctx->lock);
        }
        rcu_read_unlock();

        if (do_switch) {
                raw_spin_lock(&ctx->lock);
                ctx_sched_out(ctx, cpuctx, EVENT_ALL);
                cpuctx->task_ctx = NULL;
                raw_spin_unlock(&ctx->lock);
        }
}

#define for_each_task_context_nr(ctxn)                                  \
        for ((ctxn) = 0; (ctxn) < perf_nr_task_contexts; (ctxn)++)

/*
 * Called from scheduler to remove the events of the current task,
 * with interrupts disabled.
 *
 * We stop each event and update the event value in event->count.
 *
 * This does not protect us against NMI, but disable()
 * sets the disabled bit in the control field of event _before_
 * accessing the event control register. If a NMI hits, then it will
 * not restart the event.
 */
void __perf_event_task_sched_out(struct task_struct *task,
                                 struct task_struct *next)
{
        int ctxn;

        for_each_task_context_nr(ctxn)
                perf_event_context_sched_out(task, ctxn, next);

        /*
         * if cgroup events exist on this CPU, then we need
         * to check if we have to switch out PMU state.
         * cgroup event are system-wide mode only
         */
        if (atomic_read(&__get_cpu_var(perf_cgroup_events)))
                perf_cgroup_sched_out(task, next);
}

static void task_ctx_sched_out(struct perf_event_context *ctx)
{
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);

        if (!cpuctx->task_ctx)
                return;

        if (WARN_ON_ONCE(ctx != cpuctx->task_ctx))
                return;

        ctx_sched_out(ctx, cpuctx, EVENT_ALL);
        cpuctx->task_ctx = NULL;
}

/*
 * Called with IRQs disabled
 */
static void cpu_ctx_sched_out(struct perf_cpu_context *cpuctx,
                              enum event_type_t event_type)
{
        ctx_sched_out(&cpuctx->ctx, cpuctx, event_type);
}

static void
ctx_pinned_sched_in(struct perf_event_context *ctx,
                    struct perf_cpu_context *cpuctx)
{
        struct perf_event *event;

        list_for_each_entry(event, &ctx->pinned_groups, group_entry) {
                if (event->state <= PERF_EVENT_STATE_OFF)
                        continue;
                if (!event_filter_match(event))
                        continue;

                /* may need to reset tstamp_enabled */
                if (is_cgroup_event(event))
                        perf_cgroup_mark_enabled(event, ctx);

                if (group_can_go_on(event, cpuctx, 1))
                        group_sched_in(event, cpuctx, ctx);

                /*
                 * If this pinned group hasn't been scheduled,
                 * put it in error state.
                 */
                if (event->state == PERF_EVENT_STATE_INACTIVE) {
                        update_group_times(event);
                        event->state = PERF_EVENT_STATE_ERROR;
                }
        }
}

static void
ctx_flexible_sched_in(struct perf_event_context *ctx,
                      struct perf_cpu_context *cpuctx)
{
        struct perf_event *event;
        int can_add_hw = 1;

        list_for_each_entry(event, &ctx->flexible_groups, group_entry) {
                /* Ignore events in OFF or ERROR state */
                if (event->state <= PERF_EVENT_STATE_OFF)
                        continue;
                /*
                 * Listen to the 'cpu' scheduling filter constraint
                 * of events:
                 */
                if (!event_filter_match(event))
                        continue;

                /* may need to reset tstamp_enabled */
                if (is_cgroup_event(event))
                        perf_cgroup_mark_enabled(event, ctx);

                if (group_can_go_on(event, cpuctx, can_add_hw)) {
                        if (group_sched_in(event, cpuctx, ctx))
                                can_add_hw = 0;
                }
        }
}

static void
ctx_sched_in(struct perf_event_context *ctx,
             struct perf_cpu_context *cpuctx,
             enum event_type_t event_type,
             struct task_struct *task)
{
        u64 now;
        int is_active = ctx->is_active;

        ctx->is_active |= event_type;
        if (likely(!ctx->nr_events))
                return;

        now = perf_clock();
        ctx->timestamp = now;
        perf_cgroup_set_timestamp(task, ctx);
        /*
         * First go through the list and put on any pinned groups
         * in order to give them the best chance of going on.
         */
        if (!(is_active & EVENT_PINNED) && (event_type & EVENT_PINNED))
                ctx_pinned_sched_in(ctx, cpuctx);

        /* Then walk through the lower prio flexible groups */
        if (!(is_active & EVENT_FLEXIBLE) && (event_type & EVENT_FLEXIBLE))
                ctx_flexible_sched_in(ctx, cpuctx);
}

static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx,
                             enum event_type_t event_type,
                             struct task_struct *task)
{
        struct perf_event_context *ctx = &cpuctx->ctx;

        ctx_sched_in(ctx, cpuctx, event_type, task);
}

static void perf_event_context_sched_in(struct perf_event_context *ctx,
                                        struct task_struct *task)
{
        struct perf_cpu_context *cpuctx;

        cpuctx = __get_cpu_context(ctx);
        if (cpuctx->task_ctx == ctx)
                return;

        perf_ctx_lock(cpuctx, ctx);
        perf_pmu_disable(ctx->pmu);
        /*
         * We want to keep the following priority order:
         * cpu pinned (that don't need to move), task pinned,
         * cpu flexible, task flexible.
         */
        cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE);

        if (ctx->nr_events)
                cpuctx->task_ctx = ctx;

        perf_event_sched_in(cpuctx, cpuctx->task_ctx, task);

        perf_pmu_enable(ctx->pmu);
        perf_ctx_unlock(cpuctx, ctx);

        /*
         * Since these rotations are per-cpu, we need to ensure the
         * cpu-context we got scheduled on is actually rotating.
         */
        perf_pmu_rotate_start(ctx->pmu);
}

/*
 * Called from scheduler to add the events of the current task
 * with interrupts disabled.
 *
 * We restore the event value and then enable it.
 *
 * This does not protect us against NMI, but enable()
 * sets the enabled bit in the control field of event _before_
 * accessing the event control register. If a NMI hits, then it will
 * keep the event running.
 */
void __perf_event_task_sched_in(struct task_struct *prev,
                                struct task_struct *task)
{
        struct perf_event_context *ctx;
        int ctxn;

        for_each_task_context_nr(ctxn) {
                ctx = task->perf_event_ctxp[ctxn];
                if (likely(!ctx))
                        continue;

                perf_event_context_sched_in(ctx, task);
        }
        /*
         * if cgroup events exist on this CPU, then we need
         * to check if we have to switch in PMU state.
         * cgroup event are system-wide mode only
         */
        if (atomic_read(&__get_cpu_var(perf_cgroup_events)))
                perf_cgroup_sched_in(prev, task);
}

static u64 perf_calculate_period(struct perf_event *event, u64 nsec, u64 count)
{
        u64 frequency = event->attr.sample_freq;
        u64 sec = NSEC_PER_SEC;
        u64 divisor, dividend;

        int count_fls, nsec_fls, frequency_fls, sec_fls;

        count_fls = fls64(count);
        nsec_fls = fls64(nsec);
        frequency_fls = fls64(frequency);
        sec_fls = 30;

        /*
         * We got @count in @nsec, with a target of sample_freq HZ
         * the target period becomes:
         *
         *             @count * 10^9
         * period = -------------------
         *          @nsec * sample_freq
         *
         */

        /*
         * Reduce accuracy by one bit such that @a and @b converge
         * to a similar magnitude.
         */
#define REDUCE_FLS(a, b)                \
do {                                    \
        if (a##_fls > b##_fls) {        \
                a >>= 1;                \
                a##_fls--;              \
        } else {                        \
                b >>= 1;                \
                b##_fls--;              \
        }                               \
} while (0)

        /*
         * Reduce accuracy until either term fits in a u64, then proceed with
         * the other, so that finally we can do a u64/u64 division.
         */
        while (count_fls + sec_fls > 64 && nsec_fls + frequency_fls > 64) {
                REDUCE_FLS(nsec, frequency);
                REDUCE_FLS(sec, count);
        }

        if (count_fls + sec_fls > 64) {
                divisor = nsec * frequency;

                while (count_fls + sec_fls > 64) {
                        REDUCE_FLS(count, sec);
                        divisor >>= 1;
                }

                dividend = count * sec;
        } else {
                dividend = count * sec;

                while (nsec_fls + frequency_fls > 64) {
                        REDUCE_FLS(nsec, frequency);
                        dividend >>= 1;
                }

                divisor = nsec * frequency;
        }

        if (!divisor)
                return dividend;

        return div64_u64(dividend, divisor);
}

static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count)
{
        struct hw_perf_event *hwc = &event->hw;
        s64 period, sample_period;
        s64 delta;

        period = perf_calculate_period(event, nsec, count);

        delta = (s64)(period - hwc->sample_period);
        delta = (delta + 7) / 8; /* low pass filter */

        sample_period = hwc->sample_period + delta;

        if (!sample_period)
                sample_period = 1;

        hwc->sample_period = sample_period;

        if (local64_read(&hwc->period_left) > 8*sample_period) {
                event->pmu->stop(event, PERF_EF_UPDATE);
                local64_set(&hwc->period_left, 0);
                event->pmu->start(event, PERF_EF_RELOAD);
        }
}

static void perf_ctx_adjust_freq(struct perf_event_context *ctx, u64 period)
{
        struct perf_event *event;
        struct hw_perf_event *hwc;
        u64 interrupts, now;
        s64 delta;

        list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
                if (event->state != PERF_EVENT_STATE_ACTIVE)
                        continue;

                if (!event_filter_match(event))
                        continue;

                hwc = &event->hw;

                interrupts = hwc->interrupts;
                hwc->interrupts = 0;

                /*
                 * unthrottle events on the tick
                 */
                if (interrupts == MAX_INTERRUPTS) {
                        perf_log_throttle(event, 1);
                        event->pmu->start(event, 0);
                }

                if (!event->attr.freq || !event->attr.sample_freq)
                        continue;

                event->pmu->read(event);
                now = local64_read(&event->count);
                delta = now - hwc->freq_count_stamp;
                hwc->freq_count_stamp = now;

                if (delta > 0)
                        perf_adjust_period(event, period, delta);
        }
}

/*
 * Round-robin a context's events:
 */
static void rotate_ctx(struct perf_event_context *ctx)
{
        /*
         * Rotate the first entry last of non-pinned groups. Rotation might be
         * disabled by the inheritance code.
         */
        if (!ctx->rotate_disable)
                list_rotate_left(&ctx->flexible_groups);
}

/*
 * perf_pmu_rotate_start() and perf_rotate_context() are fully serialized
 * because they're strictly cpu affine and rotate_start is called with IRQs
 * disabled, while rotate_context is called from IRQ context.
 */
static void perf_rotate_context(struct perf_cpu_context *cpuctx)
{
        u64 interval = (u64)cpuctx->jiffies_interval * TICK_NSEC;
        struct perf_event_context *ctx = NULL;
        int rotate = 0, remove = 1;

        if (cpuctx->ctx.nr_events) {
                remove = 0;
                if (cpuctx->ctx.nr_events != cpuctx->ctx.nr_active)
                        rotate = 1;
        }

        ctx = cpuctx->task_ctx;
        if (ctx && ctx->nr_events) {
                remove = 0;
                if (ctx->nr_events != ctx->nr_active)
                        rotate = 1;
        }

        perf_ctx_lock(cpuctx, cpuctx->task_ctx);
        perf_pmu_disable(cpuctx->ctx.pmu);
        perf_ctx_adjust_freq(&cpuctx->ctx, interval);
        if (ctx)
                perf_ctx_adjust_freq(ctx, interval);

        if (!rotate)
                goto done;

        cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE);
        if (ctx)
                ctx_sched_out(ctx, cpuctx, EVENT_FLEXIBLE);

        rotate_ctx(&cpuctx->ctx);
        if (ctx)
                rotate_ctx(ctx);

        perf_event_sched_in(cpuctx, ctx, current);

done:
        if (remove)
                list_del_init(&cpuctx->rotation_list);

        perf_pmu_enable(cpuctx->ctx.pmu);
        perf_ctx_unlock(cpuctx, cpuctx->task_ctx);
}

void perf_event_task_tick(void)
{
        struct list_head *head = &__get_cpu_var(rotation_list);
        struct perf_cpu_context *cpuctx, *tmp;

        WARN_ON(!irqs_disabled());

        list_for_each_entry_safe(cpuctx, tmp, head, rotation_list) {
                if (cpuctx->jiffies_interval == 1 ||
                                !(jiffies % cpuctx->jiffies_interval))
                        perf_rotate_context(cpuctx);
        }
}

static int event_enable_on_exec(struct perf_event *event,
                                struct perf_event_context *ctx)
{
        if (!event->attr.enable_on_exec)
                return 0;

        event->attr.enable_on_exec = 0;
        if (event->state >= PERF_EVENT_STATE_INACTIVE)
                return 0;

        __perf_event_mark_enabled(event, ctx);

        return 1;
}

/*
 * Enable all of a task's events that have been marked enable-on-exec.
 * This expects task == current.
 */
static void perf_event_enable_on_exec(struct perf_event_context *ctx)
{
        struct perf_event *event;
        unsigned long flags;
        int enabled = 0;
        int ret;

        local_irq_save(flags);
        if (!ctx || !ctx->nr_events)
                goto out;

        /*
         * We must ctxsw out cgroup events to avoid conflict
         * when invoking perf_task_event_sched_in() later on
         * in this function. Otherwise we end up trying to
         * ctxswin cgroup events which are already scheduled
         * in.
         */
        perf_cgroup_sched_out(current, NULL);

        raw_spin_lock(&ctx->lock);
        task_ctx_sched_out(ctx);

        list_for_each_entry(event, &ctx->pinned_groups, group_entry) {
                ret = event_enable_on_exec(event, ctx);
                if (ret)
                        enabled = 1;
        }

        list_for_each_entry(event, &ctx->flexible_groups, group_entry) {
                ret = event_enable_on_exec(event, ctx);
                if (ret)
                        enabled = 1;
        }

        /*
         * Unclone this context if we enabled any event.
         */
        if (enabled)
                unclone_ctx(ctx);

        raw_spin_unlock(&ctx->lock);

        /*
         * Also calls ctxswin for cgroup events, if any:
         */
        perf_event_context_sched_in(ctx, ctx->task);
out:
        local_irq_restore(flags);
}

/*
 * Cross CPU call to read the hardware event
 */
static void __perf_event_read(void *info)
{
        struct perf_event *event = info;
        struct perf_event_context *ctx = event->ctx;
        struct perf_cpu_context *cpuctx = __get_cpu_context(ctx);

        /*
         * If this is a task context, we need to check whether it is
         * the current task context of this cpu.  If not it has been
         * scheduled out before the smp call arrived.  In that case
         * event->count would have been updated to a recent sample
         * when the event was scheduled out.
         */
        if (ctx->task && cpuctx->task_ctx != ctx)
                return;

        raw_spin_lock(&ctx->lock);
        if (ctx->is_active) {
                update_context_time(ctx);
                update_cgrp_time_from_event(event);
        }
        update_event_times(event);
        if (event->state == PERF_EVENT_STATE_ACTIVE)
                event->pmu->read(event);
        raw_spin_unlock(&ctx->lock);
}

static inline u64 perf_event_count(struct perf_event *event)
{
        return local64_read(&event->count) + atomic64_read(&event->child_count);
}

static u64 perf_event_read(struct perf_event *event)
{
        /*
         * If event is enabled and currently active on a CPU, update the
         * value in the event structure:
         */
        if (event->state == PERF_EVENT_STATE_ACTIVE) {
                smp_call_function_single(event->oncpu,
                                         __perf_event_read, event, 1);
        } else if (event->state == PERF_EVENT_STATE_INACTIVE) {
                struct perf_event_context *ctx = event->ctx;
                unsigned long flags;

                raw_spin_lock_irqsave(&ctx->lock, flags);
                /*
                 * may read while context is not active
                 * (e.g., thread is blocked), in that case
                 * we cannot update context time
                 */
                if (ctx->is_active) {
                        update_context_time(ctx);
                        update_cgrp_time_from_event(event);
                }
                update_event_times(event);
                raw_spin_unlock_irqrestore(&ctx->lock, flags);
        }

        return perf_event_count(event);
}

/*
 * Callchain support
 */

struct callchain_cpus_entries {
        struct rcu_head                 rcu_head;
        struct perf_callchain_entry     *cpu_entries[0];
};

static DEFINE_PER_CPU(int, callchain_recursion[PERF_NR_CONTEXTS]);
static atomic_t nr_callchain_events;
static DEFINE_MUTEX(callchain_mutex);
struct callchain_cpus_entries *callchain_cpus_entries;


__weak void perf_callchain_kernel(struct perf_callchain_entry *entry,
                                  struct pt_regs *regs)
{
}

__weak void perf_callchain_user(struct perf_callchain_entry *entry,
                                struct pt_regs *regs)
{
}

static void release_callchain_buffers_rcu(struct rcu_head *head)
{
        struct callchain_cpus_entries *entries;
        int cpu;

        entries = container_of(head, struct callchain_cpus_entries, rcu_head);

        for_each_possible_cpu(cpu)
                kfree(entries->cpu_entries[cpu]);

        kfree(entries);
}

static void release_callchain_buffers(void)
{
        struct callchain_cpus_entries *entries;

        entries = callchain_cpus_entries;
        rcu_assign_pointer(callchain_cpus_entries, NULL);
        call_rcu(&entries->rcu_head, release_callchain_buffers_rcu);
}

static int alloc_callchain_buffers(void)
{
        int cpu;
        int size;
        struct callchain_cpus_entries *entries;

