aboutsummaryrefslogtreecommitdiff
path: root/kernel/perf_counter.c
diff options
context:
space:
mode:
Diffstat (limited to 'kernel/perf_counter.c')
-rw-r--r--kernel/perf_counter.c4962
1 files changed, 0 insertions, 4962 deletions
diff --git a/kernel/perf_counter.c b/kernel/perf_counter.c
deleted file mode 100644
index e0d91fdf0c3..00000000000
--- a/kernel/perf_counter.c
+++ /dev/null
@@ -1,4962 +0,0 @@
-/*
- * Performance counter core code
- *
- * Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
- * Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar
- * Copyright (C) 2008-2009 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/file.h>
-#include <linux/poll.h>
-#include <linux/sysfs.h>
-#include <linux/dcache.h>
-#include <linux/percpu.h>
-#include <linux/ptrace.h>
-#include <linux/vmstat.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_counter.h>
-
-#include <asm/irq_regs.h>
-
-/*
- * Each CPU has a list of per CPU counters:
- */
-DEFINE_PER_CPU(struct perf_cpu_context, perf_cpu_context);
-
-int perf_max_counters __read_mostly = 1;
-static int perf_reserved_percpu __read_mostly;
-static int perf_overcommit __read_mostly = 1;
-
-static atomic_t nr_counters __read_mostly;
-static atomic_t nr_mmap_counters __read_mostly;
-static atomic_t nr_comm_counters __read_mostly;
-static atomic_t nr_task_counters __read_mostly;
-
-/*
- * perf counter paranoia level:
- * -1 - not paranoid at all
- * 0 - disallow raw tracepoint access for unpriv
- * 1 - disallow cpu counters for unpriv
- * 2 - disallow kernel profiling for unpriv
- */
-int sysctl_perf_counter_paranoid __read_mostly = 1;
-
-static inline bool perf_paranoid_tracepoint_raw(void)
-{
- return sysctl_perf_counter_paranoid > -1;
-}
-
-static inline bool perf_paranoid_cpu(void)
-{
- return sysctl_perf_counter_paranoid > 0;
-}
-
-static inline bool perf_paranoid_kernel(void)
-{
- return sysctl_perf_counter_paranoid > 1;
-}
-
-int sysctl_perf_counter_mlock __read_mostly = 512; /* 'free' kb per user */
-
-/*
- * max perf counter sample rate
- */
-int sysctl_perf_counter_sample_rate __read_mostly = 100000;
-
-static atomic64_t perf_counter_id;
-
-/*
- * Lock for (sysadmin-configurable) counter reservations:
- */
-static DEFINE_SPINLOCK(perf_resource_lock);
-
-/*
- * Architecture provided APIs - weak aliases:
- */
-extern __weak const struct pmu *hw_perf_counter_init(struct perf_counter *counter)
-{
- return NULL;
-}
-
-void __weak hw_perf_disable(void) { barrier(); }
-void __weak hw_perf_enable(void) { barrier(); }
-
-void __weak hw_perf_counter_setup(int cpu) { barrier(); }
-void __weak hw_perf_counter_setup_online(int cpu) { barrier(); }
-
-int __weak
-hw_perf_group_sched_in(struct perf_counter *group_leader,
- struct perf_cpu_context *cpuctx,
- struct perf_counter_context *ctx, int cpu)
-{
- return 0;
-}
-
-void __weak perf_counter_print_debug(void) { }
-
-static DEFINE_PER_CPU(int, disable_count);
-
-void __perf_disable(void)
-{
- __get_cpu_var(disable_count)++;
-}
-
-bool __perf_enable(void)
-{
- return !--__get_cpu_var(disable_count);
-}
-
-void perf_disable(void)
-{
- __perf_disable();
- hw_perf_disable();
-}
-
-void perf_enable(void)
-{
- if (__perf_enable())
- hw_perf_enable();
-}
-
-static void get_ctx(struct perf_counter_context *ctx)
-{
- WARN_ON(!atomic_inc_not_zero(&ctx->refcount));
-}
-
-static void free_ctx(struct rcu_head *head)
-{
- struct perf_counter_context *ctx;
-
- ctx = container_of(head, struct perf_counter_context, rcu_head);
- kfree(ctx);
-}
-
-static void put_ctx(struct perf_counter_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);
- call_rcu(&ctx->rcu_head, free_ctx);
- }
-}
-
-static void unclone_ctx(struct perf_counter_context *ctx)
-{
- if (ctx->parent_ctx) {
- put_ctx(ctx->parent_ctx);
- ctx->parent_ctx = NULL;
- }
-}
-
-/*
- * If we inherit counters we want to return the parent counter id
- * to userspace.
