#include <linux/smp.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>
*/
int sysctl_perf_event_paranoid __read_mostly = 1;
-static inline bool perf_paranoid_tracepoint_raw(void)
-{
- return sysctl_perf_event_paranoid > -1;
-}
-
-static inline bool perf_paranoid_cpu(void)
-{
- return sysctl_perf_event_paranoid > 0;
-}
-
-static inline bool perf_paranoid_kernel(void)
-{
- return sysctl_perf_event_paranoid > 1;
-}
-
int sysctl_perf_event_mlock __read_mostly = 512; /* 'free' kb per user */
/*
void __weak hw_perf_disable(void) { barrier(); }
void __weak hw_perf_enable(void) { barrier(); }
-void __weak hw_perf_event_setup(int cpu) { barrier(); }
-void __weak hw_perf_event_setup_online(int cpu) { barrier(); }
-
int __weak
hw_perf_group_sched_in(struct perf_event *group_leader,
struct perf_cpu_context *cpuctx,
- struct perf_event_context *ctx, int cpu)
+ struct perf_event_context *ctx)
{
return 0;
}
static DEFINE_PER_CPU(int, perf_disable_count);
-void __perf_disable(void)
-{
- __get_cpu_var(perf_disable_count)++;
-}
-
-bool __perf_enable(void)
-{
- return !--__get_cpu_var(perf_disable_count);
-}
-
void perf_disable(void)
{
- __perf_disable();
- hw_perf_disable();
+ if (!__get_cpu_var(perf_disable_count)++)
+ hw_perf_disable();
}
void perf_enable(void)
{
- if (__perf_enable())
+ if (!--__get_cpu_var(perf_disable_count))
hw_perf_enable();
}
static inline u64 perf_clock(void)
{
- return cpu_clock(smp_processor_id());
+ return cpu_clock(raw_smp_processor_id());
}
/*
static int
event_sched_in(struct perf_event *event,
struct perf_cpu_context *cpuctx,
- struct perf_event_context *ctx,
- int cpu)
+ struct perf_event_context *ctx)
{
if (event->state <= PERF_EVENT_STATE_OFF)
return 0;
event->state = PERF_EVENT_STATE_ACTIVE;
- event->oncpu = cpu; /* TODO: put 'cpu' into cpuctx->cpu */
+ event->oncpu = smp_processor_id();
/*
* The new state must be visible before we turn it on in the hardware:
*/
static int
group_sched_in(struct perf_event *group_event,
struct perf_cpu_context *cpuctx,
- struct perf_event_context *ctx,
- int cpu)
+ struct perf_event_context *ctx)
{
struct perf_event *event, *partial_group;
int ret;
if (group_event->state == PERF_EVENT_STATE_OFF)
return 0;
- ret = hw_perf_group_sched_in(group_event, cpuctx, ctx, cpu);
+ ret = hw_perf_group_sched_in(group_event, cpuctx, ctx);
if (ret)
return ret < 0 ? ret : 0;
- if (event_sched_in(group_event, cpuctx, ctx, cpu))
+ if (event_sched_in(group_event, cpuctx, ctx))
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, cpu)) {
+ if (event_sched_in(event, cpuctx, ctx)) {
partial_group = event;
goto group_error;
}
struct perf_event *event = info;
struct perf_event_context *ctx = event->ctx;
struct perf_event *leader = event->group_leader;
- int cpu = smp_processor_id();
int err;
/*
if (!group_can_go_on(event, cpuctx, 1))
err = -EEXIST;
else
- err = event_sched_in(event, cpuctx, ctx, cpu);
+ err = event_sched_in(event, cpuctx, ctx);
if (err) {
/*
} else {
perf_disable();
if (event == leader)
- err = group_sched_in(event, cpuctx, ctx,
- smp_processor_id());
+ err = group_sched_in(event, cpuctx, ctx);
else
- err = event_sched_in(event, cpuctx, ctx,
- smp_processor_id());
+ err = event_sched_in(event, cpuctx, ctx);
perf_enable();
}
return 0;
}
-void __perf_event_sched_out(struct perf_event_context *ctx,
- struct perf_cpu_context *cpuctx)
+enum event_type_t {
+ EVENT_FLEXIBLE = 0x1,
+ EVENT_PINNED = 0x2,