        /*
         * We can't use the percpu allocation API for data that can be
         * accessed from NMI. Use a temporary manual per cpu allocation
         * until that gets sorted out.
         */
        size = offsetof(struct callchain_cpus_entries, cpu_entries[nr_cpu_ids]);

        entries = kzalloc(size, GFP_KERNEL);
        if (!entries)
                return -ENOMEM;

        size = sizeof(struct perf_callchain_entry) * PERF_NR_CONTEXTS;

        for_each_possible_cpu(cpu) {
                entries->cpu_entries[cpu] = kmalloc_node(size, GFP_KERNEL,
                                                         cpu_to_node(cpu));
                if (!entries->cpu_entries[cpu])
                        goto fail;
        }

        rcu_assign_pointer(callchain_cpus_entries, entries);

        return 0;

fail:
        for_each_possible_cpu(cpu)
                kfree(entries->cpu_entries[cpu]);
        kfree(entries);

        return -ENOMEM;
}

static int get_callchain_buffers(void)
{
        int err = 0;
        int count;

        mutex_lock(&callchain_mutex);

        count = atomic_inc_return(&nr_callchain_events);
        if (WARN_ON_ONCE(count < 1)) {
                err = -EINVAL;
                goto exit;
        }

        if (count > 1) {
                /* If the allocation failed, give up */
                if (!callchain_cpus_entries)
                        err = -ENOMEM;
                goto exit;
        }

        err = alloc_callchain_buffers();
        if (err)
                release_callchain_buffers();
exit:
        mutex_unlock(&callchain_mutex);

        return err;
}

static void put_callchain_buffers(void)
{
        if (atomic_dec_and_mutex_lock(&nr_callchain_events, &callchain_mutex)) {
                release_callchain_buffers();
                mutex_unlock(&callchain_mutex);
        }
}

static int get_recursion_context(int *recursion)
{
        int rctx;

        if (in_nmi())
                rctx = 3;
        else if (in_irq())
                rctx = 2;
        else if (in_softirq())
                rctx = 1;
        else
                rctx = 0;

        if (recursion[rctx])
                return -1;

        recursion[rctx]++;
        barrier();

        return rctx;
}

static inline void put_recursion_context(int *recursion, int rctx)
{
        barrier();
        recursion[rctx]--;
}

static struct perf_callchain_entry *get_callchain_entry(int *rctx)
{
        int cpu;
        struct callchain_cpus_entries *entries;

        *rctx = get_recursion_context(__get_cpu_var(callchain_recursion));
        if (*rctx == -1)
                return NULL;

        entries = rcu_dereference(callchain_cpus_entries);
        if (!entries)
                return NULL;

        cpu = smp_processor_id();

        return &entries->cpu_entries[cpu][*rctx];
}

static void
put_callchain_entry(int rctx)
{
        put_recursion_context(__get_cpu_var(callchain_recursion), rctx);
}

static struct perf_callchain_entry *perf_callchain(struct pt_regs *regs)
{
        int rctx;
        struct perf_callchain_entry *entry;


        entry = get_callchain_entry(&rctx);
        if (rctx == -1)
                return NULL;

        if (!entry)
                goto exit_put;

        entry->nr = 0;

        if (!user_mode(regs)) {
                perf_callchain_store(entry, PERF_CONTEXT_KERNEL);
                perf_callchain_kernel(entry, regs);
                if (current->mm)
                        regs = task_pt_regs(current);
                else
                        regs = NULL;
        }

        if (regs) {
                perf_callchain_store(entry, PERF_CONTEXT_USER);
                perf_callchain_user(entry, regs);
        }

exit_put:
        put_callchain_entry(rctx);

        return entry;
}

/*
 * Initialize the perf_event context in a task_struct:
 */
static void __perf_event_init_context(struct perf_event_context *ctx)
{
        raw_spin_lock_init(&ctx->lock);
        mutex_init(&ctx->mutex);
        INIT_LIST_HEAD(&ctx->pinned_groups);
        INIT_LIST_HEAD(&ctx->flexible_groups);
        INIT_LIST_HEAD(&ctx->event_list);
        atomic_set(&ctx->refcount, 1);
}

static struct perf_event_context *
alloc_perf_context(struct pmu *pmu, struct task_struct *task)
{
        struct perf_event_context *ctx;

        ctx = kzalloc(sizeof(struct perf_event_context), GFP_KERNEL);
        if (!ctx)
                return NULL;

        __perf_event_init_context(ctx);
        if (task) {
                ctx->task = task;
                get_task_struct(task);
        }
        ctx->pmu = pmu;

        return ctx;
}

static struct task_struct *
find_lively_task_by_vpid(pid_t vpid)
{
        struct task_struct *task;
        int err;

        rcu_read_lock();
        if (!vpid)
                task = current;
        else
                task = find_task_by_vpid(vpid);
        if (task)
                get_task_struct(task);
        rcu_read_unlock();

        if (!task)
                return ERR_PTR(-ESRCH);

        /* Reuse ptrace permission checks for now. */
        err = -EACCES;
        if (!ptrace_may_access(task, PTRACE_MODE_READ_REALCREDS))
                goto errout;

        return task;
errout:
        put_task_struct(task);
        return ERR_PTR(err);

}

/*
 * Returns a matching context with refcount and pincount.
 */
static struct perf_event_context *
find_get_context(struct pmu *pmu, struct task_struct *task, int cpu)
{
        struct perf_event_context *ctx;
        struct perf_cpu_context *cpuctx;
        unsigned long flags;
        int ctxn, err;

        if (!task) {
                /* Must be root to operate on a CPU event: */
                if (perf_paranoid_cpu() && !capable(CAP_SYS_ADMIN))
                        return ERR_PTR(-EACCES);

                /*
                 * We could be clever and allow to attach a event to an
                 * offline CPU and activate it when the CPU comes up, but
                 * that's for later.
                 */
                if (!cpu_online(cpu))
                        return ERR_PTR(-ENODEV);

                cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu);
                ctx = &cpuctx->ctx;
                get_ctx(ctx);
                ++ctx->pin_count;

                return ctx;
        }

        err = -EINVAL;
        ctxn = pmu->task_ctx_nr;
        if (ctxn < 0)
                goto errout;

retry:
        ctx = perf_lock_task_context(task, ctxn, &flags);
        if (ctx) {
                unclone_ctx(ctx);
                ++ctx->pin_count;
                raw_spin_unlock_irqrestore(&ctx->lock, flags);
        } else {
                ctx = alloc_perf_context(pmu, task);
                err = -ENOMEM;
                if (!ctx)
                        goto errout;

                err = 0;
                mutex_lock(&task->perf_event_mutex);
                /*
                 * If it has already passed perf_event_exit_task().
                 * we must see PF_EXITING, it takes this mutex too.
                 */
                if (task->flags & PF_EXITING)
                        err = -ESRCH;
                else if (task->perf_event_ctxp[ctxn])
                        err = -EAGAIN;
                else {
                        get_ctx(ctx);
                        ++ctx->pin_count;
                        rcu_assign_pointer(task->perf_event_ctxp[ctxn], ctx);
                }
                mutex_unlock(&task->perf_event_mutex);

                if (unlikely(err)) {
                        put_ctx(ctx);

                        if (err == -EAGAIN)
                                goto retry;
                        goto errout;
                }
        }

        return ctx;

errout:
        return ERR_PTR(err);
}

static void perf_event_free_filter(struct perf_event *event);

static void free_event_rcu(struct rcu_head *head)
{
        struct perf_event *event;

        event = container_of(head, struct perf_event, rcu_head);
        if (event->ns)
                put_pid_ns(event->ns);
        perf_event_free_filter(event);
        kfree(event);
}

static void ring_buffer_put(struct ring_buffer *rb);
static void ring_buffer_detach(struct perf_event *event, struct ring_buffer *rb);

static void free_event(struct perf_event *event)
{
        irq_work_sync(&event->pending);

        if (!event->parent) {
                if (event->attach_state & PERF_ATTACH_TASK)
                        jump_label_dec(&perf_sched_events);
                if (event->attr.mmap || event->attr.mmap_data)
                        atomic_dec(&nr_mmap_events);
                if (event->attr.comm)
                        atomic_dec(&nr_comm_events);
                if (event->attr.task)
                        atomic_dec(&nr_task_events);
                if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN)
                        put_callchain_buffers();
                if (is_cgroup_event(event)) {
                        atomic_dec(&per_cpu(perf_cgroup_events, event->cpu));
                        jump_label_dec(&perf_sched_events);
                }
        }

        if (event->rb) {
                struct ring_buffer *rb;

                /*
                 * Can happen when we close an event with re-directed output.
                 *
                 * Since we have a 0 refcount, perf_mmap_close() will skip
                 * over us; possibly making our ring_buffer_put() the last.
                 */
                mutex_lock(&event->mmap_mutex);
                rb = event->rb;
                if (rb) {
                        rcu_assign_pointer(event->rb, NULL);
                        ring_buffer_detach(event, rb);
                        ring_buffer_put(rb); /* could be last */
                }
                mutex_unlock(&event->mmap_mutex);
        }

        if (is_cgroup_event(event))
                perf_detach_cgroup(event);

        if (event->destroy)
                event->destroy(event);

        if (event->ctx)
                put_ctx(event->ctx);

        call_rcu(&event->rcu_head, free_event_rcu);
}

int perf_event_release_kernel(struct perf_event *event)
{
        struct perf_event_context *ctx = event->ctx;

        WARN_ON_ONCE(ctx->parent_ctx);
        /*
         * There are two ways this annotation is useful:
         *
         *  1) there is a lock recursion from perf_event_exit_task
         *     see the comment there.
         *
         *  2) there is a lock-inversion with mmap_sem through
         *     perf_event_read_group(), which takes faults while
         *     holding ctx->mutex, however this is called after
         *     the last filedesc died, so there is no possibility
         *     to trigger the AB-BA case.
         */
        mutex_lock_nested(&ctx->mutex, SINGLE_DEPTH_NESTING);
        perf_remove_from_context(event, true);
        mutex_unlock(&ctx->mutex);

        free_event(event);

        return 0;
}
EXPORT_SYMBOL_GPL(perf_event_release_kernel);

/*
 * Called when the last reference to the file is gone.
 */
static void put_event(struct perf_event *event)
{
        struct task_struct *owner;

        if (!atomic_long_dec_and_test(&event->refcount))
                return;

        rcu_read_lock();
        owner = ACCESS_ONCE(event->owner);
        /*
         * Matches the smp_wmb() in perf_event_exit_task(). If we observe
         * !owner it means the list deletion is complete and we can indeed
         * free this event, otherwise we need to serialize on
         * owner->perf_event_mutex.
         */
        smp_read_barrier_depends();
        if (owner) {
                /*
                 * Since delayed_put_task_struct() also drops the last
                 * task reference we can safely take a new reference
                 * while holding the rcu_read_lock().
                 */
                get_task_struct(owner);
        }
        rcu_read_unlock();

        if (owner) {
                /*
                 * If we're here through perf_event_exit_task() we're already
                 * holding ctx->mutex which would be an inversion wrt. the
                 * normal lock order.
                 *
                 * However we can safely take this lock because its the child
                 * ctx->mutex.
                 */
                mutex_lock_nested(&owner->perf_event_mutex, SINGLE_DEPTH_NESTING);

                /*
                 * We have to re-check the event->owner field, if it is cleared
                 * we raced with perf_event_exit_task(), acquiring the mutex
                 * ensured they're done, and we can proceed with freeing the
                 * event.
                 */
                if (event->owner)
                        list_del_init(&event->owner_entry);
                mutex_unlock(&owner->perf_event_mutex);
                put_task_struct(owner);
        }

        perf_event_release_kernel(event);
}

static int perf_release(struct inode *inode, struct file *file)
{
        put_event(file->private_data);
        return 0;
}

u64 perf_event_read_value(struct perf_event *event, u64 *enabled, u64 *running)
{
        struct perf_event *child;
        u64 total = 0;

        *enabled = 0;
        *running = 0;

        mutex_lock(&event->child_mutex);
        total += perf_event_read(event);
        *enabled += event->total_time_enabled +
                        atomic64_read(&event->child_total_time_enabled);
        *running += event->total_time_running +
                        atomic64_read(&event->child_total_time_running);

        list_for_each_entry(child, &event->child_list, child_list) {
                total += perf_event_read(child);
                *enabled += child->total_time_enabled;
                *running += child->total_time_running;
        }
        mutex_unlock(&event->child_mutex);

        return total;
}
EXPORT_SYMBOL_GPL(perf_event_read_value);

static void __perf_read_group_add(struct perf_event *leader,
                                        u64 read_format, u64 *values)
{
        struct perf_event *sub;
        int n = 1; /* skip @nr */
        u64 count, enabled, running;

        count = perf_event_read_value(leader, &enabled, &running);

        /*
         * Since we co-schedule groups, {enabled,running} times of siblings
         * will be identical to those of the leader, so we only publish one
         * set.
         */
        if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
                values[n++] = enabled;
        if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
                values[n++] = running;

        /*
         * Write {count,id} tuples for every sibling.
         */
        values[n++] += count;
        if (read_format & PERF_FORMAT_ID)
                values[n++] = primary_event_id(leader);

        list_for_each_entry(sub, &leader->sibling_list, group_entry) {
                values[n++] = perf_event_read_value(sub, &enabled, &running);
                if (read_format & PERF_FORMAT_ID)
                        values[n++] = primary_event_id(sub);
        }
}

static int perf_event_read_group(struct perf_event *event,
                                   u64 read_format, char __user *buf)
{
        struct perf_event *leader = event->group_leader, *child;
        struct perf_event_context *ctx = leader->ctx;
        int ret = event->read_size;
        u64 *values;

        lockdep_assert_held(&ctx->mutex);

        values = kzalloc(event->read_size, GFP_KERNEL);
        if (!values)
                return -ENOMEM;

        values[0] = 1 + leader->nr_siblings;

        /*
         * By locking the child_mutex of the leader we effectively
         * lock the child list of all siblings.. XXX explain how.
         */
        mutex_lock(&leader->child_mutex);

        __perf_read_group_add(leader, read_format, values);
        list_for_each_entry(child, &leader->child_list, child_list)
                __perf_read_group_add(child, read_format, values);

        mutex_unlock(&leader->child_mutex);

        if (copy_to_user(buf, values, event->read_size))
                ret = -EFAULT;

        kfree(values);

        return ret;
}

static int perf_event_read_one(struct perf_event *event,
                                 u64 read_format, char __user *buf)
{
        u64 enabled, running;
        u64 values[4];
        int n = 0;

        values[n++] = perf_event_read_value(event, &enabled, &running);
        if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
                values[n++] = enabled;
        if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
                values[n++] = running;
        if (read_format & PERF_FORMAT_ID)
                values[n++] = primary_event_id(event);

        if (copy_to_user(buf, values, n * sizeof(u64)))
                return -EFAULT;

        return n * sizeof(u64);
}

/*
 * Read the performance event - simple non blocking version for now
 */
static ssize_t
perf_read_hw(struct perf_event *event, char __user *buf, size_t count)
{
        u64 read_format = event->attr.read_format;
        int ret;

        /*
         * Return end-of-file for a read on a event that is in
         * error state (i.e. because it was pinned but it couldn't be
         * scheduled on to the CPU at some point).
         */
        if (event->state == PERF_EVENT_STATE_ERROR)
                return 0;

        if (count < event->read_size)
                return -ENOSPC;

        WARN_ON_ONCE(event->ctx->parent_ctx);
        if (read_format & PERF_FORMAT_GROUP)
                ret = perf_event_read_group(event, read_format, buf);
        else
                ret = perf_event_read_one(event, read_format, buf);

        return ret;
}

static ssize_t
perf_read(struct file *file, char __user *buf, size_t count, loff_t *ppos)
{
        struct perf_event *event = file->private_data;
        struct perf_event_context *ctx;
        int ret;

        ctx = perf_event_ctx_lock(event);
        ret = perf_read_hw(event, buf, count);
        perf_event_ctx_unlock(event, ctx);

        return ret;
}

static unsigned int perf_poll(struct file *file, poll_table *wait)
{
        struct perf_event *event = file->private_data;
        struct ring_buffer *rb;
        unsigned int events = POLL_HUP;

        /*
         * Pin the event->rb by taking event->mmap_mutex; otherwise
         * perf_event_set_output() can swizzle our rb and make us miss wakeups.
         */
        mutex_lock(&event->mmap_mutex);
        rb = event->rb;
        if (rb)
                events = atomic_xchg(&rb->poll, 0);
        mutex_unlock(&event->mmap_mutex);

        poll_wait(file, &event->waitq, wait);

        return events;
}

static void _perf_event_reset(struct perf_event *event)
{
        (void)perf_event_read(event);
        local64_set(&event->count, 0);
        perf_event_update_userpage(event);
}

/*
 * Holding the top-level event's child_mutex means that any
 * descendant process that has inherited this event will block
 * in sync_child_event if it goes to exit, thus satisfying the
 * task existence requirements of perf_event_enable/disable.
 */
static void perf_event_for_each_child(struct perf_event *event,
                                        void (*func)(struct perf_event *))
{
        struct perf_event *child;