- */
-static u64 primary_counter_id(struct perf_counter *counter)
-{
- u64 id = counter->id;
-
- if (counter->parent)
- id = counter->parent->id;
-
- return id;
-}
-
-/*
- * Get the perf_counter_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_counter_context *
-perf_lock_task_context(struct task_struct *task, unsigned long *flags)
-{
- struct perf_counter_context *ctx;
-
- rcu_read_lock();
- retry:
- ctx = rcu_dereference(task->perf_counter_ctxp);
- if (ctx) {
- /*
- * If this context is a clone of another, it might
- * get swapped for another underneath us by
- * perf_counter_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.
- */
- spin_lock_irqsave(&ctx->lock, *flags);
- if (ctx != rcu_dereference(task->perf_counter_ctxp)) {
- spin_unlock_irqrestore(&ctx->lock, *flags);
- goto retry;
- }
-
- if (!atomic_inc_not_zero(&ctx->refcount)) {
- spin_unlock_irqrestore(&ctx->lock, *flags);
- ctx = NULL;
- }
- }
- rcu_read_unlock();
- 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_counter_context *perf_pin_task_context(struct task_struct *task)
-{
- struct perf_counter_context *ctx;
- unsigned long flags;
-
- ctx = perf_lock_task_context(task, &flags);
- if (ctx) {
- ++ctx->pin_count;
- spin_unlock_irqrestore(&ctx->lock, flags);
- }
- return ctx;
-}
-
-static void perf_unpin_context(struct perf_counter_context *ctx)
-{
- unsigned long flags;
-
- spin_lock_irqsave(&ctx->lock, flags);
- --ctx->pin_count;
- spin_unlock_irqrestore(&ctx->lock, flags);
- put_ctx(ctx);
-}
-
-/*
- * Add a counter from the lists for its context.
- * Must be called with ctx->mutex and ctx->lock held.
- */
-static void
-list_add_counter(struct perf_counter *counter, struct perf_counter_context *ctx)
-{
- struct perf_counter *group_leader = counter->group_leader;
-
- /*
- * Depending on whether it is a standalone or sibling counter,
- * add it straight to the context's counter list, or to the group
- * leader's sibling list:
- */
- if (group_leader == counter)
- list_add_tail(&counter->list_entry, &ctx->counter_list);
- else {
- list_add_tail(&counter->list_entry, &group_leader->sibling_list);
- group_leader->nr_siblings++;
- }
-
- list_add_rcu(&counter->event_entry, &ctx->event_list);
- ctx->nr_counters++;
- if (counter->attr.inherit_stat)
- ctx->nr_stat++;
-}
-
-/*
- * Remove a counter from the lists for its context.
- * Must be called with ctx->mutex and ctx->lock held.