+ EVENT_ALL = EVENT_FLEXIBLE | EVENT_PINNED,
+};
+
+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;
update_context_time(ctx);
perf_disable();
- if (ctx->nr_active) {
+ if (!ctx->nr_active)
+ goto out_enable;
+
+ if (event_type & EVENT_PINNED)
list_for_each_entry(event, &ctx->pinned_groups, group_entry)
group_sched_out(event, cpuctx, ctx);
+ if (event_type & EVENT_FLEXIBLE)
list_for_each_entry(event, &ctx->flexible_groups, group_entry)
group_sched_out(event, cpuctx, ctx);
- }
+
+ out_enable:
perf_enable();
out:
raw_spin_unlock(&ctx->lock);
struct perf_event_context *ctx = task->perf_event_ctxp;
struct perf_event_context *next_ctx;
struct perf_event_context *parent;
- struct pt_regs *regs;
int do_switch = 1;
- regs = task_pt_regs(task);
- perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, regs, 0);
+ perf_sw_event(PERF_COUNT_SW_CONTEXT_SWITCHES, 1, 1, NULL, 0);
if (likely(!ctx || !cpuctx->task_ctx))
return;
rcu_read_unlock();
if (do_switch) {
- __perf_event_sched_out(ctx, cpuctx);
+ ctx_sched_out(ctx, cpuctx, EVENT_ALL);
cpuctx->task_ctx = NULL;
}
}
-/*
- * Called with IRQs disabled
- */
-static void __perf_event_task_sched_out(struct perf_event_context *ctx)
+static void task_ctx_sched_out(struct perf_event_context *ctx,
+ enum event_type_t event_type)
{
struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
if (WARN_ON_ONCE(ctx != cpuctx->task_ctx))
return;
- __perf_event_sched_out(ctx, cpuctx);
+ ctx_sched_out(ctx, cpuctx, event_type);
cpuctx->task_ctx = NULL;
}
/*
* Called with IRQs disabled
*/
-static void perf_event_cpu_sched_out(struct perf_cpu_context *cpuctx)
+static void __perf_event_task_sched_out(struct perf_event_context *ctx)
{
- __perf_event_sched_out(&cpuctx->ctx, cpuctx);
+ task_ctx_sched_out(ctx, EVENT_ALL);
+}
+
+/*
+ * 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
-__perf_event_sched_in(struct perf_event_context *ctx,
- struct perf_cpu_context *cpuctx)
+ctx_pinned_sched_in(struct perf_event_context *ctx,
+ struct perf_cpu_context *cpuctx)
{
- int cpu = smp_processor_id();
struct perf_event *event;
- int can_add_hw = 1;
- raw_spin_lock(&ctx->lock);
- ctx->is_active = 1;
- if (likely(!ctx->nr_events))
- goto out;
-
- ctx->timestamp = perf_clock();
-
- perf_disable();
-
- /*
- * First go through the list and put on any pinned groups
- * in order to give them the best chance of going on.
- */
list_for_each_entry(event, &ctx->pinned_groups, group_entry) {
if (event->state <= PERF_EVENT_STATE_OFF)
continue;
- if (event->cpu != -1 && event->cpu != cpu)
+ if (event->cpu != -1 && event->cpu != smp_processor_id())
continue;
if (group_can_go_on(event, cpuctx, 1))
- group_sched_in(event, cpuctx, ctx, cpu);
+ group_sched_in(event, cpuctx, ctx);
/*
* If this pinned group hasn't been scheduled,
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 */
* Listen to the 'cpu' scheduling filter constraint
* of events:
*/
- if (event->cpu != -1 && event->cpu != cpu)
+ if (event->cpu != -1 && event->cpu != smp_processor_id())
continue;
if (group_can_go_on(event, cpuctx, can_add_hw))
- if (group_sched_in(event, cpuctx, ctx, cpu))
+ 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)
+{
+ raw_spin_lock(&ctx->lock);
+ ctx->is_active = 1;
+ if (likely(!ctx->nr_events))
+ goto out;
+
+ ctx->timestamp = perf_clock();
+
+ perf_disable();
+
+ /*
+ * First go through the list and put on any pinned groups
+ * in order to give them the best chance of going on.