        WARN_ON_ONCE(event->ctx->parent_ctx);

        mutex_lock(&event->child_mutex);
        func(event);
        list_for_each_entry(child, &event->child_list, child_list)
                func(child);
        mutex_unlock(&event->child_mutex);
}

static void perf_event_for_each(struct perf_event *event,
                                  void (*func)(struct perf_event *))
{
        struct perf_event_context *ctx = event->ctx;
        struct perf_event *sibling;

        lockdep_assert_held(&ctx->mutex);

        event = event->group_leader;

        perf_event_for_each_child(event, func);
        func(event);
        list_for_each_entry(sibling, &event->sibling_list, group_entry)
                perf_event_for_each_child(sibling, func);
}

static int perf_event_period(struct perf_event *event, u64 __user *arg)
{
        struct perf_event_context *ctx = event->ctx;
        int ret = 0;
        u64 value;

        if (!is_sampling_event(event))
                return -EINVAL;

        if (copy_from_user(&value, arg, sizeof(value)))
                return -EFAULT;

        if (!value)
                return -EINVAL;

        raw_spin_lock_irq(&ctx->lock);
        if (event->attr.freq) {
                if (value > sysctl_perf_event_sample_rate) {
                        ret = -EINVAL;
                        goto unlock;
                }

                event->attr.sample_freq = value;
        } else {
                event->attr.sample_period = value;
                event->hw.sample_period = value;
        }
unlock:
        raw_spin_unlock_irq(&ctx->lock);

        return ret;
}

static const struct file_operations perf_fops;

static struct file *perf_fget_light(int fd, int *fput_needed)
{
        struct file *file;

        file = fget_light(fd, fput_needed);
        if (!file)
                return ERR_PTR(-EBADF);

        if (file->f_op != &perf_fops) {
                fput_light(file, *fput_needed);
                *fput_needed = 0;
                return ERR_PTR(-EBADF);
        }

        return file;
}

static int perf_event_set_output(struct perf_event *event,
                                 struct perf_event *output_event);
static int perf_event_set_filter(struct perf_event *event, void __user *arg);

static long _perf_ioctl(struct perf_event *event, unsigned int cmd, unsigned long arg)
{
        void (*func)(struct perf_event *);
        u32 flags = arg;

        switch (cmd) {
        case PERF_EVENT_IOC_ENABLE:
                func = _perf_event_enable;
                break;
        case PERF_EVENT_IOC_DISABLE:
                func = _perf_event_disable;
                break;
        case PERF_EVENT_IOC_RESET:
                func = _perf_event_reset;
                break;

        case PERF_EVENT_IOC_REFRESH:
                return _perf_event_refresh(event, arg);

        case PERF_EVENT_IOC_PERIOD:
                return perf_event_period(event, (u64 __user *)arg);

        case PERF_EVENT_IOC_SET_OUTPUT:
        {
                struct file *output_file = NULL;
                struct perf_event *output_event = NULL;
                int fput_needed = 0;
                int ret;

                if (arg != -1) {
                        output_file = perf_fget_light(arg, &fput_needed);
                        if (IS_ERR(output_file))
                                return PTR_ERR(output_file);
                        output_event = output_file->private_data;
                }

                ret = perf_event_set_output(event, output_event);
                if (output_event)
                        fput_light(output_file, fput_needed);

                return ret;
        }

        case PERF_EVENT_IOC_SET_FILTER:
                return perf_event_set_filter(event, (void __user *)arg);

        default:
                return -ENOTTY;
        }

        if (flags & PERF_IOC_FLAG_GROUP)
                perf_event_for_each(event, func);
        else
                perf_event_for_each_child(event, func);

        return 0;
}

static long perf_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
        struct perf_event *event = file->private_data;
        struct perf_event_context *ctx;
        long ret;

        ctx = perf_event_ctx_lock(event);
        ret = _perf_ioctl(event, cmd, arg);
        perf_event_ctx_unlock(event, ctx);

        return ret;
}

#ifdef CONFIG_COMPAT
static long perf_compat_ioctl(struct file *file, unsigned int cmd,
                                unsigned long arg)
{
        switch (_IOC_NR(cmd)) {
        case _IOC_NR(PERF_EVENT_IOC_SET_FILTER):
                /* Fix up pointer size (usually 4 -> 8 in 32-on-64-bit case */
                if (_IOC_SIZE(cmd) == sizeof(compat_uptr_t)) {
                        cmd &= ~IOCSIZE_MASK;
                        cmd |= sizeof(void *) << IOCSIZE_SHIFT;
                }
                break;
        }
        return perf_ioctl(file, cmd, arg);
}
#else
# define perf_compat_ioctl NULL
#endif

int perf_event_task_enable(void)
{
        struct perf_event_context *ctx;
        struct perf_event *event;

        mutex_lock(&current->perf_event_mutex);
        list_for_each_entry(event, &current->perf_event_list, owner_entry) {
                ctx = perf_event_ctx_lock(event);
                perf_event_for_each_child(event, _perf_event_enable);
                perf_event_ctx_unlock(event, ctx);
        }
        mutex_unlock(&current->perf_event_mutex);

        return 0;
}

int perf_event_task_disable(void)
{
        struct perf_event_context *ctx;
        struct perf_event *event;

        mutex_lock(&current->perf_event_mutex);
        list_for_each_entry(event, &current->perf_event_list, owner_entry) {
                ctx = perf_event_ctx_lock(event);
                perf_event_for_each_child(event, _perf_event_disable);
                perf_event_ctx_unlock(event, ctx);
        }
        mutex_unlock(&current->perf_event_mutex);

        return 0;
}

#ifndef PERF_EVENT_INDEX_OFFSET
# define PERF_EVENT_INDEX_OFFSET 0
#endif

static int perf_event_index(struct perf_event *event)
{
        if (event->hw.state & PERF_HES_STOPPED)
                return 0;

        if (event->state != PERF_EVENT_STATE_ACTIVE)
                return 0;

        return event->hw.idx + 1 - PERF_EVENT_INDEX_OFFSET;
}

static void calc_timer_values(struct perf_event *event,
                                u64 *enabled,
                                u64 *running)
{
        u64 now, ctx_time;

        now = perf_clock();
        ctx_time = event->shadow_ctx_time + now;
        *enabled = ctx_time - event->tstamp_enabled;
        *running = ctx_time - event->tstamp_running;
}

/*
 * Callers need to ensure there can be no nesting of this function, otherwise
 * the seqlock logic goes bad. We can not serialize this because the arch
 * code calls this from NMI context.
 */
void perf_event_update_userpage(struct perf_event *event)
{
        struct perf_event_mmap_page *userpg;
        struct ring_buffer *rb;
        u64 enabled, running;

        rcu_read_lock();
        /*
         * compute total_time_enabled, total_time_running
         * based on snapshot values taken when the event
         * was last scheduled in.
         *
         * we cannot simply called update_context_time()
         * because of locking issue as we can be called in
         * NMI context
         */
        calc_timer_values(event, &enabled, &running);
        rb = rcu_dereference(event->rb);
        if (!rb)
                goto unlock;

        userpg = rb->user_page;

        /*
         * Disable preemption so as to not let the corresponding user-space
         * spin too long if we get preempted.
         */
        preempt_disable();
        ++userpg->lock;
        barrier();
        userpg->index = perf_event_index(event);
        userpg->offset = perf_event_count(event);
        if (event->state == PERF_EVENT_STATE_ACTIVE)
                userpg->offset -= local64_read(&event->hw.prev_count);

        userpg->time_enabled = enabled +
                        atomic64_read(&event->child_total_time_enabled);

        userpg->time_running = running +
                        atomic64_read(&event->child_total_time_running);

        barrier();
        ++userpg->lock;
        preempt_enable();
unlock:
        rcu_read_unlock();
}

static int perf_mmap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
        struct perf_event *event = vma->vm_file->private_data;
        struct ring_buffer *rb;
        int ret = VM_FAULT_SIGBUS;

        if (vmf->flags & FAULT_FLAG_MKWRITE) {
                if (vmf->pgoff == 0)
                        ret = 0;
                return ret;
        }

        rcu_read_lock();
        rb = rcu_dereference(event->rb);
        if (!rb)
                goto unlock;

        if (vmf->pgoff && (vmf->flags & FAULT_FLAG_WRITE))
                goto unlock;

        vmf->page = perf_mmap_to_page(rb, vmf->pgoff);
        if (!vmf->page)
                goto unlock;

        get_page(vmf->page);
        vmf->page->mapping = vma->vm_file->f_mapping;
        vmf->page->index   = vmf->pgoff;

        ret = 0;
unlock:
        rcu_read_unlock();

        return ret;
}

static void ring_buffer_attach(struct perf_event *event,
                               struct ring_buffer *rb)
{
        unsigned long flags;

        if (!list_empty(&event->rb_entry))
                return;

        spin_lock_irqsave(&rb->event_lock, flags);
        if (list_empty(&event->rb_entry))
                list_add(&event->rb_entry, &rb->event_list);
        spin_unlock_irqrestore(&rb->event_lock, flags);
}

static void ring_buffer_detach(struct perf_event *event, struct ring_buffer *rb)
{
        unsigned long flags;

        if (list_empty(&event->rb_entry))
                return;

        spin_lock_irqsave(&rb->event_lock, flags);
        list_del_init(&event->rb_entry);
        wake_up_all(&event->waitq);
        spin_unlock_irqrestore(&rb->event_lock, flags);
}

static void ring_buffer_wakeup(struct perf_event *event)
{
        struct ring_buffer *rb;

        rcu_read_lock();
        rb = rcu_dereference(event->rb);
        if (rb) {
                list_for_each_entry_rcu(event, &rb->event_list, rb_entry)
                        wake_up_all(&event->waitq);
        }
        rcu_read_unlock();
}

static void rb_free_rcu(struct rcu_head *rcu_head)
{
        struct ring_buffer *rb;

        rb = container_of(rcu_head, struct ring_buffer, rcu_head);
        rb_free(rb);
}

static struct ring_buffer *ring_buffer_get(struct perf_event *event)
{
        struct ring_buffer *rb;

        rcu_read_lock();
        rb = rcu_dereference(event->rb);
        if (rb) {
                if (!atomic_inc_not_zero(&rb->refcount))
                        rb = NULL;
        }
        rcu_read_unlock();

        return rb;
}

static void ring_buffer_put(struct ring_buffer *rb)
{
        if (!atomic_dec_and_test(&rb->refcount))
                return;

        WARN_ON_ONCE(!list_empty(&rb->event_list));

        call_rcu(&rb->rcu_head, rb_free_rcu);
}

static void perf_mmap_open(struct vm_area_struct *vma)
{
        struct perf_event *event = vma->vm_file->private_data;

        atomic_inc(&event->mmap_count);
        atomic_inc(&event->rb->mmap_count);
}

/*
 * A buffer can be mmap()ed multiple times; either directly through the same
 * event, or through other events by use of perf_event_set_output().
 *
 * In order to undo the VM accounting done by perf_mmap() we need to destroy
 * the buffer here, where we still have a VM context. This means we need
 * to detach all events redirecting to us.
 */
static void perf_mmap_close(struct vm_area_struct *vma)
{
        struct perf_event *event = vma->vm_file->private_data;

        struct ring_buffer *rb = event->rb;
        struct user_struct *mmap_user = rb->mmap_user;
        int mmap_locked = rb->mmap_locked;
        unsigned long size = perf_data_size(rb);

        atomic_dec(&rb->mmap_count);

        if (!atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex))
                return;

        /* Detach current event from the buffer. */
        rcu_assign_pointer(event->rb, NULL);
        ring_buffer_detach(event, rb);
        mutex_unlock(&event->mmap_mutex);

        /* If there's still other mmap()s of this buffer, we're done. */
        if (atomic_read(&rb->mmap_count)) {
                ring_buffer_put(rb); /* can't be last */
                return;
        }

        /*
         * No other mmap()s, detach from all other events that might redirect
         * into the now unreachable buffer. Somewhat complicated by the
         * fact that rb::event_lock otherwise nests inside mmap_mutex.
         */
again:
        rcu_read_lock();
        list_for_each_entry_rcu(event, &rb->event_list, rb_entry) {
                if (!atomic_long_inc_not_zero(&event->refcount)) {
                        /*
                         * This event is en-route to free_event() which will
                         * detach it and remove it from the list.
                         */
                        continue;
                }
                rcu_read_unlock();

                mutex_lock(&event->mmap_mutex);
                /*
                 * Check we didn't race with perf_event_set_output() which can
                 * swizzle the rb from under us while we were waiting to
                 * acquire mmap_mutex.
                 *
                 * If we find a different rb; ignore this event, a next
                 * iteration will no longer find it on the list. We have to
                 * still restart the iteration to make sure we're not now
                 * iterating the wrong list.
                 */
                if (event->rb == rb) {
                        rcu_assign_pointer(event->rb, NULL);
                        ring_buffer_detach(event, rb);
                        ring_buffer_put(rb); /* can't be last, we still have one */
                }
                mutex_unlock(&event->mmap_mutex);
                put_event(event);

                /*
                 * Restart the iteration; either we're on the wrong list or
                 * destroyed its integrity by doing a deletion.
                 */
                goto again;
        }
        rcu_read_unlock();

        /*
         * It could be there's still a few 0-ref events on the list; they'll
         * get cleaned up by free_event() -- they'll also still have their
         * ref on the rb and will free it whenever they are done with it.
         *
         * Aside from that, this buffer is 'fully' detached and unmapped,
         * undo the VM accounting.
         */

        atomic_long_sub((size >> PAGE_SHIFT) + 1, &mmap_user->locked_vm);
        vma->vm_mm->pinned_vm -= mmap_locked;
        free_uid(mmap_user);

        ring_buffer_put(rb); /* could be last */
}

static const struct vm_operations_struct perf_mmap_vmops = {
        .open           = perf_mmap_open,
        .close          = perf_mmap_close,
        .fault          = perf_mmap_fault,
        .page_mkwrite   = perf_mmap_fault,
};

static int perf_mmap(struct file *file, struct vm_area_struct *vma)
{
        struct perf_event *event = file->private_data;
        unsigned long user_locked, user_lock_limit;
        struct user_struct *user = current_user();
        unsigned long locked, lock_limit;
        struct ring_buffer *rb;
        unsigned long vma_size;
        unsigned long nr_pages;
        long user_extra, extra;
        int ret = 0, flags = 0;

        /*
         * Don't allow mmap() of inherited per-task counters. This would
         * create a performance issue due to all children writing to the
         * same rb.
         */
        if (event->cpu == -1 && event->attr.inherit)
                return -EINVAL;

        if (!(vma->vm_flags & VM_SHARED))
                return -EINVAL;

        vma_size = vma->vm_end - vma->vm_start;
        nr_pages = (vma_size / PAGE_SIZE) - 1;

        /*
         * If we have rb pages ensure they're a power-of-two number, so we
         * can do bitmasks instead of modulo.
         */
        if (nr_pages != 0 && !is_power_of_2(nr_pages))
                return -EINVAL;

        if (vma_size != PAGE_SIZE * (1 + nr_pages))
                return -EINVAL;

        if (vma->vm_pgoff != 0)
                return -EINVAL;

        WARN_ON_ONCE(event->ctx->parent_ctx);
again:
        mutex_lock(&event->mmap_mutex);
        if (event->rb) {
                if (event->rb->nr_pages != nr_pages) {
                        ret = -EINVAL;
                        goto unlock;
                }

                if (!atomic_inc_not_zero(&event->rb->mmap_count)) {
                        /*
                         * Raced against perf_mmap_close() through
                         * perf_event_set_output(). Try again, hope for better
                         * luck.
                         */
                        mutex_unlock(&event->mmap_mutex);
                        goto again;
                }

                goto unlock;
        }

        user_extra = nr_pages + 1;
        user_lock_limit = sysctl_perf_event_mlock >> (PAGE_SHIFT - 10);

        /*
         * Increase the limit linearly with more CPUs:
         */
        user_lock_limit *= num_online_cpus();

        user_locked = atomic_long_read(&user->locked_vm) + user_extra;

        extra = 0;
        if (user_locked > user_lock_limit)
                extra = user_locked - user_lock_limit;

        lock_limit = rlimit(RLIMIT_MEMLOCK);
        lock_limit >>= PAGE_SHIFT;
        locked = vma->vm_mm->pinned_vm + extra;

        if ((locked > lock_limit) && perf_paranoid_tracepoint_raw() &&
                !capable(CAP_IPC_LOCK)) {
                ret = -EPERM;
                goto unlock;
        }

        WARN_ON(event->rb);

        if (vma->vm_flags & VM_WRITE)
                flags |= RING_BUFFER_WRITABLE;

        rb = rb_alloc(nr_pages, 
                event->attr.watermark ? event->attr.wakeup_watermark : 0,
                event->cpu, flags);

        if (!rb) {
                ret = -ENOMEM;
                goto unlock;
        }

        atomic_set(&rb->mmap_count, 1);
        rb->mmap_locked = extra;
        rb->mmap_user = get_current_user();

        atomic_long_add(user_extra, &user->locked_vm);
        vma->vm_mm->pinned_vm += extra;

        ring_buffer_attach(event, rb);
        rcu_assign_pointer(event->rb, rb);

unlock:
        if (!ret)
                atomic_inc(&event->mmap_count);
        mutex_unlock(&event->mmap_mutex);

        /*
         * Since pinned accounting is per vm we cannot allow fork() to copy our
         * vma.
         */
        vma->vm_flags |= VM_DONTCOPY | VM_RESERVED;
        vma->vm_ops = &perf_mmap_vmops;

        return ret;
}

static int perf_fasync(int fd, struct file *filp, int on)
{
        struct inode *inode = filp->f_path.dentry->d_inode;
        struct perf_event *event = filp->private_data;
        int retval;

        mutex_lock(&inode->i_mutex);
        retval = fasync_helper(fd, filp, on, &event->fasync);
        mutex_unlock(&inode->i_mutex);

        if (retval < 0)
                return retval;

        return 0;
}

static const struct file_operations perf_fops = {
        .llseek                 = no_llseek,
        .release                = perf_release,
        .read                   = perf_read,
        .poll                   = perf_poll,
        .unlocked_ioctl         = perf_ioctl,
        .compat_ioctl           = perf_compat_ioctl,
        .mmap                   = perf_mmap,
        .fasync                 = perf_fasync,
};