- */
-static void
-list_del_counter(struct perf_counter *counter, struct perf_counter_context *ctx)
-{
- struct perf_counter *sibling, *tmp;
-
- if (list_empty(&counter->list_entry))
- return;
- ctx->nr_counters--;
- if (counter->attr.inherit_stat)
- ctx->nr_stat--;
-
- list_del_init(&counter->list_entry);
- list_del_rcu(&counter->event_entry);
-
- if (counter->group_leader != counter)
- counter->group_leader->nr_siblings--;
-
- /*
- * If this was a group counter with sibling counters then
- * upgrade the siblings to singleton counters by adding them
- * to the context list directly:
- */
- list_for_each_entry_safe(sibling, tmp,
- &counter->sibling_list, list_entry) {
-
- list_move_tail(&sibling->list_entry, &ctx->counter_list);
- sibling->group_leader = sibling;
- }
-}
-
-static void
-counter_sched_out(struct perf_counter *counter,
- struct perf_cpu_context *cpuctx,
- struct perf_counter_context *ctx)
-{
- if (counter->state != PERF_COUNTER_STATE_ACTIVE)
- return;
-
- counter->state = PERF_COUNTER_STATE_INACTIVE;
- if (counter->pending_disable) {
- counter->pending_disable = 0;
- counter->state = PERF_COUNTER_STATE_OFF;
- }
- counter->tstamp_stopped = ctx->time;
- counter->pmu->disable(counter);
- counter->oncpu = -1;
-
- if (!is_software_counter(counter))
- cpuctx->active_oncpu--;
- ctx->nr_active--;
- if (counter->attr.exclusive || !cpuctx->active_oncpu)
- cpuctx->exclusive = 0;
-}
-
-static void
-group_sched_out(struct perf_counter *group_counter,
- struct perf_cpu_context *cpuctx,
- struct perf_counter_context *ctx)
-{
- struct perf_counter *counter;
-
- if (group_counter->state != PERF_COUNTER_STATE_ACTIVE)
- return;
-
- counter_sched_out(group_counter, cpuctx, ctx);
-
- /*
- * Schedule out siblings (if any):
- */
- list_for_each_entry(counter, &group_counter->sibling_list, list_entry)
- counter_sched_out(counter, cpuctx, ctx);
-
- if (group_counter->attr.exclusive)
- cpuctx->exclusive = 0;
-}
-
-/*
- * Cross CPU call to remove a performance counter
- *
- * We disable the counter on the hardware level first. After that we
- * remove it from the context list.
- */
-static void __perf_counter_remove_from_context(void *info)
-{
- struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
- struct perf_counter *counter = info;
- struct perf_counter_context *ctx = counter->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.
- */
- if (ctx->task && cpuctx->task_ctx != ctx)
- return;
-
- spin_lock(&ctx->lock);
- /*
- * Protect the list operation against NMI by disabling the
- * counters on a global level.
- */
- perf_disable();
-
- counter_sched_out(counter, cpuctx, ctx);
-
- list_del_counter(counter, ctx);
-
- if (!ctx->task) {
- /*
- * Allow more per task counters with respect to the
- * reservation:
- */
- cpuctx->max_pertask =
- min(perf_max_counters - ctx->nr_counters,
- perf_max_counters - perf_reserved_percpu);
- }
-
- perf_enable();
- spin_unlock(&ctx->lock);
-}
-
-
-/*
- * Remove the counter from a task's (or a CPU's) list of counters.
- *
- * Must be called with ctx->mutex held.
- *
- * CPU counters are removed with a smp call. For task counters we only
- * call when the task is on a CPU.
- *
- * If counter->ctx is a cloned context, callers must make sure that
- * every task struct that counter->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_counter_exit_task, it's OK because the
- * context has been detached from its task.
- */
-static void perf_counter_remove_from_context(struct perf_counter *counter)
-{
- struct perf_counter_context *ctx = counter->ctx;
- struct task_struct *task = ctx->task;
-
- if (!task) {
- /*
- * Per cpu counters are removed via an smp call and
- * the removal is always sucessful.
- */
- smp_call_function_single(counter->cpu,
- __perf_counter_remove_from_context,
- counter, 1);
- return;
- }
-
-retry:
- task_oncpu_function_call(task, __perf_counter_remove_from_context,
- counter);
-
- spin_lock_irq(&ctx->lock);
- /*
- * If the context is active we need to retry the smp call.
- */
- if (ctx->nr_active && !list_empty(&counter->list_entry)) {
- spin_unlock_irq(&ctx->lock);
- goto retry;
- }
-
- /*
- * The lock prevents that this context is scheduled in so we
- * can remove the counter safely, if the call above did not
- * succeed.