+ */
+ if (event_type & EVENT_PINNED)
+ ctx_pinned_sched_in(ctx, cpuctx);
+
+ /* Then walk through the lower prio flexible groups */
+ if (event_type & EVENT_FLEXIBLE)
+ ctx_flexible_sched_in(ctx, cpuctx);
+
perf_enable();
out:
raw_spin_unlock(&ctx->lock);
}
+static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx,
+ enum event_type_t event_type)
+{
+ struct perf_event_context *ctx = &cpuctx->ctx;
+
+ ctx_sched_in(ctx, cpuctx, event_type);
+}
+
+static void task_ctx_sched_in(struct task_struct *task,
+ enum event_type_t event_type)
+{
+ struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
+ struct perf_event_context *ctx = task->perf_event_ctxp;
+
+ if (likely(!ctx))
+ return;
+ if (cpuctx->task_ctx == ctx)
+ return;
+ ctx_sched_in(ctx, cpuctx, event_type);
+ cpuctx->task_ctx = ctx;
+}
/*
* Called from scheduler to add the events of the current task
* with interrupts disabled.
if (likely(!ctx))
return;
+
if (cpuctx->task_ctx == ctx)
return;
- __perf_event_sched_in(ctx, cpuctx);
+
+ perf_disable();
+
+ /*
+ * 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);
+
+ ctx_sched_in(ctx, cpuctx, EVENT_PINNED);
+ cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE);
+ ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE);
+
cpuctx->task_ctx = ctx;
+
+ perf_enable();
}
-static void perf_event_cpu_sched_in(struct perf_cpu_context *cpuctx)
+#define MAX_INTERRUPTS (~0ULL)
+
+static void perf_log_throttle(struct perf_event *event, int enable);
+
+static u64 perf_calculate_period(struct perf_event *event, u64 nsec, u64 count)
{
- struct perf_event_context *ctx = &cpuctx->ctx;
+ 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
+ *
+ */
- __perf_event_sched_in(ctx, cpuctx);
+ /*
+ * 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;
+ }
+
+ return div64_u64(dividend, divisor);
}
-#define MAX_INTERRUPTS (~0ULL)
+static void perf_event_stop(struct perf_event *event)
+{
+ if (!event->pmu->stop)
+ return event->pmu->disable(event);
-static void perf_log_throttle(struct perf_event *event, int enable);
+ return event->pmu->stop(event);
+}
+
+static int perf_event_start(struct perf_event *event)
+{
+ if (!event->pmu->start)
+ return event->pmu->enable(event);
-static void perf_adjust_period(struct perf_event *event, u64 events)
+ return event->pmu->start(event);
+}
+
+static void perf_adjust_period(struct perf_event *event, u64 nsec, u64 count)
{
struct hw_perf_event *hwc = &event->hw;
u64 period, sample_period;
s64 delta;
- events *= hwc->sample_period;
- period = div64_u64(events, event->attr.sample_freq);
+ period = perf_calculate_period(event, nsec, count);
delta = (s64)(period - hwc->sample_period);
delta = (delta + 7) / 8; /* low pass filter */
sample_period = 1;
hwc->sample_period = sample_period;
+
+ if (atomic64_read(&hwc->period_left) > 8*sample_period) {
+ perf_disable();
+ perf_event_stop(event);
+ atomic64_set(&hwc->period_left, 0);
+ perf_event_start(event);
+ perf_enable();
+ }
}
static void perf_ctx_adjust_freq(struct perf_event_context *ctx)
{
struct perf_event *event;
struct hw_perf_event *hwc;
- u64 interrupts, freq;
+ u64 interrupts, now;
+ s64 delta;
raw_spin_lock(&ctx->lock);
list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
*/
if (interrupts == MAX_INTERRUPTS) {
perf_log_throttle(event, 1);
+ perf_disable();
event->pmu->unthrottle(event);