/*
 * Perf event wakeup
 *
 * If there's data, ensure we set the poll() state and publish everything
 * to user-space before waking everybody up.
 */

static inline struct fasync_struct **perf_event_fasync(struct perf_event *event)
{
        /* only the parent has fasync state */
        if (event->parent)
                event = event->parent;
        return &event->fasync;
}

void perf_event_wakeup(struct perf_event *event)
{
        ring_buffer_wakeup(event);

        if (event->pending_kill) {
                kill_fasync(perf_event_fasync(event), SIGIO, event->pending_kill);
                event->pending_kill = 0;
        }
}

static void perf_pending_event(struct irq_work *entry)
{
        struct perf_event *event = container_of(entry,
                        struct perf_event, pending);
        int rctx;

        rctx = perf_swevent_get_recursion_context();
        /*
         * If we 'fail' here, that's OK, it means recursion is already disabled
         * and we won't recurse 'further'.
         */

        if (event->pending_disable) {
                event->pending_disable = 0;
                __perf_event_disable(event);
        }

        if (event->pending_wakeup) {
                event->pending_wakeup = 0;
                perf_event_wakeup(event);
        }

        if (rctx >= 0)
                perf_swevent_put_recursion_context(rctx);
}

/*
 * We assume there is only KVM supporting the callbacks.
 * Later on, we might change it to a list if there is
 * another virtualization implementation supporting the callbacks.
 */
struct perf_guest_info_callbacks *perf_guest_cbs;

int perf_register_guest_info_callbacks(struct perf_guest_info_callbacks *cbs)
{
        perf_guest_cbs = cbs;
        return 0;
}
EXPORT_SYMBOL_GPL(perf_register_guest_info_callbacks);

int perf_unregister_guest_info_callbacks(struct perf_guest_info_callbacks *cbs)
{
        perf_guest_cbs = NULL;
        return 0;
}
EXPORT_SYMBOL_GPL(perf_unregister_guest_info_callbacks);

static void __perf_event_header__init_id(struct perf_event_header *header,
                                         struct perf_sample_data *data,
                                         struct perf_event *event)
{
        u64 sample_type = event->attr.sample_type;

        data->type = sample_type;
        header->size += event->id_header_size;

        if (sample_type & PERF_SAMPLE_TID) {
                /* namespace issues */
                data->tid_entry.pid = perf_event_pid(event, current);
                data->tid_entry.tid = perf_event_tid(event, current);
        }

        if (sample_type & PERF_SAMPLE_TIME)
                data->time = perf_clock();

        if (sample_type & PERF_SAMPLE_ID)
                data->id = primary_event_id(event);

        if (sample_type & PERF_SAMPLE_STREAM_ID)
                data->stream_id = event->id;

        if (sample_type & PERF_SAMPLE_CPU) {
                data->cpu_entry.cpu      = raw_smp_processor_id();
                data->cpu_entry.reserved = 0;
        }
}

void perf_event_header__init_id(struct perf_event_header *header,
                                struct perf_sample_data *data,
                                struct perf_event *event)
{
        if (event->attr.sample_id_all)
                __perf_event_header__init_id(header, data, event);
}

static void __perf_event__output_id_sample(struct perf_output_handle *handle,
                                           struct perf_sample_data *data)
{
        u64 sample_type = data->type;

        if (sample_type & PERF_SAMPLE_TID)
                perf_output_put(handle, data->tid_entry);

        if (sample_type & PERF_SAMPLE_TIME)
                perf_output_put(handle, data->time);

        if (sample_type & PERF_SAMPLE_ID)
                perf_output_put(handle, data->id);

        if (sample_type & PERF_SAMPLE_STREAM_ID)
                perf_output_put(handle, data->stream_id);

        if (sample_type & PERF_SAMPLE_CPU)
                perf_output_put(handle, data->cpu_entry);
}

void perf_event__output_id_sample(struct perf_event *event,
                                  struct perf_output_handle *handle,
                                  struct perf_sample_data *sample)
{
        if (event->attr.sample_id_all)
                __perf_event__output_id_sample(handle, sample);
}

static void perf_output_read_one(struct perf_output_handle *handle,
                                 struct perf_event *event,
                                 u64 enabled, u64 running)
{
        u64 read_format = event->attr.read_format;
        u64 values[4];
        int n = 0;

        values[n++] = perf_event_count(event);
        if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) {
                values[n++] = enabled +
                        atomic64_read(&event->child_total_time_enabled);
        }
        if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) {
                values[n++] = running +
                        atomic64_read(&event->child_total_time_running);
        }
        if (read_format & PERF_FORMAT_ID)
                values[n++] = primary_event_id(event);

        __output_copy(handle, values, n * sizeof(u64));
}

static void perf_output_read_group(struct perf_output_handle *handle,
                            struct perf_event *event,
                            u64 enabled, u64 running)
{
        struct perf_event *leader = event->group_leader, *sub;
        u64 read_format = event->attr.read_format;
        u64 values[5];
        int n = 0;

        values[n++] = 1 + leader->nr_siblings;

        if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
                values[n++] = enabled;

        if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
                values[n++] = running;

        if (leader != event)
                leader->pmu->read(leader);

        values[n++] = perf_event_count(leader);
        if (read_format & PERF_FORMAT_ID)
                values[n++] = primary_event_id(leader);

        __output_copy(handle, values, n * sizeof(u64));

        list_for_each_entry(sub, &leader->sibling_list, group_entry) {
                n = 0;

                if (sub != event)
                        sub->pmu->read(sub);

                values[n++] = perf_event_count(sub);
                if (read_format & PERF_FORMAT_ID)
                        values[n++] = primary_event_id(sub);

                __output_copy(handle, values, n * sizeof(u64));
        }
}

#define PERF_FORMAT_TOTAL_TIMES (PERF_FORMAT_TOTAL_TIME_ENABLED|\
                                 PERF_FORMAT_TOTAL_TIME_RUNNING)

/*
 * XXX PERF_SAMPLE_READ vs inherited events seems difficult.
 *
 * The problem is that its both hard and excessively expensive to iterate the
 * child list, not to mention that its impossible to IPI the children running
 * on another CPU, from interrupt/NMI context.
 */
static void perf_output_read(struct perf_output_handle *handle,
                             struct perf_event *event)
{
        u64 enabled = 0, running = 0;
        u64 read_format = event->attr.read_format;

        /*
         * compute total_time_enabled, total_time_running
         * based on snapshot values taken when the event
         * was last scheduled in.
         *
         * we cannot simply called update_context_time()
         * because of locking issue as we are called in
         * NMI context
         */
        if (read_format & PERF_FORMAT_TOTAL_TIMES)
                calc_timer_values(event, &enabled, &running);

        if (event->attr.read_format & PERF_FORMAT_GROUP)
                perf_output_read_group(handle, event, enabled, running);
        else
                perf_output_read_one(handle, event, enabled, running);
}

void perf_output_sample(struct perf_output_handle *handle,
                        struct perf_event_header *header,
                        struct perf_sample_data *data,
                        struct perf_event *event)
{
        u64 sample_type = data->type;

        perf_output_put(handle, *header);

        if (sample_type & PERF_SAMPLE_IP)
                perf_output_put(handle, data->ip);

        if (sample_type & PERF_SAMPLE_TID)
                perf_output_put(handle, data->tid_entry);

        if (sample_type & PERF_SAMPLE_TIME)
                perf_output_put(handle, data->time);

        if (sample_type & PERF_SAMPLE_ADDR)
                perf_output_put(handle, data->addr);

        if (sample_type & PERF_SAMPLE_ID)
                perf_output_put(handle, data->id);

        if (sample_type & PERF_SAMPLE_STREAM_ID)
                perf_output_put(handle, data->stream_id);

        if (sample_type & PERF_SAMPLE_CPU)
                perf_output_put(handle, data->cpu_entry);

        if (sample_type & PERF_SAMPLE_PERIOD)
                perf_output_put(handle, data->period);

        if (sample_type & PERF_SAMPLE_READ)
                perf_output_read(handle, event);

        if (sample_type & PERF_SAMPLE_CALLCHAIN) {
                if (data->callchain) {
                        int size = 1;

                        if (data->callchain)
                                size += data->callchain->nr;

                        size *= sizeof(u64);

                        __output_copy(handle, data->callchain, size);
                } else {
                        u64 nr = 0;
                        perf_output_put(handle, nr);
                }
        }

        if (sample_type & PERF_SAMPLE_RAW) {
                if (data->raw) {
                        perf_output_put(handle, data->raw->size);
                        __output_copy(handle, data->raw->data,
                                           data->raw->size);
                } else {
                        struct {
                                u32     size;
                                u32     data;
                        } raw = {
                                .size = sizeof(u32),
                                .data = 0,
                        };
                        perf_output_put(handle, raw);
                }
        }

        if (!event->attr.watermark) {
                int wakeup_events = event->attr.wakeup_events;

                if (wakeup_events) {
                        struct ring_buffer *rb = handle->rb;
                        int events = local_inc_return(&rb->events);

                        if (events >= wakeup_events) {
                                local_sub(wakeup_events, &rb->events);
                                local_inc(&rb->wakeup);
                        }
                }
        }
}

void perf_prepare_sample(struct perf_event_header *header,
                         struct perf_sample_data *data,
                         struct perf_event *event,
                         struct pt_regs *regs)
{
        u64 sample_type = event->attr.sample_type;

        header->type = PERF_RECORD_SAMPLE;
        header->size = sizeof(*header) + event->header_size;

        header->misc = 0;
        header->misc |= perf_misc_flags(regs);

        __perf_event_header__init_id(header, data, event);

        if (sample_type & PERF_SAMPLE_IP)
                data->ip = perf_instruction_pointer(regs);

        if (sample_type & PERF_SAMPLE_CALLCHAIN) {
                int size = 1;

                data->callchain = perf_callchain(regs);

                if (data->callchain)
                        size += data->callchain->nr;

                header->size += size * sizeof(u64);
        }

        if (sample_type & PERF_SAMPLE_RAW) {
                int size = sizeof(u32);

                if (data->raw)
                        size += data->raw->size;
                else
                        size += sizeof(u32);

                WARN_ON_ONCE(size & (sizeof(u64)-1));
                header->size += size;
        }
}

static void perf_event_output(struct perf_event *event,
                                struct perf_sample_data *data,
                                struct pt_regs *regs)
{
        struct perf_output_handle handle;
        struct perf_event_header header;

        /* protect the callchain buffers */
        rcu_read_lock();

        perf_prepare_sample(&header, data, event, regs);

        if (perf_output_begin(&handle, event, header.size))
                goto exit;

        perf_output_sample(&handle, &header, data, event);

        perf_output_end(&handle);

exit:
        rcu_read_unlock();
}

/*
 * read event_id
 */

struct perf_read_event {
        struct perf_event_header        header;

        u32                             pid;
        u32                             tid;
};

static void
perf_event_read_event(struct perf_event *event,
                        struct task_struct *task)
{
        struct perf_output_handle handle;
        struct perf_sample_data sample;
        struct perf_read_event read_event = {
                .header = {
                        .type = PERF_RECORD_READ,
                        .misc = 0,
                        .size = sizeof(read_event) + event->read_size,
                },
                .pid = perf_event_pid(event, task),
                .tid = perf_event_tid(event, task),
        };
        int ret;

        perf_event_header__init_id(&read_event.header, &sample, event);
        ret = perf_output_begin(&handle, event, read_event.header.size);
        if (ret)
                return;

        perf_output_put(&handle, read_event);
        perf_output_read(&handle, event);
        perf_event__output_id_sample(event, &handle, &sample);

        perf_output_end(&handle);
}

/*
 * task tracking -- fork/exit
 *
 * enabled by: attr.comm | attr.mmap | attr.mmap_data | attr.task
 */

struct perf_task_event {
        struct task_struct              *task;
        struct perf_event_context       *task_ctx;

        struct {
                struct perf_event_header        header;

                u32                             pid;
                u32                             ppid;
                u32                             tid;
                u32                             ptid;
                u64                             time;
        } event_id;
};

static void perf_event_task_output(struct perf_event *event,
                                     struct perf_task_event *task_event)
{
        struct perf_output_handle handle;
        struct perf_sample_data sample;
        struct task_struct *task = task_event->task;
        int ret, size = task_event->event_id.header.size;

        perf_event_header__init_id(&task_event->event_id.header, &sample, event);

        ret = perf_output_begin(&handle, event,
                                task_event->event_id.header.size);
        if (ret)
                goto out;

        task_event->event_id.pid = perf_event_pid(event, task);
        task_event->event_id.ppid = perf_event_pid(event, current);

        task_event->event_id.tid = perf_event_tid(event, task);
        task_event->event_id.ptid = perf_event_tid(event, current);

        perf_output_put(&handle, task_event->event_id);

        perf_event__output_id_sample(event, &handle, &sample);

        perf_output_end(&handle);
out:
        task_event->event_id.header.size = size;
}

static int perf_event_task_match(struct perf_event *event)
{
        if (event->state < PERF_EVENT_STATE_INACTIVE)
                return 0;

        if (!event_filter_match(event))
                return 0;

        if (event->attr.comm || event->attr.mmap ||
            event->attr.mmap_data || event->attr.task)
                return 1;

        return 0;
}

static void perf_event_task_ctx(struct perf_event_context *ctx,
                                  struct perf_task_event *task_event)
{
        struct perf_event *event;

        list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
                if (perf_event_task_match(event))
                        perf_event_task_output(event, task_event);
        }
}

static void perf_event_task_event(struct perf_task_event *task_event)
{
        struct perf_cpu_context *cpuctx;
        struct perf_event_context *ctx;
        struct pmu *pmu;
        int ctxn;

        rcu_read_lock();
        list_for_each_entry_rcu(pmu, &pmus, entry) {
                cpuctx = get_cpu_ptr(pmu->pmu_cpu_context);
                if (cpuctx->unique_pmu != pmu)
                        goto next;
                perf_event_task_ctx(&cpuctx->ctx, task_event);

                ctx = task_event->task_ctx;
                if (!ctx) {
                        ctxn = pmu->task_ctx_nr;
                        if (ctxn < 0)
                                goto next;
                        ctx = rcu_dereference(current->perf_event_ctxp[ctxn]);
                }
                if (ctx)
                        perf_event_task_ctx(ctx, task_event);
next:
                put_cpu_ptr(pmu->pmu_cpu_context);
        }
        rcu_read_unlock();
}

static void perf_event_task(struct task_struct *task,
                              struct perf_event_context *task_ctx,
                              int new)
{
        struct perf_task_event task_event;

        if (!atomic_read(&nr_comm_events) &&
            !atomic_read(&nr_mmap_events) &&
            !atomic_read(&nr_task_events))
                return;

        task_event = (struct perf_task_event){
                .task     = task,
                .task_ctx = task_ctx,
                .event_id    = {
                        .header = {
                                .type = new ? PERF_RECORD_FORK : PERF_RECORD_EXIT,
                                .misc = 0,
                                .size = sizeof(task_event.event_id),
                        },
                        /* .pid  */
                        /* .ppid */
                        /* .tid  */
                        /* .ptid */
                        .time = perf_clock(),
                },
        };

        perf_event_task_event(&task_event);
}

void perf_event_fork(struct task_struct *task)
{
        perf_event_task(task, NULL, 1);
}

/*
 * comm tracking
 */

struct perf_comm_event {
        struct task_struct      *task;
        char                    *comm;
        int                     comm_size;

        struct {
                struct perf_event_header        header;

                u32                             pid;
                u32                             tid;
        } event_id;
};

static void perf_event_comm_output(struct perf_event *event,
                                     struct perf_comm_event *comm_event)
{
        struct perf_output_handle handle;
        struct perf_sample_data sample;
        int size = comm_event->event_id.header.size;
        int ret;

        perf_event_header__init_id(&comm_event->event_id.header, &sample, event);
        ret = perf_output_begin(&handle, event,
                                comm_event->event_id.header.size);

        if (ret)
                goto out;

        comm_event->event_id.pid = perf_event_pid(event, comm_event->task);
        comm_event->event_id.tid = perf_event_tid(event, comm_event->task);

        perf_output_put(&handle, comm_event->event_id);
        __output_copy(&handle, comm_event->comm,
                                   comm_event->comm_size);

        perf_event__output_id_sample(event, &handle, &sample);

        perf_output_end(&handle);
out:
        comm_event->event_id.header.size = size;
}

static int perf_event_comm_match(struct perf_event *event)
{
        if (event->state < PERF_EVENT_STATE_INACTIVE)
                return 0;

        if (!event_filter_match(event))
                return 0;

        if (event->attr.comm)
                return 1;

        return 0;
}

static void perf_event_comm_ctx(struct perf_event_context *ctx,
                                  struct perf_comm_event *comm_event)
{
        struct perf_event *event;

        list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
                if (perf_event_comm_match(event))
                        perf_event_comm_output(event, comm_event);
        }
}

static void perf_event_comm_event(struct perf_comm_event *comm_event)
{
        struct perf_cpu_context *cpuctx;
        struct perf_event_context *ctx;
        char comm[TASK_COMM_LEN];
        unsigned int size;
        struct pmu *pmu;
        int ctxn;

        memset(comm, 0, sizeof(comm));
        strlcpy(comm, comm_event->task->comm, sizeof(comm));
        size = ALIGN(strlen(comm)+1, sizeof(u64));

        comm_event->comm = comm;
        comm_event->comm_size = size;

        comm_event->event_id.header.size = sizeof(comm_event->event_id) + size;
        rcu_read_lock();
        list_for_each_entry_rcu(pmu, &pmus, entry) {
                cpuctx = get_cpu_ptr(pmu->pmu_cpu_context);
                if (cpuctx->unique_pmu != pmu)
                        goto next;
                perf_event_comm_ctx(&cpuctx->ctx, comm_event);

                ctxn = pmu->task_ctx_nr;
                if (ctxn < 0)
                        goto next;

                ctx = rcu_dereference(current->perf_event_ctxp[ctxn]);
                if (ctx)
                        perf_event_comm_ctx(ctx, comm_event);
next:
                put_cpu_ptr(pmu->pmu_cpu_context);
        }
        rcu_read_unlock();
}

void perf_event_comm(struct task_struct *task)
{
        struct perf_comm_event comm_event;
        struct perf_event_context *ctx;
        int ctxn;

        for_each_task_context_nr(ctxn) {
                ctx = task->perf_event_ctxp[ctxn];
                if (!ctx)
                        continue;

                perf_event_enable_on_exec(ctx);
        }

        if (!atomic_read(&nr_comm_events))
                return;

        comm_event = (struct perf_comm_event){
                .task   = task,
                /* .comm      */
                /* .comm_size */
                .event_id  = {
                        .header = {
                                .type = PERF_RECORD_COMM,
                                .misc = 0,
                                /* .size */
                        },
                        /* .pid */
                        /* .tid */
                },
        };