- */
- if (!list_empty(&counter->list_entry)) {
- list_del_counter(counter, ctx);
- }
- spin_unlock_irq(&ctx->lock);
-}
-
-static inline u64 perf_clock(void)
-{
- return cpu_clock(smp_processor_id());
-}
-
-/*
- * Update the record of the current time in a context.
- */
-static void update_context_time(struct perf_counter_context *ctx)
-{
- u64 now = perf_clock();
-
- ctx->time += now - ctx->timestamp;
- ctx->timestamp = now;
-}
-
-/*
- * Update the total_time_enabled and total_time_running fields for a counter.
- */
-static void update_counter_times(struct perf_counter *counter)
-{
- struct perf_counter_context *ctx = counter->ctx;
- u64 run_end;
-
- if (counter->state < PERF_COUNTER_STATE_INACTIVE ||
- counter->group_leader->state < PERF_COUNTER_STATE_INACTIVE)
- return;
-
- counter->total_time_enabled = ctx->time - counter->tstamp_enabled;
-
- if (counter->state == PERF_COUNTER_STATE_INACTIVE)
- run_end = counter->tstamp_stopped;
- else
- run_end = ctx->time;
-
- counter->total_time_running = run_end - counter->tstamp_running;
-}
-
-/*
- * Update total_time_enabled and total_time_running for all counters in a group.
- */
-static void update_group_times(struct perf_counter *leader)
-{
- struct perf_counter *counter;
-
- update_counter_times(leader);
- list_for_each_entry(counter, &leader->sibling_list, list_entry)
- update_counter_times(counter);
-}
-
-/*
- * Cross CPU call to disable a performance counter
- */
-static void __perf_counter_disable(void *info)
-{
- struct perf_counter *counter = info;
- struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
- struct perf_counter_context *ctx = counter->ctx;
-
- /*
- * If this is a per-task counter, need to check whether this
- * counter's task is the current task on this cpu.
- */
- if (ctx->task && cpuctx->task_ctx != ctx)
- return;
-
- spin_lock(&ctx->lock);
-
- /*
- * If the counter is on, turn it off.
- * If it is in error state, leave it in error state.
- */
- if (counter->state >= PERF_COUNTER_STATE_INACTIVE) {
- update_context_time(ctx);
- update_group_times(counter);
- if (counter == counter->group_leader)
- group_sched_out(counter, cpuctx, ctx);
- else
- counter_sched_out(counter, cpuctx, ctx);
- counter->state = PERF_COUNTER_STATE_OFF;
- }
-
- spin_unlock(&ctx->lock);
-}
-
-/*
- * Disable a counter.
- *
- * If counter->ctx is a cloned context, callers must make sure that
- * every task struct that counter->ctx->task could possibly point to
- * remains valid. This condition is satisifed when called through
- * perf_counter_for_each_child or perf_counter_for_each because they
- * hold the top-level counter's child_mutex, so any descendant that
- * goes to exit will block in sync_child_counter.
- * When called from perf_pending_counter it's OK because counter->ctx
- * is the current context on this CPU and preemption is disabled,
- * hence we can't get into perf_counter_task_sched_out for this context.
- */
-static void perf_counter_disable(struct perf_counter *counter)
-{
- struct perf_counter_context *ctx = counter->ctx;
- struct task_struct *task = ctx->task;
-
- if (!task) {
- /*
- * Disable the counter on the cpu that it's on
- */
- smp_call_function_single(counter->cpu, __perf_counter_disable,
- counter, 1);
- return;
- }
-
- retry:
- task_oncpu_function_call(task, __perf_counter_disable, counter);
-
- spin_lock_irq(&ctx->lock);
- /*
- * If the counter is still active, we need to retry the cross-call.
- */
- if (counter->state == PERF_COUNTER_STATE_ACTIVE) {
- spin_unlock_irq(&ctx->lock);
- goto retry;
- }
-
- /*
- * Since we have the lock this context can't be scheduled
- * in, so we can change the state safely.