- interrupts = 2*sysctl_perf_event_sample_rate/HZ;
+ perf_enable();
}
if (!event->attr.freq || !event->attr.sample_freq)
continue;
- /*
- * if the specified freq < HZ then we need to skip ticks
- */
- if (event->attr.sample_freq < HZ) {
- freq = event->attr.sample_freq;
-
- hwc->freq_count += freq;
- hwc->freq_interrupts += interrupts;
-
- if (hwc->freq_count < HZ)
- continue;
-
- interrupts = hwc->freq_interrupts;
- hwc->freq_interrupts = 0;
- hwc->freq_count -= HZ;
- } else
- freq = HZ;
-
- perf_adjust_period(event, freq * interrupts);
+ perf_disable();
+ event->pmu->read(event);
+ now = atomic64_read(&event->count);
+ delta = now - hwc->freq_count_stamp;
+ hwc->freq_count_stamp = now;
- /*
- * In order to avoid being stalled by an (accidental) huge
- * sample period, force reset the sample period if we didn't
- * get any events in this freq period.
- */
- if (!interrupts) {
- perf_disable();
- event->pmu->disable(event);
- atomic64_set(&hwc->period_left, 0);
- event->pmu->enable(event);
- perf_enable();
- }
+ if (delta > 0)
+ perf_adjust_period(event, TICK_NSEC, delta);
+ perf_enable();
}
raw_spin_unlock(&ctx->lock);
}
*/
static void rotate_ctx(struct perf_event_context *ctx)
{
- if (!ctx->nr_events)
- return;
-
raw_spin_lock(&ctx->lock);
/* Rotate the first entry last of non-pinned groups */
- perf_disable();
-
list_rotate_left(&ctx->flexible_groups);
- perf_enable();
-
raw_spin_unlock(&ctx->lock);
}
{
struct perf_cpu_context *cpuctx;
struct perf_event_context *ctx;
+ int rotate = 0;
if (!atomic_read(&nr_events))
return;
cpuctx = &__get_cpu_var(perf_cpu_context);
+ if (cpuctx->ctx.nr_events &&
+ cpuctx->ctx.nr_events != cpuctx->ctx.nr_active)
+ rotate = 1;
+
ctx = curr->perf_event_ctxp;
+ if (ctx && ctx->nr_events && ctx->nr_events != ctx->nr_active)
+ rotate = 1;
perf_ctx_adjust_freq(&cpuctx->ctx);
if (ctx)
perf_ctx_adjust_freq(ctx);
- perf_event_cpu_sched_out(cpuctx);
+ if (!rotate)
+ return;
+
+ perf_disable();
+ cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE);
if (ctx)
- __perf_event_task_sched_out(ctx);
+ task_ctx_sched_out(ctx, EVENT_FLEXIBLE);
rotate_ctx(&cpuctx->ctx);
if (ctx)
rotate_ctx(ctx);
- perf_event_cpu_sched_in(cpuctx);
+ cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE);
if (ctx)
- perf_event_task_sched_in(curr);
+ task_ctx_sched_in(curr, EVENT_FLEXIBLE);
+ perf_enable();
}
static int event_enable_on_exec(struct perf_event *event,
if (user_locked > user_lock_limit)
extra = user_locked - user_lock_limit;
- lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
+ lock_limit = rlimit(RLIMIT_MEMLOCK);
lock_limit >>= PAGE_SHIFT;
locked = vma->vm_mm->locked_vm + extra;
}
static const struct file_operations perf_fops = {
+ .llseek = no_llseek,
.release = perf_release,
.read = perf_read,
.poll = perf_poll,
return NULL;
}
+__weak
+void perf_arch_fetch_caller_regs(struct pt_regs *regs, unsigned long ip, int skip)
+{
+}
+
+
+/*
+ * 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);
+
/*
* Output
*/
struct perf_task_event *task_event)
{
struct perf_output_handle handle;
- int size;
struct task_struct *task = task_event->task;
- int ret;
+ unsigned long flags;
+ int size, ret;
+
+ /*
+ * If this CPU attempts to acquire an rq lock held by a CPU spinning
+ * in perf_output_lock() from interrupt context, it's game over.