        perf_event_comm_event(&comm_event);
}

/*
 * mmap tracking
 */

struct perf_mmap_event {
        struct vm_area_struct   *vma;

        const char              *file_name;
        int                     file_size;

        struct {
                struct perf_event_header        header;

                u32                             pid;
                u32                             tid;
                u64                             start;
                u64                             len;
                u64                             pgoff;
        } event_id;
};

static void perf_event_mmap_output(struct perf_event *event,
                                     struct perf_mmap_event *mmap_event)
{
        struct perf_output_handle handle;
        struct perf_sample_data sample;
        int size = mmap_event->event_id.header.size;
        int ret;

        perf_event_header__init_id(&mmap_event->event_id.header, &sample, event);
        ret = perf_output_begin(&handle, event,
                                mmap_event->event_id.header.size);
        if (ret)
                goto out;

        mmap_event->event_id.pid = perf_event_pid(event, current);
        mmap_event->event_id.tid = perf_event_tid(event, current);

        perf_output_put(&handle, mmap_event->event_id);
        __output_copy(&handle, mmap_event->file_name,
                                   mmap_event->file_size);

        perf_event__output_id_sample(event, &handle, &sample);

        perf_output_end(&handle);
out:
        mmap_event->event_id.header.size = size;
}

static int perf_event_mmap_match(struct perf_event *event,
                                   struct perf_mmap_event *mmap_event,
                                   int executable)
{
        if (event->state < PERF_EVENT_STATE_INACTIVE)
                return 0;

        if (!event_filter_match(event))
                return 0;

        if ((!executable && event->attr.mmap_data) ||
            (executable && event->attr.mmap))
                return 1;

        return 0;
}

static void perf_event_mmap_ctx(struct perf_event_context *ctx,
                                  struct perf_mmap_event *mmap_event,
                                  int executable)
{
        struct perf_event *event;

        list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
                if (perf_event_mmap_match(event, mmap_event, executable))
                        perf_event_mmap_output(event, mmap_event);
        }
}

static void perf_event_mmap_event(struct perf_mmap_event *mmap_event)
{
        struct perf_cpu_context *cpuctx;
        struct perf_event_context *ctx;
        struct vm_area_struct *vma = mmap_event->vma;
        struct file *file = vma->vm_file;
        unsigned int size;
        char tmp[16];
        char *buf = NULL;
        const char *name;
        struct pmu *pmu;
        int ctxn;

        memset(tmp, 0, sizeof(tmp));

        if (file) {
                /*
                 * d_path works from the end of the rb backwards, so we
                 * need to add enough zero bytes after the string to handle
                 * the 64bit alignment we do later.
                 */
                buf = kzalloc(PATH_MAX + sizeof(u64), GFP_KERNEL);
                if (!buf) {
                        name = strncpy(tmp, "//enomem", sizeof(tmp));
                        goto got_name;
                }
                name = d_path(&file->f_path, buf, PATH_MAX);
                if (IS_ERR(name)) {
                        name = strncpy(tmp, "//toolong", sizeof(tmp));
                        goto got_name;
                }
        } else {
                if (arch_vma_name(mmap_event->vma)) {
                        name = strncpy(tmp, arch_vma_name(mmap_event->vma),
                                       sizeof(tmp));
                        goto got_name;
                }

                if (!vma->vm_mm) {
                        name = strncpy(tmp, "[vdso]", sizeof(tmp));
                        goto got_name;
                } else if (vma->vm_start <= vma->vm_mm->start_brk &&
                                vma->vm_end >= vma->vm_mm->brk) {
                        name = strncpy(tmp, "[heap]", sizeof(tmp));
                        goto got_name;
                } else if (vma->vm_start <= vma->vm_mm->start_stack &&
                                vma->vm_end >= vma->vm_mm->start_stack) {
                        name = strncpy(tmp, "[stack]", sizeof(tmp));
                        goto got_name;
                }

                name = strncpy(tmp, "//anon", sizeof(tmp));
                goto got_name;
        }

got_name:
        size = ALIGN(strlen(name)+1, sizeof(u64));

        mmap_event->file_name = name;
        mmap_event->file_size = size;

        mmap_event->event_id.header.size = sizeof(mmap_event->event_id) + size;

        rcu_read_lock();
        list_for_each_entry_rcu(pmu, &pmus, entry) {
                cpuctx = get_cpu_ptr(pmu->pmu_cpu_context);
                if (cpuctx->unique_pmu != pmu)
                        goto next;
                perf_event_mmap_ctx(&cpuctx->ctx, mmap_event,
                                        vma->vm_flags & VM_EXEC);

                ctxn = pmu->task_ctx_nr;
                if (ctxn < 0)
                        goto next;

                ctx = rcu_dereference(current->perf_event_ctxp[ctxn]);
                if (ctx) {
                        perf_event_mmap_ctx(ctx, mmap_event,
                                        vma->vm_flags & VM_EXEC);
                }
next:
                put_cpu_ptr(pmu->pmu_cpu_context);
        }
        rcu_read_unlock();

        kfree(buf);
}

void perf_event_mmap(struct vm_area_struct *vma)
{
        struct perf_mmap_event mmap_event;

        if (!atomic_read(&nr_mmap_events))
                return;

        mmap_event = (struct perf_mmap_event){
                .vma    = vma,
                /* .file_name */
                /* .file_size */
                .event_id  = {
                        .header = {
                                .type = PERF_RECORD_MMAP,
                                .misc = PERF_RECORD_MISC_USER,
                                /* .size */
                        },
                        /* .pid */
                        /* .tid */
                        .start  = vma->vm_start,
                        .len    = vma->vm_end - vma->vm_start,
                        .pgoff  = (u64)vma->vm_pgoff << PAGE_SHIFT,
                },
        };

        perf_event_mmap_event(&mmap_event);
}

/*
 * IRQ throttle logging
 */

static void perf_log_throttle(struct perf_event *event, int enable)
{
        struct perf_output_handle handle;
        struct perf_sample_data sample;
        int ret;

        struct {
                struct perf_event_header        header;
                u64                             time;
                u64                             id;
                u64                             stream_id;
        } throttle_event = {
                .header = {
                        .type = PERF_RECORD_THROTTLE,
                        .misc = 0,
                        .size = sizeof(throttle_event),
                },
                .time           = perf_clock(),
                .id             = primary_event_id(event),
                .stream_id      = event->id,
        };

        if (enable)
                throttle_event.header.type = PERF_RECORD_UNTHROTTLE;

        perf_event_header__init_id(&throttle_event.header, &sample, event);

        ret = perf_output_begin(&handle, event,
                                throttle_event.header.size);
        if (ret)
                return;

        perf_output_put(&handle, throttle_event);
        perf_event__output_id_sample(event, &handle, &sample);
        perf_output_end(&handle);
}

/*
 * Generic event overflow handling, sampling.
 */

static int __perf_event_overflow(struct perf_event *event,
                                   int throttle, struct perf_sample_data *data,
                                   struct pt_regs *regs)
{
        int events = atomic_read(&event->event_limit);
        struct hw_perf_event *hwc = &event->hw;
        int ret = 0;

        /*
         * Non-sampling counters might still use the PMI to fold short
         * hardware counters, ignore those.
         */
        if (unlikely(!is_sampling_event(event)))
                return 0;

        if (unlikely(hwc->interrupts >= max_samples_per_tick)) {
                if (throttle) {
                        hwc->interrupts = MAX_INTERRUPTS;
                        perf_log_throttle(event, 0);
                        ret = 1;
                }
        } else
                hwc->interrupts++;

        if (event->attr.freq) {
                u64 now = perf_clock();
                s64 delta = now - hwc->freq_time_stamp;

                hwc->freq_time_stamp = now;

                if (delta > 0 && delta < 2*TICK_NSEC)
                        perf_adjust_period(event, delta, hwc->last_period);
        }

        /*
         * XXX event_limit might not quite work as expected on inherited
         * events
         */

        event->pending_kill = POLL_IN;
        if (events && atomic_dec_and_test(&event->event_limit)) {
                ret = 1;
                event->pending_kill = POLL_HUP;
                event->pending_disable = 1;
                irq_work_queue(&event->pending);
        }

        if (event->overflow_handler)
                event->overflow_handler(event, data, regs);
        else
                perf_event_output(event, data, regs);

        if (*perf_event_fasync(event) && event->pending_kill) {
                event->pending_wakeup = 1;
                irq_work_queue(&event->pending);
        }

        return ret;
}

int perf_event_overflow(struct perf_event *event,
                          struct perf_sample_data *data,
                          struct pt_regs *regs)
{
        return __perf_event_overflow(event, 1, data, regs);
}

/*
 * Generic software event infrastructure
 */

struct swevent_htable {
        struct swevent_hlist            *swevent_hlist;
        struct mutex                    hlist_mutex;
        int                             hlist_refcount;

        /* Recursion avoidance in each contexts */
        int                             recursion[PERF_NR_CONTEXTS];
};

static DEFINE_PER_CPU(struct swevent_htable, swevent_htable);

/*
 * We directly increment event->count and keep a second value in
 * event->hw.period_left to count intervals. This period event
 * is kept in the range [-sample_period, 0] so that we can use the
 * sign as trigger.
 */

static u64 perf_swevent_set_period(struct perf_event *event)
{
        struct hw_perf_event *hwc = &event->hw;
        u64 period = hwc->last_period;
        u64 nr, offset;
        s64 old, val;

        hwc->last_period = hwc->sample_period;

again:
        old = val = local64_read(&hwc->period_left);
        if (val < 0)
                return 0;

        nr = div64_u64(period + val, period);
        offset = nr * period;
        val -= offset;
        if (local64_cmpxchg(&hwc->period_left, old, val) != old)
                goto again;

        return nr;
}

static void perf_swevent_overflow(struct perf_event *event, u64 overflow,
                                    struct perf_sample_data *data,
                                    struct pt_regs *regs)
{
        struct hw_perf_event *hwc = &event->hw;
        int throttle = 0;

        data->period = event->hw.last_period;
        if (!overflow)
                overflow = perf_swevent_set_period(event);

        if (hwc->interrupts == MAX_INTERRUPTS)
                return;

        for (; overflow; overflow--) {
                if (__perf_event_overflow(event, throttle,
                                            data, regs)) {
                        /*
                         * We inhibit the overflow from happening when
                         * hwc->interrupts == MAX_INTERRUPTS.
                         */
                        break;
                }
                throttle = 1;
        }
}

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 (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 int perf_exclude_event(struct perf_event *event,
                              struct pt_regs *regs)
{
        if (event->hw.state & PERF_HES_STOPPED)
                return 1;

        if (regs) {
                if (event->attr.exclude_user && user_mode(regs))
                        return 1;

                if (event->attr.exclude_kernel && !user_mode(regs))
                        return 1;
        }

        return 0;
}

static int perf_swevent_match(struct perf_event *event,
                                enum perf_type_id type,
                                u32 event_id,
                                struct perf_sample_data *data,
                                struct pt_regs *regs)
{
        if (event->attr.type != type)
                return 0;

        if (event->attr.config != event_id)
                return 0;

        if (perf_exclude_event(event, regs))
                return 0;

        return 1;
}

static inline u64 swevent_hash(u64 type, u32 event_id)
{
        u64 val = event_id | (type << 32);

        return hash_64(val, SWEVENT_HLIST_BITS);
}

static inline struct hlist_head *
__find_swevent_head(struct swevent_hlist *hlist, u64 type, u32 event_id)
{
        u64 hash = swevent_hash(type, event_id);

        return &hlist->heads[hash];
}

/* For the read side: events when they trigger */
static inline struct hlist_head *
find_swevent_head_rcu(struct swevent_htable *swhash, u64 type, u32 event_id)
{
        struct swevent_hlist *hlist;

        hlist = rcu_dereference(swhash->swevent_hlist);
        if (!hlist)
                return NULL;

        return __find_swevent_head(hlist, type, event_id);
}

/* For the event head insertion and removal in the hlist */
static inline struct hlist_head *
find_swevent_head(struct swevent_htable *swhash, struct perf_event *event)
{
        struct swevent_hlist *hlist;
        u32 event_id = event->attr.config;
        u64 type = event->attr.type;

        /*
         * Event scheduling is always serialized against hlist allocation
         * and release. Which makes the protected version suitable here.
         * The context lock guarantees that.
         */
        hlist = rcu_dereference_protected(swhash->swevent_hlist,
                                          lockdep_is_held(&event->ctx->lock));
        if (!hlist)
                return NULL;

        return __find_swevent_head(hlist, type, event_id);
}

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 = &__get_cpu_var(swevent_htable);
        struct perf_event *event;
        struct hlist_node *node;
        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, node, head, hlist_entry) {
                if (perf_swevent_match(event, type, event_id, data, regs))
                        perf_swevent_event(event, nr, data, regs);
        }
end:
        rcu_read_unlock();
}

int perf_swevent_get_recursion_context(void)
{
        struct swevent_htable *swhash = &__get_cpu_var(swevent_htable);

        return get_recursion_context(swhash->recursion);
}
EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context);

inline void perf_swevent_put_recursion_context(int rctx)
{
        struct swevent_htable *swhash = &__get_cpu_var(swevent_htable);

        put_recursion_context(swhash->recursion, rctx);
}

void __perf_sw_event(u32 event_id, u64 nr, struct pt_regs *regs, u64 addr)
{
        struct perf_sample_data data;
        int rctx;

        preempt_disable_notrace();
        rctx = perf_swevent_get_recursion_context();
        if (rctx < 0)
                return;

        perf_sample_data_init(&data, addr);

        do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, &data, regs);

        perf_swevent_put_recursion_context(rctx);
        preempt_enable_notrace();
}

static void perf_swevent_read(struct perf_event *event)
{
}

static int perf_swevent_add(struct perf_event *event, int flags)
{
        struct swevent_htable *swhash = &__get_cpu_var(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);

        return 0;
}

static void perf_swevent_del(struct perf_event *event, int flags)
{
        hlist_del_rcu(&event->hlist_entry);
}

static void perf_swevent_start(struct perf_event *event, int flags)
{
        event->hw.state = 0;
}

static void perf_swevent_stop(struct perf_event *event, int flags)
{
        event->hw.state = PERF_HES_STOPPED;
}

/* Deref the hlist from the update side */
static inline struct swevent_hlist *
swevent_hlist_deref(struct swevent_htable *swhash)
{
        return rcu_dereference_protected(swhash->swevent_hlist,
                                         lockdep_is_held(&swhash->hlist_mutex));
}

static void swevent_hlist_release(struct swevent_htable *swhash)
{
        struct swevent_hlist *hlist = swevent_hlist_deref(swhash);

        if (!hlist)
                return;

        rcu_assign_pointer(swhash->swevent_hlist, NULL);
        kfree_rcu(hlist, rcu_head);
}

static void swevent_hlist_put_cpu(struct perf_event *event, int cpu)
{
        struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu);

        mutex_lock(&swhash->hlist_mutex);

        if (!--swhash->hlist_refcount)
                swevent_hlist_release(swhash);

        mutex_unlock(&swhash->hlist_mutex);
}

static void swevent_hlist_put(struct perf_event *event)
{
        int cpu;

        if (event->cpu != -1) {
                swevent_hlist_put_cpu(event, event->cpu);
                return;
        }

        for_each_possible_cpu(cpu)
                swevent_hlist_put_cpu(event, cpu);
}

static int swevent_hlist_get_cpu(struct perf_event *event, int cpu)
{
        struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu);
        int err = 0;

        mutex_lock(&swhash->hlist_mutex);
        if (!swevent_hlist_deref(swhash) && cpu_online(cpu)) {
                struct swevent_hlist *hlist;

                hlist = kzalloc(sizeof(*hlist), GFP_KERNEL);
                if (!hlist) {
                        err = -ENOMEM;
                        goto exit;
                }
                rcu_assign_pointer(swhash->swevent_hlist, hlist);
        }
        swhash->hlist_refcount++;
exit:
        mutex_unlock(&swhash->hlist_mutex);

        return err;
}

static int swevent_hlist_get(struct perf_event *event)
{
        int err;
        int cpu, failed_cpu;

        if (event->cpu != -1)
                return swevent_hlist_get_cpu(event, event->cpu);

        get_online_cpus();
        for_each_possible_cpu(cpu) {
                err = swevent_hlist_get_cpu(event, cpu);
                if (err) {
                        failed_cpu = cpu;
                        goto fail;
                }
        }
        put_online_cpus();

        return 0;
fail:
        for_each_possible_cpu(cpu) {
                if (cpu == failed_cpu)
                        break;
                swevent_hlist_put_cpu(event, cpu);
        }

        put_online_cpus();
        return err;
}

struct jump_label_key perf_swevent_enabled[PERF_COUNT_SW_MAX];

static void sw_perf_event_destroy(struct perf_event *event)
{
        u64 event_id = event->attr.config;

        WARN_ON(event->parent);

        jump_label_dec(&perf_swevent_enabled[event_id]);
        swevent_hlist_put(event);
}

static int perf_swevent_init(struct perf_event *event)
{
        u64 event_id = event->attr.config;

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

        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(event);
                if (err)
                        return err;

                jump_label_inc(&perf_swevent_enabled[event_id]);
                event->destroy = sw_perf_event_destroy;
        }

        return 0;
}

static struct pmu perf_swevent = {
        .task_ctx_nr    = perf_sw_context,

        .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,
};

#ifdef CONFIG_EVENT_TRACING

static int perf_tp_filter_match(struct perf_event *event,
                                struct perf_sample_data *data)
{
        void *record = data->raw->data;