- */
- if (counter->state == PERF_COUNTER_STATE_INACTIVE) {
- update_group_times(counter);
- counter->state = PERF_COUNTER_STATE_OFF;
- }
-
- spin_unlock_irq(&ctx->lock);
-}
-
-static int
-counter_sched_in(struct perf_counter *counter,
- struct perf_cpu_context *cpuctx,
- struct perf_counter_context *ctx,
- int cpu)
-{
- if (counter->state <= PERF_COUNTER_STATE_OFF)
- return 0;
-
- counter->state = PERF_COUNTER_STATE_ACTIVE;
- counter->oncpu = cpu; /* TODO: put 'cpu' into cpuctx->cpu */
- /*
- * The new state must be visible before we turn it on in the hardware:
- */
- smp_wmb();
-
- if (counter->pmu->enable(counter)) {
- counter->state = PERF_COUNTER_STATE_INACTIVE;
- counter->oncpu = -1;
- return -EAGAIN;
- }
-
- counter->tstamp_running += ctx->time - counter->tstamp_stopped;
-
- if (!is_software_counter(counter))
- cpuctx->active_oncpu++;
- ctx->nr_active++;
-
- if (counter->attr.exclusive)
- cpuctx->exclusive = 1;
-
- return 0;
-}
-
-static int
-group_sched_in(struct perf_counter *group_counter,
- struct perf_cpu_context *cpuctx,
- struct perf_counter_context *ctx,
- int cpu)
-{
- struct perf_counter *counter, *partial_group;
- int ret;
-
- if (group_counter->state == PERF_COUNTER_STATE_OFF)
- return 0;
-
- ret = hw_perf_group_sched_in(group_counter, cpuctx, ctx, cpu);
- if (ret)
- return ret < 0 ? ret : 0;
-
- if (counter_sched_in(group_counter, cpuctx, ctx, cpu))
- return -EAGAIN;
-
- /*
- * Schedule in siblings as one group (if any):
- */
- list_for_each_entry(counter, &group_counter->sibling_list, list_entry) {
- if (counter_sched_in(counter, cpuctx, ctx, cpu)) {
- partial_group = counter;
- goto group_error;
- }
- }
-
- return 0;
-
-group_error:
- /*
- * Groups can be scheduled in as one unit only, so undo any
- * partial group before returning:
- */
- list_for_each_entry(counter, &group_counter->sibling_list, list_entry) {
- if (counter == partial_group)
- break;
- counter_sched_out(counter, cpuctx, ctx);
- }
- counter_sched_out(group_counter, cpuctx, ctx);
-
- return -EAGAIN;
-}
-
-/*
- * Return 1 for a group consisting entirely of software counters,
- * 0 if the group contains any hardware counters.
- */
-static int is_software_only_group(struct perf_counter *leader)
-{
- struct perf_counter *counter;
-
- if (!is_software_counter(leader))
- return 0;
-
- list_for_each_entry(counter, &leader->sibling_list, list_entry)
- if (!is_software_counter(counter))
- return 0;
-
- return 1;
-}
-
-/*
- * Work out whether we can put this counter group on the CPU now.
- */
-static int group_can_go_on(struct perf_counter *counter,
- struct perf_cpu_context *cpuctx,
- int can_add_hw)
-{
- /*
- * Groups consisting entirely of software counters can always go on.
- */
- if (is_software_only_group(counter))
- return 1;
- /*
- * If an exclusive group is already on, no other hardware
- * counters can go on.
- */
- if (cpuctx->exclusive)
- return 0;
- /*
- * If this group is exclusive and there are already
- * counters on the CPU, it can't go on.
- */
- if (counter->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_counter_to_ctx(struct perf_counter *counter,
- struct perf_counter_context *ctx)
-{
- list_add_counter(counter, ctx);
- counter->tstamp_enabled = ctx->time;
- counter->tstamp_running = ctx->time;
- counter->tstamp_stopped = ctx->time;
-}
-
-/*
- * Cross CPU call to install and enable a performance counter
- *
- * Must be called with ctx->mutex held
- */
-static void __perf_install_in_context(void *info)
-{
- struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
- struct perf_counter *counter = info;
- struct perf_counter_context *ctx = counter->ctx;
- struct perf_counter *leader = counter->group_leader;
- int cpu = smp_processor_id();
- int err;
-
- /*
- * 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.