+ */
+ local_irq_save(flags);
size = task_event->event_id.header.size;
ret = perf_output_begin(&handle, event, size, 0, 0);
- if (ret)
+ if (ret) {
+ local_irq_restore(flags);
return;
+ }
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);
- task_event->event_id.time = perf_clock();
-
perf_output_put(&handle, task_event->event_id);
perf_output_end(&handle);
+ local_irq_restore(flags);
}
static int perf_event_task_match(struct perf_event *event)
{
+ if (event->state < PERF_EVENT_STATE_INACTIVE)
+ return 0;
+
if (event->cpu != -1 && event->cpu != smp_processor_id())
return 0;
cpuctx = &get_cpu_var(perf_cpu_context);
perf_event_task_ctx(&cpuctx->ctx, task_event);
if (!ctx)
- ctx = rcu_dereference(task_event->task->perf_event_ctxp);
+ ctx = rcu_dereference(current->perf_event_ctxp);
if (ctx)
perf_event_task_ctx(ctx, task_event);
put_cpu_var(perf_cpu_context);
/* .ppid */
/* .tid */
/* .ptid */
+ .time = perf_clock(),
},
};
static int perf_event_comm_match(struct perf_event *event)
{
+ if (event->state < PERF_EVENT_STATE_INACTIVE)
+ return 0;
+
if (event->cpu != -1 && event->cpu != smp_processor_id())
return 0;
static int perf_event_mmap_match(struct perf_event *event,
struct perf_mmap_event *mmap_event)
{
+ if (event->state < PERF_EVENT_STATE_INACTIVE)
+ return 0;
+
if (event->cpu != -1 && event->cpu != smp_processor_id())
return 0;
.event_id = {
.header = {
.type = PERF_RECORD_MMAP,
- .misc = 0,
+ .misc = PERF_RECORD_MISC_USER,
/* .size */
},
/* .pid */
/* .tid */
.start = vma->vm_start,
.len = vma->vm_end - vma->vm_start,
- .pgoff = vma->vm_pgoff,
+ .pgoff = (u64)vma->vm_pgoff << PAGE_SHIFT,
},
};
if (event->attr.freq) {
u64 now = perf_clock();
- s64 delta = now - hwc->freq_stamp;
+ s64 delta = now - hwc->freq_time_stamp;
- hwc->freq_stamp = now;
+ hwc->freq_time_stamp = now;
- if (delta > 0 && delta < TICK_NSEC)
- perf_adjust_period(event, NSEC_PER_SEC / (int)delta);
+ if (delta > 0 && delta < 2*TICK_NSEC)
+ perf_adjust_period(event, delta, hwc->last_period);
}
/*
perf_swevent_overflow(event, 0, nmi, data, regs);
}
-static int perf_swevent_is_counting(struct perf_event *event)
-{
- /*
- * The event is active, we're good!
- */
- if (event->state == PERF_EVENT_STATE_ACTIVE)
- return 1;
-
- /*
- * The event is off/error, not counting.
- */
- if (event->state != PERF_EVENT_STATE_INACTIVE)
- return 0;
-
- /*
- * The event is inactive, if the context is active
- * we're part of a group that didn't make it on the 'pmu',
- * not counting.
- */
- if (event->ctx->is_active)
- return 0;
-
- /*
- * We're inactive and the context is too, this means the
- * task is scheduled out, we're counting events that happen
- * to us, like migration events.