        /* only top level events have filters set */
        if (event->parent)
                event = event->parent;

        if (likely(!event->filter) || filter_match_preds(event->filter, record))
                return 1;
        return 0;
}

static int perf_tp_event_match(struct perf_event *event,
                                struct perf_sample_data *data,
                                struct pt_regs *regs)
{
        if (event->hw.state & PERF_HES_STOPPED)
                return 0;
        /*
         * All tracepoints are from kernel-space.
         */
        if (event->attr.exclude_kernel)
                return 0;

        if (!perf_tp_filter_match(event, data))
                return 0;

        return 1;
}

void perf_tp_event(u64 addr, u64 count, void *record, int entry_size,
                   struct pt_regs *regs, struct hlist_head *head, int rctx)
{
        struct perf_sample_data data;
        struct perf_event *event;
        struct hlist_node *node;

        struct perf_raw_record raw = {
                .size = entry_size,
                .data = record,
        };

        perf_sample_data_init(&data, addr);
        data.raw = &raw;

        hlist_for_each_entry_rcu(event, node, head, hlist_entry) {
                if (perf_tp_event_match(event, &data, regs))
                        perf_swevent_event(event, count, &data, regs);
        }

        perf_swevent_put_recursion_context(rctx);
}
EXPORT_SYMBOL_GPL(perf_tp_event);

static void tp_perf_event_destroy(struct perf_event *event)
{
        perf_trace_destroy(event);
}

static int perf_tp_event_init(struct perf_event *event)
{
        int err;

        if (event->attr.type != PERF_TYPE_TRACEPOINT)
                return -ENOENT;

        err = perf_trace_init(event);
        if (err)
                return err;

        event->destroy = tp_perf_event_destroy;

        return 0;
}

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,
};

static inline void perf_tp_register(void)
{
        perf_pmu_register(&perf_tracepoint, "tracepoint", PERF_TYPE_TRACEPOINT);
}

static int perf_event_set_filter(struct perf_event *event, void __user *arg)
{
        char *filter_str;
        int ret;

        if (event->attr.type != PERF_TYPE_TRACEPOINT)
                return -EINVAL;

        filter_str = strndup_user(arg, PAGE_SIZE);
        if (IS_ERR(filter_str))
                return PTR_ERR(filter_str);

        ret = ftrace_profile_set_filter(event, event->attr.config, filter_str);

        kfree(filter_str);
        return ret;
}

static void perf_event_free_filter(struct perf_event *event)
{
        ftrace_profile_free_filter(event);
}

#else

static inline void perf_tp_register(void)
{
}

static int perf_event_set_filter(struct perf_event *event, void __user *arg)
{
        return -ENOENT;
}

static void perf_event_free_filter(struct perf_event *event)
{
}

#endif /* CONFIG_EVENT_TRACING */

#ifdef CONFIG_HAVE_HW_BREAKPOINT
void perf_bp_event(struct perf_event *bp, void *data)
{
        struct perf_sample_data sample;
        struct pt_regs *regs = data;

        perf_sample_data_init(&sample, bp->attr.bp_addr);

        if (!bp->hw.state && !perf_exclude_event(bp, regs))
                perf_swevent_event(bp, 1, &sample, regs);
}
#endif

/*
 * hrtimer based swevent callback
 */

static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer)
{
        enum hrtimer_restart ret = HRTIMER_RESTART;
        struct perf_sample_data data;
        struct pt_regs *regs;
        struct perf_event *event;
        u64 period;

        event = container_of(hrtimer, struct perf_event, hw.hrtimer);

        if (event->state != PERF_EVENT_STATE_ACTIVE)
                return HRTIMER_NORESTART;

        event->pmu->read(event);

        perf_sample_data_init(&data, 0);
        data.period = event->hw.last_period;
        regs = get_irq_regs();

        if (regs && !perf_exclude_event(event, regs)) {
                if (!(event->attr.exclude_idle && current->pid == 0))
                        if (perf_event_overflow(event, &data, regs))
                                ret = HRTIMER_NORESTART;
        }

        period = max_t(u64, 10000, event->hw.sample_period);
        hrtimer_forward_now(hrtimer, ns_to_ktime(period));

        return ret;
}

static void perf_swevent_start_hrtimer(struct perf_event *event)
{
        struct hw_perf_event *hwc = &event->hw;
        s64 period;

        if (!is_sampling_event(event))
                return;

        period = local64_read(&hwc->period_left);
        if (period) {
                if (period < 0)
                        period = 10000;

                local64_set(&hwc->period_left, 0);
        } else {
                period = max_t(u64, 10000, hwc->sample_period);
        }
        __hrtimer_start_range_ns(&hwc->hrtimer,
                                ns_to_ktime(period), 0,
                                HRTIMER_MODE_REL_PINNED, 0);
}

static void perf_swevent_cancel_hrtimer(struct perf_event *event)
{
        struct hw_perf_event *hwc = &event->hw;

        if (is_sampling_event(event)) {
                ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer);
                local64_set(&hwc->period_left, ktime_to_ns(remaining));

                hrtimer_cancel(&hwc->hrtimer);
        }
}

static void perf_swevent_init_hrtimer(struct perf_event *event)
{
        struct hw_perf_event *hwc = &event->hw;

        if (!is_sampling_event(event))
                return;

        hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
        hwc->hrtimer.function = perf_swevent_hrtimer;

        /*
         * Since hrtimers have a fixed rate, we can do a static freq->period
         * mapping and avoid the whole period adjust feedback stuff.
         */
        if (event->attr.freq) {
                long freq = event->attr.sample_freq;

                event->attr.sample_period = NSEC_PER_SEC / freq;
                hwc->sample_period = event->attr.sample_period;
                local64_set(&hwc->period_left, hwc->sample_period);
                event->attr.freq = 0;
        }
}

/*
 * Software event: cpu wall time clock
 */

static void cpu_clock_event_update(struct perf_event *event)
{
        s64 prev;
        u64 now;

        now = local_clock();
        prev = local64_xchg(&event->hw.prev_count, now);
        local64_add(now - prev, &event->count);
}

static void cpu_clock_event_start(struct perf_event *event, int flags)
{
        local64_set(&event->hw.prev_count, local_clock());
        perf_swevent_start_hrtimer(event);
}

static void cpu_clock_event_stop(struct perf_event *event, int flags)
{
        perf_swevent_cancel_hrtimer(event);
        cpu_clock_event_update(event);
}

static int cpu_clock_event_add(struct perf_event *event, int flags)
{
        if (flags & PERF_EF_START)
                cpu_clock_event_start(event, flags);

        return 0;
}

static void cpu_clock_event_del(struct perf_event *event, int flags)
{
        cpu_clock_event_stop(event, flags);
}

static void cpu_clock_event_read(struct perf_event *event)
{
        cpu_clock_event_update(event);
}

static int cpu_clock_event_init(struct perf_event *event)
{
        if (event->attr.type != PERF_TYPE_SOFTWARE)
                return -ENOENT;

        if (event->attr.config != PERF_COUNT_SW_CPU_CLOCK)
                return -ENOENT;

        perf_swevent_init_hrtimer(event);

        return 0;
}

static struct pmu perf_cpu_clock = {
        .task_ctx_nr    = perf_sw_context,

        .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,
};

/*
 * Software event: task time clock
 */

static void task_clock_event_update(struct perf_event *event, u64 now)
{
        u64 prev;
        s64 delta;

        prev = local64_xchg(&event->hw.prev_count, now);
        delta = now - prev;
        local64_add(delta, &event->count);
}

static void task_clock_event_start(struct perf_event *event, int flags)
{
        local64_set(&event->hw.prev_count, event->ctx->time);
        perf_swevent_start_hrtimer(event);
}

static void task_clock_event_stop(struct perf_event *event, int flags)
{
        perf_swevent_cancel_hrtimer(event);
        task_clock_event_update(event, event->ctx->time);
}

static int task_clock_event_add(struct perf_event *event, int flags)
{
        if (flags & PERF_EF_START)
                task_clock_event_start(event, flags);

        return 0;
}

static void task_clock_event_del(struct perf_event *event, int flags)
{
        task_clock_event_stop(event, PERF_EF_UPDATE);
}

static void task_clock_event_read(struct perf_event *event)
{
        u64 now = perf_clock();
        u64 delta = now - event->ctx->timestamp;
        u64 time = event->ctx->time + delta;

        task_clock_event_update(event, time);
}

static int task_clock_event_init(struct perf_event *event)
{
        if (event->attr.type != PERF_TYPE_SOFTWARE)
                return -ENOENT;

        if (event->attr.config != PERF_COUNT_SW_TASK_CLOCK)
                return -ENOENT;

        perf_swevent_init_hrtimer(event);

        return 0;
}

static struct pmu perf_task_clock = {
        .task_ctx_nr    = perf_sw_context,

        .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,
};

static void perf_pmu_nop_void(struct pmu *pmu)
{
}

static int perf_pmu_nop_int(struct pmu *pmu)
{
        return 0;
}

static void perf_pmu_start_txn(struct pmu *pmu)
{
        perf_pmu_disable(pmu);
}

static int perf_pmu_commit_txn(struct pmu *pmu)
{
        perf_pmu_enable(pmu);
        return 0;
}

static void perf_pmu_cancel_txn(struct pmu *pmu)
{
        perf_pmu_enable(pmu);
}

/*
 * Ensures all contexts with the same task_ctx_nr have the same
 * pmu_cpu_context too.
 */
static void *find_pmu_context(int ctxn)
{
        struct pmu *pmu;

        if (ctxn < 0)
                return NULL;

        list_for_each_entry(pmu, &pmus, entry) {
                if (pmu->task_ctx_nr == ctxn)
                        return pmu->pmu_cpu_context;
        }

        return NULL;
}

static void update_pmu_context(struct pmu *pmu, struct pmu *old_pmu)
{
        int cpu;

        for_each_possible_cpu(cpu) {
                struct perf_cpu_context *cpuctx;

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

                if (cpuctx->unique_pmu == old_pmu)
                        cpuctx->unique_pmu = pmu;
        }
}

static void free_pmu_context(struct pmu *pmu)
{
        struct pmu *i;

        mutex_lock(&pmus_lock);
        /*
         * Like a real lame refcount.
         */
        list_for_each_entry(i, &pmus, entry) {
                if (i->pmu_cpu_context == pmu->pmu_cpu_context) {
                        update_pmu_context(i, pmu);
                        goto out;
                }
        }

        free_percpu(pmu->pmu_cpu_context);
out:
        mutex_unlock(&pmus_lock);
}
static struct idr pmu_idr;

static ssize_t
type_show(struct device *dev, struct device_attribute *attr, char *page)
{
        struct pmu *pmu = dev_get_drvdata(dev);

        return snprintf(page, PAGE_SIZE-1, "%d\n", pmu->type);
}

static struct device_attribute pmu_dev_attrs[] = {
       __ATTR_RO(type),
       __ATTR_NULL,
};

static int pmu_bus_running;
static struct bus_type pmu_bus = {
        .name           = "event_source",
        .dev_attrs      = pmu_dev_attrs,
};

static void pmu_dev_release(struct device *dev)
{
        kfree(dev);
}

static int pmu_dev_alloc(struct pmu *pmu)
{
        int ret = -ENOMEM;

        pmu->dev = kzalloc(sizeof(struct device), GFP_KERNEL);
        if (!pmu->dev)
                goto out;

        device_initialize(pmu->dev);
        ret = dev_set_name(pmu->dev, "%s", pmu->name);
        if (ret)
                goto free_dev;

        dev_set_drvdata(pmu->dev, pmu);
        pmu->dev->bus = &pmu_bus;
        pmu->dev->release = pmu_dev_release;
        ret = device_add(pmu->dev);
        if (ret)
                goto free_dev;

out:
        return ret;

free_dev:
        put_device(pmu->dev);
        goto out;
}

static struct lock_class_key cpuctx_mutex;
static struct lock_class_key cpuctx_lock;

int perf_pmu_register(struct pmu *pmu, char *name, int type)
{
        int cpu, ret;

        mutex_lock(&pmus_lock);
        ret = -ENOMEM;
        pmu->pmu_disable_count = alloc_percpu(int);
        if (!pmu->pmu_disable_count)
                goto unlock;

        pmu->type = -1;
        if (!name)
                goto skip_type;
        pmu->name = name;

        if (type < 0) {
                int err = idr_pre_get(&pmu_idr, GFP_KERNEL);
                if (!err)
                        goto free_pdc;

                err = idr_get_new_above(&pmu_idr, pmu, PERF_TYPE_MAX, &type);
                if (err) {
                        ret = err;
                        goto free_pdc;
                }
        }
        pmu->type = type;

        if (pmu_bus_running) {
                ret = pmu_dev_alloc(pmu);
                if (ret)
                        goto free_idr;
        }

skip_type:
        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.type = cpu_context;
                cpuctx->ctx.pmu = pmu;
                cpuctx->jiffies_interval = 1;
                INIT_LIST_HEAD(&cpuctx->rotation_list);
                cpuctx->unique_pmu = pmu;
        }

got_cpu_context:
        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_void;
                        pmu->commit_txn = perf_pmu_nop_int;
                        pmu->cancel_txn = perf_pmu_nop_void;
                }
        }

        if (!pmu->pmu_enable) {
                pmu->pmu_enable  = perf_pmu_nop_void;
                pmu->pmu_disable = perf_pmu_nop_void;
        }

        list_add_rcu(&pmu->entry, &pmus);
        ret = 0;
unlock:
        mutex_unlock(&pmus_lock);

        return ret;

free_dev:
        device_del(pmu->dev);
        put_device(pmu->dev);

free_idr:
        if (pmu->type >= PERF_TYPE_MAX)
                idr_remove(&pmu_idr, pmu->type);

free_pdc:
        free_percpu(pmu->pmu_disable_count);
        goto unlock;
}

void perf_pmu_unregister(struct pmu *pmu)
{
        mutex_lock(&pmus_lock);
        list_del_rcu(&pmu->entry);
        mutex_unlock(&pmus_lock);

        /*
         * We dereference the pmu list under both SRCU and regular RCU, so
         * synchronize against both of those.
         */
        synchronize_srcu(&pmus_srcu);
        synchronize_rcu();

        free_percpu(pmu->pmu_disable_count);
        if (pmu->type >= PERF_TYPE_MAX)
                idr_remove(&pmu_idr, pmu->type);
        device_del(pmu->dev);
        put_device(pmu->dev);
        free_pmu_context(pmu);
}

struct pmu *perf_init_event(struct perf_event *event)
{
        struct pmu *pmu = NULL;
        int idx;
        int ret;

        idx = srcu_read_lock(&pmus_srcu);

        rcu_read_lock();
        pmu = idr_find(&pmu_idr, event->attr.type);
        rcu_read_unlock();
        if (pmu) {
                event->pmu = pmu;
                ret = pmu->event_init(event);
                if (ret)
                        pmu = ERR_PTR(ret);
                goto unlock;
        }

        list_for_each_entry_rcu(pmu, &pmus, entry) {
                event->pmu = pmu;
                ret = pmu->event_init(event);
                if (!ret)
                        goto unlock;

                if (ret != -ENOENT) {
                        pmu = ERR_PTR(ret);
                        goto unlock;
                }
        }
        pmu = ERR_PTR(-ENOENT);
unlock:
        srcu_read_unlock(&pmus_srcu, idx);

        return pmu;
}

/*
 * Allocate and initialize a event structure
 */
static struct perf_event *
perf_event_alloc(struct perf_event_attr *attr, int cpu,
                 struct task_struct *task,
                 struct perf_event *group_leader,
                 struct perf_event *parent_event,
                 perf_overflow_handler_t overflow_handler,
                 void *context)
{
        struct pmu *pmu;
        struct perf_event *event;
        struct hw_perf_event *hwc;
        long err;

        if ((unsigned)cpu >= nr_cpu_ids) {
                if (!task || cpu != -1)
                        return ERR_PTR(-EINVAL);
        }

        event = kzalloc(sizeof(*event), GFP_KERNEL);
        if (!event)
                return ERR_PTR(-ENOMEM);

        /*
         * Single events are their own group leaders, with an
         * empty sibling list:
         */
        if (!group_leader)
                group_leader = event;

        mutex_init(&event->child_mutex);
        INIT_LIST_HEAD(&event->child_list);

        INIT_LIST_HEAD(&event->group_entry);
        INIT_LIST_HEAD(&event->event_entry);
        INIT_LIST_HEAD(&event->sibling_list);
        INIT_LIST_HEAD(&event->rb_entry);

        init_waitqueue_head(&event->waitq);
        init_irq_work(&event->pending, perf_pending_event);

        mutex_init(&event->mmap_mutex);

        atomic_long_set(&event->refcount, 1);
        event->cpu              = cpu;
        event->attr             = *attr;
        event->group_leader     = group_leader;
        event->pmu              = NULL;
        event->oncpu            = -1;

        event->parent           = parent_event;

        event->ns               = get_pid_ns(current->nsproxy->pid_ns);
        event->id               = atomic64_inc_return(&perf_event_id);

        event->state            = PERF_EVENT_STATE_INACTIVE;

        if (task) {
                event->attach_state = PERF_ATTACH_TASK;
#ifdef CONFIG_HAVE_HW_BREAKPOINT
                /*
                 * hw_breakpoint is a bit difficult here..
                 */
                if (attr->type == PERF_TYPE_BREAKPOINT)
                        event->hw.bp_target = task;
#endif
        }

        if (!overflow_handler && parent_event) {
                overflow_handler = parent_event->overflow_handler;
                context = parent_event->overflow_handler_context;
        }

        event->overflow_handler = overflow_handler;
        event->overflow_handler_context = context;

        perf_event__state_init(event);

        pmu = NULL;

        hwc = &event->hw;
        hwc->sample_period = attr->sample_period;
        if (attr->freq && attr->sample_freq)
                hwc->sample_period = 1;
        hwc->last_period = hwc->sample_period;

        local64_set(&hwc->period_left, hwc->sample_period);