- * Or possibly this is the right context but it isn't
- * on this cpu because it had no counters.
- */
- if (ctx->task && cpuctx->task_ctx != ctx) {
- if (cpuctx->task_ctx || ctx->task != current)
- return;
- cpuctx->task_ctx = ctx;
- }
-
- spin_lock(&ctx->lock);
- ctx->is_active = 1;
- update_context_time(ctx);
-
- /*
- * Protect the list operation against NMI by disabling the
- * counters on a global level. NOP for non NMI based counters.
- */
- perf_disable();
-
- add_counter_to_ctx(counter, ctx);
-
- /*
- * Don't put the counter on if it is disabled or if
- * it is in a group and the group isn't on.
- */
- if (counter->state != PERF_COUNTER_STATE_INACTIVE ||
- (leader != counter && leader->state != PERF_COUNTER_STATE_ACTIVE))
- goto unlock;
-
- /*
- * An exclusive counter can't go on if there are already active
- * hardware counters, and no hardware counter can go on if there
- * is already an exclusive counter on.
- */
- if (!group_can_go_on(counter, cpuctx, 1))
- err = -EEXIST;
- else
- err = counter_sched_in(counter, cpuctx, ctx, cpu);
-
- if (err) {
- /*
- * This counter couldn't go on. If it is in a group
- * then we have to pull the whole group off.
- * If the counter group is pinned then put it in error state.
- */
- if (leader != counter)
- group_sched_out(leader, cpuctx, ctx);
- if (leader->attr.pinned) {
- update_group_times(leader);
- leader->state = PERF_COUNTER_STATE_ERROR;
- }
- }
-
- if (!err && !ctx->task && cpuctx->max_pertask)
- cpuctx->max_pertask--;
-
- unlock:
- perf_enable();
-
- spin_unlock(&ctx->lock);
-}
-
-/*
- * Attach a performance counter to a context
- *
- * First we add the counter to the list with the hardware enable bit
- * in counter->hw_config cleared.
- *
- * If the counter 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.
- *
- * Must be called with ctx->mutex held.
- */
-static void
-perf_install_in_context(struct perf_counter_context *ctx,
- struct perf_counter *counter,
- int cpu)
-{
- struct task_struct *task = ctx->task;
-
- if (!task) {
- /*
- * Per cpu counters are installed via an smp call and
- * the install is always sucessful.
- */
- smp_call_function_single(cpu, __perf_install_in_context,
- counter, 1);
- return;
- }
-
-retry:
- task_oncpu_function_call(task, __perf_install_in_context,
- counter);
-
- spin_lock_irq(&ctx->lock);
- /*
- * we need to retry the smp call.
- */
- if (ctx->is_active && list_empty(&counter->list_entry)) {
- spin_unlock_irq(&ctx->lock);
- goto retry;
- }
-
- /*
- * The lock prevents that this context is scheduled in so we
- * can add the counter safely, if it the call above did not
- * succeed.
- */
- if (list_empty(&counter->list_entry))
- add_counter_to_ctx(counter, ctx);
- spin_unlock_irq(&ctx->lock);
-}
-
-/*
- * Put a counter 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_counter_mark_enabled(struct perf_counter *counter,
- struct perf_counter_context *ctx)
-{
- struct perf_counter *sub;
-
- counter->state = PERF_COUNTER_STATE_INACTIVE;
- counter->tstamp_enabled = ctx->time - counter->total_time_enabled;
- list_for_each_entry(sub, &counter->sibling_list, list_entry)
- if (sub->state >= PERF_COUNTER_STATE_INACTIVE)
- sub->tstamp_enabled =
- ctx->time - sub->total_time_enabled;
-}
-
-/*
- * Cross CPU call to enable a performance counter
- */
-static void __perf_counter_enable(void *info)
-{
- struct perf_counter *counter = info;
- struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
- struct perf_counter_context *ctx = counter->ctx;
- struct perf_counter *leader = counter->group_leader;
- int err;
-
- /*
- * If this is a per-task counter, need to check whether this
- * counter's task is the current task on this cpu.
- */
- if (ctx->task && cpuctx->task_ctx != ctx) {
- if (cpuctx->task_ctx || ctx->task != current)
- return;
- cpuctx->task_ctx = ctx;
- }
-
- spin_lock(&ctx->lock);
- ctx->is_active = 1;
- update_context_time(ctx);
-
- if (counter->state >= PERF_COUNTER_STATE_INACTIVE)
- goto unlock;
- __perf_counter_mark_enabled(counter, ctx);
-
- /*
- * If the counter is in a group and isn't the group leader,
- * then don't put it on unless the group is on.
- */
- if (leader != counter && leader->state != PERF_COUNTER_STATE_ACTIVE)
- goto unlock;
-
- if (!group_can_go_on(counter, cpuctx, 1)) {
- err = -EEXIST;
- } else {
- perf_disable();
- if (counter == leader)
- err = group_sched_in(counter, cpuctx, ctx,
- smp_processor_id());
- else
- err = counter_sched_in(counter, cpuctx, ctx,
- smp_processor_id());
- perf_enable();
- }
-
- if (err) {
- /*
- * If this counter can't go on and it's part of a
- * group, then the whole group has to come off.
- */
- if (leader != counter)
- group_sched_out(leader, cpuctx, ctx);
- if (leader->attr.pinned) {
- update_group_times(leader);
- leader->state = PERF_COUNTER_STATE_ERROR;
- }
- }
-
- unlock:
- spin_unlock(&ctx->lock);
-}
-
-/*
- * Enable a counter.
- *
- * If counter->ctx is a cloned context, callers must make sure that
- * every task struct that counter->ctx->task could possibly point to
- * remains valid. This condition is satisfied when called through
- * perf_counter_for_each_child or perf_counter_for_each as described
- * for perf_counter_disable.
- */
-static void perf_counter_enable(struct perf_counter *counter)
-{
- struct perf_counter_context *ctx = counter->ctx;
- struct task_struct *task = ctx->task;
-
- if (!task) {
- /*
- * Enable the counter on the cpu that it's on
- */
- smp_call_function_single(counter->cpu, __perf_counter_enable,
- counter, 1);
- return;
- }
-
- spin_lock_irq(&ctx->lock);
- if (counter->state >= PERF_COUNTER_STATE_INACTIVE)
- goto out;
-
- /*
- * If the counter is in error state, clear that first.
- * That way, if we see the counter 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 (counter->state == PERF_COUNTER_STATE_ERROR)
- counter->state = PERF_COUNTER_STATE_OFF;
-
- retry:
- spin_unlock_irq(&ctx->lock);
- task_oncpu_function_call(task, __perf_counter_enable, counter);
-
- spin_lock_irq(&ctx->lock);
-
- /*
- * If the context is active and the counter is still off,
- * we need to retry the cross-call.
- */
- if (ctx->is_active && counter->state == PERF_COUNTER_STATE_OFF)
- goto retry;
-
- /*
- * Since we have the lock this context can't be scheduled
- * in, so we can change the state safely.
- */
- if (counter->state == PERF_COUNTER_STATE_OFF)
- __perf_counter_mark_enabled(counter, ctx);
-
- out:
- spin_unlock_irq(&ctx->lock);
-}
-
-static int perf_counter_refresh(struct perf_counter *counter, int refresh)
-{
- /*
- * not supported on inherited counters
- */
- if (counter->attr.inherit)
- return -EINVAL;
-
- atomic_add(refresh, &counter->event_limit);
- perf_counter_enable(counter);
-
- return 0;
-}
-
-void __perf_counter_sched_out(struct perf_counter_context *ctx,
- struct perf_cpu_context *cpuctx)
-{
- struct perf_counter *counter;
-
- spin_lock(&ctx->lock);
- ctx->is_active = 0;
- if (likely(!ctx->nr_counters))
- goto out;
- update_context_time(ctx);
-
- perf_disable();
- if (ctx->nr_active) {
- list_for_each_entry(counter, &ctx->counter_list, list_entry) {
- if (counter != counter->group_leader)
- counter_sched_out(counter, cpuctx, ctx);
- else
- group_sched_out(counter, cpuctx, ctx);
- }
- }
- perf_enable();
- out:
- spin_unlock(&ctx->lock);
-}
-
-/*
- * 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 counters.