        /*
         * We currently do not support PERF_SAMPLE_READ on inherited events.
         * See perf_output_read().
         */
        if (attr->inherit && (attr->sample_type & PERF_SAMPLE_READ))
                goto done;

        pmu = perf_init_event(event);

done:
        err = 0;
        if (!pmu)
                err = -EINVAL;
        else if (IS_ERR(pmu))
                err = PTR_ERR(pmu);

        if (err) {
                if (event->ns)
                        put_pid_ns(event->ns);
                kfree(event);
                return ERR_PTR(err);
        }

        if (!event->parent) {
                if (event->attach_state & PERF_ATTACH_TASK)
                        jump_label_inc(&perf_sched_events);
                if (event->attr.mmap || event->attr.mmap_data)
                        atomic_inc(&nr_mmap_events);
                if (event->attr.comm)
                        atomic_inc(&nr_comm_events);
                if (event->attr.task)
                        atomic_inc(&nr_task_events);
                if (event->attr.sample_type & PERF_SAMPLE_CALLCHAIN) {
                        err = get_callchain_buffers();
                        if (err) {
                                free_event(event);
                                return ERR_PTR(err);
                        }
                }
        }

        return event;
}

static int perf_copy_attr(struct perf_event_attr __user *uattr,
                          struct perf_event_attr *attr)
{
        u32 size;
        int ret;

        if (!access_ok(VERIFY_WRITE, uattr, PERF_ATTR_SIZE_VER0))
                return -EFAULT;

        /*
         * zero the full structure, so that a short copy will be nice.
         */
        memset(attr, 0, sizeof(*attr));

        ret = get_user(size, &uattr->size);
        if (ret)
                return ret;

        if (size > PAGE_SIZE)   /* silly large */
                goto err_size;

        if (!size)              /* abi compat */
                size = PERF_ATTR_SIZE_VER0;

        if (size < PERF_ATTR_SIZE_VER0)
                goto err_size;

        /*
         * If we're handed a bigger struct than we know of,
         * ensure all the unknown bits are 0 - i.e. new
         * user-space does not rely on any kernel feature
         * extensions we dont know about yet.
         */
        if (size > sizeof(*attr)) {
                unsigned char __user *addr;
                unsigned char __user *end;
                unsigned char val;

                addr = (void __user *)uattr + sizeof(*attr);
                end  = (void __user *)uattr + size;

                for (; addr < end; addr++) {
                        ret = get_user(val, addr);
                        if (ret)
                                return ret;
                        if (val)
                                goto err_size;
                }
                size = sizeof(*attr);
        }

        ret = copy_from_user(attr, uattr, size);
        if (ret)
                return -EFAULT;

        if (attr->__reserved_1)
                return -EINVAL;

        if (attr->sample_type & ~(PERF_SAMPLE_MAX-1))
                return -EINVAL;

        if (attr->read_format & ~(PERF_FORMAT_MAX-1))
                return -EINVAL;

out:
        return ret;

err_size:
        put_user(sizeof(*attr), &uattr->size);
        ret = -E2BIG;
        goto out;
}

static int
perf_event_set_output(struct perf_event *event, struct perf_event *output_event)
{
        struct ring_buffer *rb = NULL, *old_rb = NULL;
        int ret = -EINVAL;

        if (!output_event)
                goto set;

        /* don't allow circular references */
        if (event == output_event)
                goto out;

        /*
         * Don't allow cross-cpu buffers
         */
        if (output_event->cpu != event->cpu)
                goto out;

        /*
         * If its not a per-cpu rb, it must be the same task.
         */
        if (output_event->cpu == -1 && output_event->ctx != event->ctx)
                goto out;

set:
        mutex_lock(&event->mmap_mutex);
        /* Can't redirect output if we've got an active mmap() */
        if (atomic_read(&event->mmap_count))
                goto unlock;

        old_rb = event->rb;

        if (output_event) {
                /* get the rb we want to redirect to */
                rb = ring_buffer_get(output_event);
                if (!rb)
                        goto unlock;
        }

        if (old_rb)
                ring_buffer_detach(event, old_rb);

        if (rb)
                ring_buffer_attach(event, rb);

        rcu_assign_pointer(event->rb, rb);

        if (old_rb) {
                ring_buffer_put(old_rb);
                /*
                 * Since we detached before setting the new rb, so that we
                 * could attach the new rb, we could have missed a wakeup.
                 * Provide it now.
                 */
                wake_up_all(&event->waitq);
        }

        ret = 0;
unlock:
        mutex_unlock(&event->mmap_mutex);

out:
        return ret;
}

static void mutex_lock_double(struct mutex *a, struct mutex *b)
{
        if (b < a)
                swap(a, b);

        mutex_lock(a);
        mutex_lock_nested(b, SINGLE_DEPTH_NESTING);
}

/*
 * Variation on perf_event_ctx_lock_nested(), except we take two context
 * mutexes.
 */
static struct perf_event_context *
__perf_event_ctx_lock_double(struct perf_event *group_leader,
                             struct perf_event_context *ctx)
{
        struct perf_event_context *gctx;

again:
        rcu_read_lock();
        gctx = ACCESS_ONCE(group_leader->ctx);
        if (!atomic_inc_not_zero(&gctx->refcount)) {
                rcu_read_unlock();
                goto again;
        }
        rcu_read_unlock();

        mutex_lock_double(&gctx->mutex, &ctx->mutex);

        if (group_leader->ctx != gctx) {
                mutex_unlock(&ctx->mutex);
                mutex_unlock(&gctx->mutex);
                put_ctx(gctx);
                goto again;
        }

        return gctx;
}

/**
 * sys_perf_event_open - open a performance event, associate it to a task/cpu
 *
 * @attr_uptr:  event_id type attributes for monitoring/sampling
 * @pid:                target pid
 * @cpu:                target cpu
 * @group_fd:           group leader event fd
 */
SYSCALL_DEFINE5(perf_event_open,
                struct perf_event_attr __user *, attr_uptr,
                pid_t, pid, int, cpu, int, group_fd, unsigned long, flags)
{
        struct perf_event *group_leader = NULL, *output_event = NULL;
        struct perf_event *event, *sibling;
        struct perf_event_attr attr;
        struct perf_event_context *ctx, *uninitialized_var(gctx);
        struct file *event_file = NULL;
        struct file *group_file = NULL;
        struct task_struct *task = NULL;
        struct pmu *pmu;
        int event_fd;
        int move_group = 0;
        int fput_needed = 0;
        int err;

        /* for future expandability... */
        if (flags & ~PERF_FLAG_ALL)
                return -EINVAL;

        err = perf_copy_attr(attr_uptr, &attr);
        if (err)
                return err;

        if (!attr.exclude_kernel) {
                if (perf_paranoid_kernel() && !capable(CAP_SYS_ADMIN))
                        return -EACCES;
        }

        if (attr.freq) {
                if (attr.sample_freq > sysctl_perf_event_sample_rate)
                        return -EINVAL;
        } else {
                if (attr.sample_period & (1ULL << 63))
                        return -EINVAL;
        }

        /*
         * In cgroup mode, the pid argument is used to pass the fd
         * opened to the cgroup directory in cgroupfs. The cpu argument
         * designates the cpu on which to monitor threads from that
         * cgroup.
         */
        if ((flags & PERF_FLAG_PID_CGROUP) && (pid == -1 || cpu == -1))
                return -EINVAL;

        event_fd = get_unused_fd_flags(O_RDWR);
        if (event_fd < 0)
                return event_fd;

        if (group_fd != -1) {
                group_file = perf_fget_light(group_fd, &fput_needed);
                if (IS_ERR(group_file)) {
                        err = PTR_ERR(group_file);
                        goto err_fd;
                }
                group_leader = group_file->private_data;
                if (flags & PERF_FLAG_FD_OUTPUT)
                        output_event = group_leader;
                if (flags & PERF_FLAG_FD_NO_GROUP)
                        group_leader = NULL;
        }

        if (pid != -1 && !(flags & PERF_FLAG_PID_CGROUP)) {
                task = find_lively_task_by_vpid(pid);
                if (IS_ERR(task)) {
                        err = PTR_ERR(task);
                        goto err_group_fd;
                }
        }

        event = perf_event_alloc(&attr, cpu, task, group_leader, NULL,
                                 NULL, NULL);
        if (IS_ERR(event)) {
                err = PTR_ERR(event);
                goto err_task;
        }

        if (flags & PERF_FLAG_PID_CGROUP) {
                err = perf_cgroup_connect(pid, event, &attr, group_leader);
                if (err)
                        goto err_alloc;
                /*
                 * one more event:
                 * - that has cgroup constraint on event->cpu
                 * - that may need work on context switch
                 */
                atomic_inc(&per_cpu(perf_cgroup_events, event->cpu));
                jump_label_inc(&perf_sched_events);
        }

        /*
         * Special case software events and allow them to be part of
         * any hardware group.
         */
        pmu = event->pmu;

        if (group_leader &&
            (is_software_event(event) != is_software_event(group_leader))) {
                if (is_software_event(event)) {
                        /*
                         * If event and group_leader are not both a software
                         * event, and event is, then group leader is not.
                         *
                         * Allow the addition of software events to !software
                         * groups, this is safe because software events never
                         * fail to schedule.
                         */
                        pmu = group_leader->pmu;
                } else if (is_software_event(group_leader) &&
                           (group_leader->group_flags & PERF_GROUP_SOFTWARE)) {
                        /*
                         * In case the group is a pure software group, and we
                         * try to add a hardware event, move the whole group to
                         * the hardware context.
                         */
                        move_group = 1;
                }
        }

        /*
         * Get the target context (task or percpu):
         */
        ctx = find_get_context(pmu, task, cpu);
        if (IS_ERR(ctx)) {
                err = PTR_ERR(ctx);
                goto err_alloc;
        }

        if (task) {
                put_task_struct(task);
                task = NULL;
        }

        /*
         * Look up the group leader (we will attach this event to it):
         */
        if (group_leader) {
                err = -EINVAL;

                /*
                 * Do not allow a recursive hierarchy (this new sibling
                 * becoming part of another group-sibling):
                 */
                if (group_leader->group_leader != group_leader)
                        goto err_context;
                /*
                 * Do not allow to attach to a group in a different
                 * task or CPU context:
                 */
                if (move_group) {
                        if (group_leader->ctx->type != ctx->type)
                                goto err_context;
                } else {
                        if (group_leader->ctx != ctx)
                                goto err_context;
                }

                /*
                 * Only a group leader can be exclusive or pinned
                 */
                if (attr.exclusive || attr.pinned)
                        goto err_context;
        }

        if (output_event) {
                err = perf_event_set_output(event, output_event);
                if (err)
                        goto err_context;
        }

        event_file = anon_inode_getfile("[perf_event]", &perf_fops, event, O_RDWR);
        if (IS_ERR(event_file)) {
                err = PTR_ERR(event_file);
                goto err_context;
        }

        if (move_group) {
                gctx = __perf_event_ctx_lock_double(group_leader, ctx);

                /*
                 * Check if we raced against another sys_perf_event_open() call
                 * moving the software group underneath us.
                 */
                if (!(group_leader->group_flags & PERF_GROUP_SOFTWARE)) {
                        /*
                         * If someone moved the group out from under us, check
                         * if this new event wound up on the same ctx, if so
                         * its the regular !move_group case, otherwise fail.
                         */
                        if (gctx != ctx) {
                                err = -EINVAL;
                                goto err_locked;
                        } else {
                                perf_event_ctx_unlock(group_leader, gctx);
                                move_group = 0;
                        }
                }

                /*
                 * See perf_event_ctx_lock() for comments on the details
                 * of swizzling perf_event::ctx.
                 */
                perf_remove_from_context(group_leader, false);

                /*
                 * Removing from the context ends up with disabled
                 * event. What we want here is event in the initial
                 * startup state, ready to be add into new context.
                 */
                perf_event__state_init(group_leader);
                list_for_each_entry(sibling, &group_leader->sibling_list,
                                    group_entry) {
                        perf_remove_from_context(sibling, false);
                        perf_event__state_init(sibling);
                        put_ctx(gctx);
                }
        } else {
                mutex_lock(&ctx->mutex);
        }

        WARN_ON_ONCE(ctx->parent_ctx);

        if (move_group) {
                /*
                 * Wait for everybody to stop referencing the events through
                 * the old lists, before installing it on new lists.
                 */
                synchronize_rcu();

                perf_install_in_context(ctx, group_leader, cpu);
                get_ctx(ctx);
                list_for_each_entry(sibling, &group_leader->sibling_list,
                                    group_entry) {
                        perf_install_in_context(ctx, sibling, cpu);
                        get_ctx(ctx);
                }
        }

        perf_install_in_context(ctx, event, cpu);
        ++ctx->generation;
        perf_unpin_context(ctx);

        if (move_group) {
                perf_event_ctx_unlock(group_leader, gctx);
                put_ctx(gctx);
        }
        mutex_unlock(&ctx->mutex);

        event->owner = current;

        mutex_lock(&current->perf_event_mutex);
        list_add_tail(&event->owner_entry, &current->perf_event_list);
        mutex_unlock(&current->perf_event_mutex);

        /*
         * Precalculate sample_data sizes
         */
        perf_event__header_size(event);
        perf_event__id_header_size(event);

        /*
         * Drop the reference on the group_event after placing the
         * new event on the sibling_list. This ensures destruction
         * of the group leader will find the pointer to itself in
         * perf_group_detach().
         */
        fput_light(group_file, fput_needed);
        fd_install(event_fd, event_file);
        return event_fd;

err_locked:
        if (move_group)
                perf_event_ctx_unlock(group_leader, gctx);
        mutex_unlock(&ctx->mutex);
        fput(event_file);
err_context:
        perf_unpin_context(ctx);
        put_ctx(ctx);
err_alloc:
        /*
         * If event_file is set, the fput() above will have called ->release()
         * and that will take care of freeing the event.
         */
        if (!event_file)
                free_event(event);
err_task:
        if (task)
                put_task_struct(task);
err_group_fd:
        fput_light(group_file, fput_needed);
err_fd:
        put_unused_fd(event_fd);
        return err;
}

/**
 * perf_event_create_kernel_counter
 *
 * @attr: attributes of the counter to create
 * @cpu: cpu in which the counter is bound
 * @task: task to profile (NULL for percpu)
 */
struct perf_event *
perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
                                 struct task_struct *task,
                                 perf_overflow_handler_t overflow_handler,
                                 void *context)
{
        struct perf_event_context *ctx;
        struct perf_event *event;
        int err;

        /*
         * Get the target context (task or percpu):
         */

        event = perf_event_alloc(attr, cpu, task, NULL, NULL,
                                 overflow_handler, context);
        if (IS_ERR(event)) {
                err = PTR_ERR(event);
                goto err;
        }

        ctx = find_get_context(event->pmu, task, cpu);
        if (IS_ERR(ctx)) {
                err = PTR_ERR(ctx);
                goto err_free;
        }

        WARN_ON_ONCE(ctx->parent_ctx);
        mutex_lock(&ctx->mutex);
        perf_install_in_context(ctx, event, cpu);
        ++ctx->generation;
        perf_unpin_context(ctx);
        mutex_unlock(&ctx->mutex);

        return event;

err_free:
        free_event(event);
err:
        return ERR_PTR(err);
}
EXPORT_SYMBOL_GPL(perf_event_create_kernel_counter);

static void sync_child_event(struct perf_event *child_event,
                               struct task_struct *child)
{
        struct perf_event *parent_event = child_event->parent;
        u64 child_val;

        if (child_event->attr.inherit_stat)
                perf_event_read_event(child_event, child);

        child_val = perf_event_count(child_event);

        /*
         * Add back the child's count to the parent's count:
         */
        atomic64_add(child_val, &parent_event->child_count);
        atomic64_add(child_event->total_time_enabled,
                     &parent_event->child_total_time_enabled);
        atomic64_add(child_event->total_time_running,
                     &parent_event->child_total_time_running);

        /*
         * Remove this event from the parent's list
         */
        WARN_ON_ONCE(parent_event->ctx->parent_ctx);
        mutex_lock(&parent_event->child_mutex);
        list_del_init(&child_event->child_list);
        mutex_unlock(&parent_event->child_mutex);

        /*
         * Release the parent event, if this was the last
         * reference to it.
         */
        put_event(parent_event);
}

static void
__perf_event_exit_task(struct perf_event *child_event,
                         struct perf_event_context *child_ctx,
                         struct task_struct *child)
{
        perf_remove_from_context(child_event, !!child_event->parent);

        /*
         * It can happen that the parent exits first, and has events
         * that are still around due to the child reference. These
         * events need to be zapped.
         */
        if (child_event->parent) {
                sync_child_event(child_event, child);
                free_event(child_event);
        }
}

static void perf_event_exit_task_context(struct task_struct *child, int ctxn)
{
        struct perf_event *child_event, *tmp;
        struct perf_event_context *child_ctx;
        unsigned long flags;

        if (likely(!child->perf_event_ctxp[ctxn])) {
                perf_event_task(child, NULL, 0);
                return;
        }

        local_irq_save(flags);
        /*
         * We can't reschedule here because interrupts are disabled,
         * and either child is current or it is a task that can't be
         * scheduled, so we are now safe from rescheduling changing
         * our context.
         */
        child_ctx = rcu_dereference_raw(child->perf_event_ctxp[ctxn]);

        /*
         * Take the context lock here so that if find_get_context is
         * reading child->perf_event_ctxp, we wait until it has
         * incremented the context's refcount before we do put_ctx below.
         */
        raw_spin_lock(&child_ctx->lock);
        task_ctx_sched_out(child_ctx);
        child->perf_event_ctxp[ctxn] = NULL;
        /*
         * If this context is a clone; unclone it so it can't get
         * swapped to another process while we're removing all
         * the events from it.
         */
        unclone_ctx(child_ctx);
        update_context_time(child_ctx);
        raw_spin_unlock_irqrestore(&child_ctx->lock, flags);

        /*
         * Report the task dead after unscheduling the events so that we
         * won't get any samples after PERF_RECORD_EXIT. We can however still
         * get a few PERF_RECORD_READ events.
         */
        perf_event_task(child, child_ctx, 0);

        /*
         * We can recurse on the same lock type through:
         *
         *   __perf_event_exit_task()
         *     sync_child_event()
         *       put_event()
         *         mutex_lock(&ctx->mutex)
         *
         * But since its the parent context it won't be the same instance.
         */
        mutex_lock(&child_ctx->mutex);

again:
        list_for_each_entry_safe(child_event, tmp, &child_ctx->pinned_groups,
                                 group_entry)
                __perf_event_exit_task(child_event, child_ctx, child);

        list_for_each_entry_safe(child_event, tmp, &child_ctx->flexible_groups,
                                 group_entry)
                __perf_event_exit_task(child_event, child_ctx, child);

        /*
         * If the last event was a group event, it will have appended all
         * its siblings to the list, but we obtained 'tmp' before that which
         * will still point to the list head terminating the iteration.
         */
        if (!list_empty(&child_ctx->pinned_groups) ||
            !list_empty(&child_ctx->flexible_groups))
                goto again;

        mutex_unlock(&child_ctx->mutex);

        put_ctx(child_ctx);
}

/*
 * When a child task exits, feed back event values to parent events.
 */
void perf_event_exit_task(struct task_struct *child)
{
        struct perf_event *event, *tmp;
        int ctxn;

        mutex_lock(&child->perf_event_mutex);
        list_for_each_entry_safe(event, tmp, &child->perf_event_list,
                                 owner_entry) {
                list_del_init(&event->owner_entry);

                /*
                 * Ensure the list deletion is visible before we clear
                 * the owner, closes a race against perf_release() where
                 * we need to serialize on the owner->perf_event_mutex.
                 */
                smp_wmb();
                event->owner = NULL;
        }
        mutex_unlock(&child->perf_event_mutex);

        for_each_task_context_nr(ctxn)
                perf_event_exit_task_context(child, ctxn);
}

static void perf_free_event(struct perf_event *event,
                            struct perf_event_context *ctx)
{
        struct perf_event *parent = event->parent;

        if (WARN_ON_ONCE(!parent))
                return;

        mutex_lock(&parent->child_mutex);
        list_del_init(&event->child_list);
        mutex_unlock(&parent->child_mutex);

        put_event(parent);

        perf_group_detach(event);
        list_del_event(event, ctx);
        free_event(event);
}

/*
 * free an unexposed, unused context as created by inheritance by
 * perf_event_init_task below, used by fork() in case of fail.
 */
void perf_event_free_task(struct task_struct *task)
{
        struct perf_event_context *ctx;
        struct perf_event *event, *tmp;
        int ctxn;

        for_each_task_context_nr(ctxn) {
                ctx = task->perf_event_ctxp[ctxn];
                if (!ctx)
                        continue;

                mutex_lock(&ctx->mutex);
again:
                list_for_each_entry_safe(event, tmp, &ctx->pinned_groups,
                                group_entry)
                        perf_free_event(event, ctx);

                list_for_each_entry_safe(event, tmp, &ctx->flexible_groups,
                                group_entry)
                        perf_free_event(event, ctx);

                if (!list_empty(&ctx->pinned_groups) ||
                                !list_empty(&ctx->flexible_groups))
                        goto again;

                mutex_unlock(&ctx->mutex);

                put_ctx(ctx);
        }
}

void perf_event_delayed_put(struct task_struct *task)
{
        int ctxn;

        for_each_task_context_nr(ctxn)
                WARN_ON_ONCE(task->perf_event_ctxp[ctxn]);
}

/*
 * inherit a event from parent task to child task:
 */
static struct perf_event *
inherit_event(struct perf_event *parent_event,
              struct task_struct *parent,
              struct perf_event_context *parent_ctx,
              struct task_struct *child,
              struct perf_event *group_leader,
              struct perf_event_context *child_ctx)
{
        struct perf_event *child_event;
        unsigned long flags;

        /*
         * Instead of creating recursive hierarchies of events,
         * we link inherited events back to the original parent,
         * which has a filp for sure, which we use as the reference
         * count:
         */
        if (parent_event->parent)
                parent_event = parent_event->parent;

        child_event = perf_event_alloc(&parent_event->attr,
                                           parent_event->cpu,
                                           child,
                                           group_leader, parent_event,
                                           NULL, NULL);
        if (IS_ERR(child_event))
                return child_event;

        if (!atomic_long_inc_not_zero(&parent_event->refcount)) {
                free_event(child_event);
                return NULL;
        }

        get_ctx(child_ctx);

        /*
         * Make the child state follow the state of the parent event,
         * not its attr.disabled bit.  We hold the parent's mutex,
         * so we won't race with perf_event_{en, dis}able_family.
         */
        if (parent_event->state >= PERF_EVENT_STATE_INACTIVE)
                child_event->state = PERF_EVENT_STATE_INACTIVE;
        else
                child_event->state = PERF_EVENT_STATE_OFF;

        if (parent_event->attr.freq) {
                u64 sample_period = parent_event->hw.sample_period;
                struct hw_perf_event *hwc = &child_event->hw;

                hwc->sample_period = sample_period;
                hwc->last_period   = sample_period;

                local64_set(&hwc->period_left, sample_period);
        }

        child_event->ctx = child_ctx;
        child_event->overflow_handler = parent_event->overflow_handler;
        child_event->overflow_handler_context
                = parent_event->overflow_handler_context;

        /*
         * Precalculate sample_data sizes
         */
        perf_event__header_size(child_event);
        perf_event__id_header_size(child_event);

        /*
         * Link it up in the child's context:
         */
        raw_spin_lock_irqsave(&child_ctx->lock, flags);
        add_event_to_ctx(child_event, child_ctx);
        raw_spin_unlock_irqrestore(&child_ctx->lock, flags);

        /*
         * Link this into the parent event's child list
         */
        WARN_ON_ONCE(parent_event->ctx->parent_ctx);
        mutex_lock(&parent_event->child_mutex);
        list_add_tail(&child_event->child_list, &parent_event->child_list);
        mutex_unlock(&parent_event->child_mutex);

        return child_event;
}

static int inherit_group(struct perf_event *parent_event,
              struct task_struct *parent,
              struct perf_event_context *parent_ctx,
              struct task_struct *child,
              struct perf_event_context *child_ctx)
{
        struct perf_event *leader;
        struct perf_event *sub;
        struct perf_event *child_ctr;

        leader = inherit_event(parent_event, parent, parent_ctx,
                                 child, NULL, child_ctx);
        if (IS_ERR(leader))
                return PTR_ERR(leader);
        list_for_each_entry(sub, &parent_event->sibling_list, group_entry) {
                child_ctr = inherit_event(sub, parent, parent_ctx,
                                            child, leader, child_ctx);
                if (IS_ERR(child_ctr))
                        return PTR_ERR(child_ctr);
        }
        return 0;
}

static int
inherit_task_group(struct perf_event *event, struct task_struct *parent,
                   struct perf_event_context *parent_ctx,
                   struct task_struct *child, int ctxn,
                   int *inherited_all)
{
        int ret;
        struct perf_event_context *child_ctx;

        if (!event->attr.inherit) {
                *inherited_all = 0;
                return 0;
        }

        child_ctx = child->perf_event_ctxp[ctxn];
        if (!child_ctx) {
                /*
                 * This is executed from the parent task context, so
                 * inherit events that have been marked for cloning.
                 * First allocate and initialize a context for the
                 * child.
                 */

                child_ctx = alloc_perf_context(parent_ctx->pmu, child);
                if (!child_ctx)
                        return -ENOMEM;

                child->perf_event_ctxp[ctxn] = child_ctx;
        }

        ret = inherit_group(event, parent, parent_ctx,
                            child, child_ctx);

        if (ret)
                *inherited_all = 0;

        return ret;
}

/*
 * Initialize the perf_event context in task_struct
 */
int perf_event_init_context(struct task_struct *child, int ctxn)
{
        struct perf_event_context *child_ctx, *parent_ctx;
        struct perf_event_context *cloned_ctx;
        struct perf_event *event;
        struct task_struct *parent = current;
        int inherited_all = 1;
        unsigned long flags;
        int ret = 0;

        if (likely(!parent->perf_event_ctxp[ctxn]))
                return 0;

        /*
         * If the parent's context is a clone, pin it so it won't get
         * swapped under us.
         */
        parent_ctx = perf_pin_task_context(parent, ctxn);

        /*
         * No need to check if parent_ctx != NULL here; since we saw
         * it non-NULL earlier, the only reason for it to become NULL
         * is if we exit, and since we're currently in the middle of
         * a fork we can't be exiting at the same time.
         */

        /*
         * Lock the parent list. No need to lock the child - not PID
         * hashed yet and not running, so nobody can access it.
         */
        mutex_lock(&parent_ctx->mutex);

        /*
         * We dont have to disable NMIs - we are only looking at
         * the list, not manipulating it:
         */
        list_for_each_entry(event, &parent_ctx->pinned_groups, group_entry) {
                ret = inherit_task_group(event, parent, parent_ctx,
                                         child, ctxn, &inherited_all);
                if (ret)
                        goto out_unlock;
        }

        /*
         * We can't hold ctx->lock when iterating the ->flexible_group list due
         * to allocations, but we need to prevent rotation because
         * rotate_ctx() will change the list from interrupt context.
         */
        raw_spin_lock_irqsave(&parent_ctx->lock, flags);
        parent_ctx->rotate_disable = 1;
        raw_spin_unlock_irqrestore(&parent_ctx->lock, flags);

        list_for_each_entry(event, &parent_ctx->flexible_groups, group_entry) {
                ret = inherit_task_group(event, parent, parent_ctx,
                                         child, ctxn, &inherited_all);
                if (ret)
                        goto out_unlock;
        }

        raw_spin_lock_irqsave(&parent_ctx->lock, flags);
        parent_ctx->rotate_disable = 0;

        child_ctx = child->perf_event_ctxp[ctxn];

        if (child_ctx && inherited_all) {
                /*
                 * Mark the child context as a clone of the parent
                 * context, or of whatever the parent is a clone of.
                 *
                 * Note that if the parent is a clone, the holding of
                 * parent_ctx->lock avoids it from being uncloned.
                 */
                cloned_ctx = parent_ctx->parent_ctx;
                if (cloned_ctx) {
                        child_ctx->parent_ctx = cloned_ctx;
                        child_ctx->parent_gen = parent_ctx->parent_gen;
                } else {
                        child_ctx->parent_ctx = parent_ctx;
                        child_ctx->parent_gen = parent_ctx->generation;
                }
                get_ctx(child_ctx->parent_ctx);
        }

        raw_spin_unlock_irqrestore(&parent_ctx->lock, flags);
out_unlock:
        mutex_unlock(&parent_ctx->mutex);

        perf_unpin_context(parent_ctx);
        put_ctx(parent_ctx);

        return ret;
}

/*
 * Initialize the perf_event context in task_struct
 */
int perf_event_init_task(struct task_struct *child)
{
        int ctxn, ret;

        memset(child->perf_event_ctxp, 0, sizeof(child->perf_event_ctxp));
        mutex_init(&child->perf_event_mutex);
        INIT_LIST_HEAD(&child->perf_event_list);

        for_each_task_context_nr(ctxn) {
                ret = perf_event_init_context(child, ctxn);
                if (ret) {
                        perf_event_free_task(child);
                        return ret;
                }
        }

        return 0;
}

static void __init perf_event_init_all_cpus(void)
{
        struct swevent_htable *swhash;
        int cpu;

        for_each_possible_cpu(cpu) {
                swhash = &per_cpu(swevent_htable, cpu);
                mutex_init(&swhash->hlist_mutex);
                INIT_LIST_HEAD(&per_cpu(rotation_list, cpu));
        }
}

static void __cpuinit perf_event_init_cpu(int cpu)
{
        struct swevent_htable *swhash = &per_cpu(swevent_htable, cpu);

        mutex_lock(&swhash->hlist_mutex);
        if (swhash->hlist_refcount > 0) {
                struct swevent_hlist *hlist;

                hlist = kzalloc_node(sizeof(*hlist), GFP_KERNEL, cpu_to_node(cpu));
                WARN_ON(!hlist);
                rcu_assign_pointer(swhash->swevent_hlist, hlist);
        }
        mutex_unlock(&swhash->hlist_mutex);
}

#if defined CONFIG_HOTPLUG_CPU || defined CONFIG_KEXEC
static void perf_pmu_rotate_stop(struct pmu *pmu)
{
        struct perf_cpu_context *cpuctx = this_cpu_ptr(pmu->pmu_cpu_context);

        WARN_ON(!irqs_disabled());

        list_del_init(&cpuctx->rotation_list);
}

static void __perf_event_exit_context(void *__info)
{
        struct remove_event re = { .detach_group = false };
        struct perf_event_context *ctx = __info;

        perf_pmu_rotate_stop(ctx->pmu);

        rcu_read_lock();
        list_for_each_entry_rcu(re.event, &ctx->event_list, event_entry)
                __perf_remove_from_context(&re);
        rcu_read_unlock();
}

static void perf_event_exit_cpu_context(int cpu)
{
        struct perf_event_context *ctx;
        struct pmu *pmu;
        int idx;

        idx = srcu_read_lock(&pmus_srcu);
        list_for_each_entry_rcu(pmu, &pmus, entry) {
                ctx = &per_cpu_ptr(pmu->pmu_cpu_context, cpu)->ctx;

                mutex_lock(&ctx->mutex);
                smp_call_function_single(cpu, __perf_event_exit_context, ctx, 1);
                mutex_unlock(&ctx->mutex);
        }
        srcu_read_unlock(&pmus_srcu, idx);
}

static void perf_event_exit_cpu(int cpu)
{
        perf_event_exit_cpu_context(cpu);
}
#else
static inline void perf_event_exit_cpu(int cpu) { }
#endif

static int
perf_reboot(struct notifier_block *notifier, unsigned long val, void *v)
{
        int cpu;

        for_each_online_cpu(cpu)
                perf_event_exit_cpu(cpu);

        return NOTIFY_OK;
}

/*
 * Run the perf reboot notifier at the very last possible moment so that
 * the generic watchdog code runs as long as possible.
 */
static struct notifier_block perf_reboot_notifier = {
        .notifier_call = perf_reboot,
        .priority = INT_MIN,
};

static int __cpuinit
perf_cpu_notify(struct notifier_block *self, unsigned long action, void *hcpu)
{
        unsigned int cpu = (long)hcpu;

        switch (action & ~CPU_TASKS_FROZEN) {

        case CPU_UP_PREPARE:
        case CPU_DOWN_FAILED:
                perf_event_init_cpu(cpu);
                break;

        case CPU_UP_CANCELED:
        case CPU_DOWN_PREPARE:
                perf_event_exit_cpu(cpu);
                break;

        default:
                break;
        }

        return NOTIFY_OK;
}

void __init perf_event_init(void)
{
        int ret;

        idr_init(&pmu_idr);

        perf_event_init_all_cpus();
        init_srcu_struct(&pmus_srcu);
        perf_pmu_register(&perf_swevent, "software", PERF_TYPE_SOFTWARE);
        perf_pmu_register(&perf_cpu_clock, NULL, -1);
        perf_pmu_register(&perf_task_clock, NULL, -1);
        perf_tp_register();
        perf_cpu_notifier(perf_cpu_notify);
        register_reboot_notifier(&perf_reboot_notifier);

        ret = init_hw_breakpoint();
        WARN(ret, "hw_breakpoint initialization failed with: %d", ret);
}

static int __init perf_event_sysfs_init(void)
{
        struct pmu *pmu;
        int ret;

        mutex_lock(&pmus_lock);

        ret = bus_register(&pmu_bus);
        if (ret)
                goto unlock;

        list_for_each_entry(pmu, &pmus, entry) {
                if (!pmu->name || pmu->type < 0)
                        continue;

                ret = pmu_dev_alloc(pmu);
                WARN(ret, "Failed to register pmu: %s, reason %d\n", pmu->name, ret);
        }
        pmu_bus_running = 1;
        ret = 0;

unlock:
        mutex_unlock(&pmus_lock);

        return ret;
}
device_initcall(perf_event_sysfs_init);

#ifdef CONFIG_CGROUP_PERF
static struct cgroup_subsys_state *perf_cgroup_create(
        struct cgroup_subsys *ss, struct cgroup *cont)
{
        struct perf_cgroup *jc;

        jc = kzalloc(sizeof(*jc), GFP_KERNEL);
        if (!jc)
                return ERR_PTR(-ENOMEM);

        jc->info = alloc_percpu(struct perf_cgroup_info);
        if (!jc->info) {
                kfree(jc);
                return ERR_PTR(-ENOMEM);
        }

        return &jc->css;
}

static void perf_cgroup_destroy(struct cgroup_subsys *ss,
                                struct cgroup *cont)
{
        struct perf_cgroup *jc;
        jc = container_of(cgroup_subsys_state(cont, perf_subsys_id),
                          struct perf_cgroup, css);
        free_percpu(jc->info);
        kfree(jc);
}

static int __perf_cgroup_move(void *info)
{
        struct task_struct *task = info;
        perf_cgroup_switch(task, PERF_CGROUP_SWOUT | PERF_CGROUP_SWIN);
        return 0;
}

static void
perf_cgroup_attach_task(struct cgroup *cgrp, struct task_struct *task)
{
        task_function_call(task, __perf_cgroup_move, task);
}

static void perf_cgroup_exit(struct cgroup_subsys *ss, struct cgroup *cgrp,
                struct cgroup *old_cgrp, struct task_struct *task)
{
        /*
         * cgroup_exit() is called in the copy_process() failure path.
         * Ignore this case since the task hasn't ran yet, this avoids
         * trying to poke a half freed task state from generic code.
         */
        if (!(task->flags & PF_EXITING))
                return;

        perf_cgroup_attach_task(cgrp, task);
}

struct cgroup_subsys perf_subsys = {
        .name           = "perf_event",
        .subsys_id      = perf_subsys_id,
        .create         = perf_cgroup_create,
        .destroy        = perf_cgroup_destroy,
        .exit           = perf_cgroup_exit,
        .attach_task    = perf_cgroup_attach_task,
};
#endif /* CONFIG_CGROUP_PERF */