- * If the number of enabled counters is the same, then the set
- * of enabled counters should be the same, because these are both
- * inherited contexts, therefore we can't access individual counters
- * in them directly with an fd; we can only enable/disable all
- * counters via prctl, or enable/disable all counters in a family
- * via ioctl, which will have the same effect on both contexts.
- */
-static int context_equiv(struct perf_counter_context *ctx1,
- struct perf_counter_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_counter_read(void *counter);
-
-static void __perf_counter_sync_stat(struct perf_counter *counter,
- struct perf_counter *next_counter)
-{
- u64 value;
-
- if (!counter->attr.inherit_stat)
- return;
-
- /*
- * Update the counter value, we cannot use perf_counter_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 counter must be on the current CPU, therefore we
- * don't need to use it.
- */
- switch (counter->state) {
- case PERF_COUNTER_STATE_ACTIVE:
- __perf_counter_read(counter);
- break;
-
- case PERF_COUNTER_STATE_INACTIVE:
- update_counter_times(counter);
- break;
-
- default:
- break;
- }
-
- /*
- * In order to keep per-task stats reliable we need to flip the counter
- * values when we flip the contexts.
- */
- value = atomic64_read(&next_counter->count);
- value = atomic64_xchg(&counter->count, value);
- atomic64_set(&next_counter->count, value);
-
- swap(counter->total_time_enabled, next_counter->total_time_enabled);
- swap(counter->total_time_running, next_counter->total_time_running);
-
- /*
- * Since we swizzled the values, update the user visible data too.
- */
- perf_counter_update_userpage(counter);
- perf_counter_update_userpage(next_counter);
-}
-
-#define list_next_entry(pos, member) \
- list_entry(pos->member.next, typeof(*pos), member)
-
-static void perf_counter_sync_stat(struct perf_counter_context *ctx,
- struct perf_counter_context *next_ctx)
-{
- struct perf_counter *counter, *next_counter;
-
- if (!ctx->nr_stat)
- return;
-
- counter = list_first_entry(&ctx->event_list,
- struct perf_counter, event_entry);
-
- next_counter = list_first_entry(&next_ctx->event_list,
- struct perf_counter, event_entry);
-
- while (&counter->event_entry != &ctx->event_list &&
- &next_counter->event_entry != &next_ctx->event_list) {
-
- __perf_counter_sync_stat(counter, next_counter);
-
- counter = list_next_entry(counter, event_entry);
- next_counter = list_next_entry(next_counter, event_entry);
- }
-}
-
-/*
- * Called from scheduler to remove the counters of the current task,
- * with interrupts disabled.
- *
- * We stop each counter and update the counter value in counter->count.
- *
- * This does not protect us against NMI, but disable()
- * sets the disabled bit in the control field of counter _before_
- * accessing the counter control register. If a NMI hits, then it will
- * not restart the counter.
- */
-void perf_counter_task_sched_out(struct task_struct *task,
- struct task_struct *next, int cpu)
-{
- struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu);
- struct perf_counter_context *ctx = task->perf_counter_ctxp;
- struct perf_counter_context *next_ctx;
- struct perf_counter_context *parent;
- struct pt_regs *regs;
- int do_switch = 1;
-
- regs = task_pt_regs(task);
- perf_swcounter_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, regs, 0);
-
- if (likely(!ctx || !cpuctx->task_ctx))
- return;
-
- update_context_time(ctx);
-
- rcu_read_lock();
- parent = rcu_dereference(ctx->parent_ctx);
- next_ctx = next->perf_counter_ctxp;
- 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.
- */
- spin_lock(&ctx->lock);
- 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_counter_ctxp
- */
- task->perf_counter_ctxp = next_ctx;
- next->perf_counter_ctxp = ctx;
- ctx->task = next;
- next_ctx->task = task;
- do_switch = 0;
-
- perf_counter_sync_stat(ctx, nex