- */
- return 1;
-}
-
static int perf_tp_event_match(struct perf_event *event,
struct perf_sample_data *data);
struct perf_sample_data *data,
struct pt_regs *regs)
{
- if (event->cpu != -1 && event->cpu != smp_processor_id())
- return 0;
-
- if (!perf_swevent_is_counting(event))
- return 0;
-
if (event->attr.type != type)
return 0;
return 1;
}
-static void perf_swevent_ctx_event(struct perf_event_context *ctx,
- enum perf_type_id type,
- u32 event_id, u64 nr, int nmi,
- struct perf_sample_data *data,
- struct pt_regs *regs)
+static inline u64 swevent_hash(u64 type, u32 event_id)
+{
+ u64 val = event_id | (type << 32);
+
+ return hash_64(val, SWEVENT_HLIST_BITS);
+}
+
+static struct hlist_head *
+find_swevent_head(struct perf_cpu_context *ctx, u64 type, u32 event_id)
+{
+ u64 hash;
+ struct swevent_hlist *hlist;
+
+ hash = swevent_hash(type, event_id);
+
+ hlist = rcu_dereference(ctx->swevent_hlist);
+ if (!hlist)
+ return NULL;
+
+ return &hlist->heads[hash];
+}
+
+static void do_perf_sw_event(enum perf_type_id type, u32 event_id,
+ u64 nr, int nmi,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
{
+ struct perf_cpu_context *cpuctx;
struct perf_event *event;
+ struct hlist_node *node;
+ struct hlist_head *head;
- list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
+ cpuctx = &__get_cpu_var(perf_cpu_context);
+
+ rcu_read_lock();
+
+ head = find_swevent_head(cpuctx, 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_add(event, nr, nmi, data, regs);
}
+end:
+ rcu_read_unlock();
}
int perf_swevent_get_recursion_context(void)
}
EXPORT_SYMBOL_GPL(perf_swevent_put_recursion_context);
-static void do_perf_sw_event(enum perf_type_id type, u32 event_id,
- u64 nr, int nmi,
- struct perf_sample_data *data,
- struct pt_regs *regs)
-{
- struct perf_cpu_context *cpuctx;
- struct perf_event_context *ctx;
-
- cpuctx = &__get_cpu_var(perf_cpu_context);
- rcu_read_lock();
- perf_swevent_ctx_event(&cpuctx->ctx, type, event_id,
- nr, nmi, data, regs);
- /*
- * doesn't really matter which of the child contexts the
- * events ends up in.
- */
- ctx = rcu_dereference(current->perf_event_ctxp);
- if (ctx)
- perf_swevent_ctx_event(ctx, type, event_id, nr, nmi, data, regs);
- rcu_read_unlock();
-}
void __perf_sw_event(u32 event_id, u64 nr, int nmi,
struct pt_regs *regs, u64 addr)
if (rctx < 0)
return;
- data.addr = addr;
- data.raw = NULL;
+ perf_sample_data_init(&data, addr);
do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, nmi, &data, regs);
static int perf_swevent_enable(struct perf_event *event)
{
struct hw_perf_event *hwc = &event->hw;
+ struct perf_cpu_context *cpuctx;
+ struct hlist_head *head;
+
+ cpuctx = &__get_cpu_var(perf_cpu_context);
if (hwc->sample_period) {
hwc->last_period = hwc->sample_period;
perf_swevent_set_period(event);
}
+
+ head = find_swevent_head(cpuctx, event->attr.type, event->attr.config);
+ if (WARN_ON_ONCE(!head))
+ return -EINVAL;
+
+ hlist_add_head_rcu(&event->hlist_entry, head);
+
return 0;
}
static void perf_swevent_disable(struct perf_event *event)
{
+ hlist_del_rcu(&event->hlist_entry);
}
static const struct pmu perf_ops_generic = {
struct perf_event *event;
u64 period;
- event = container_of(hrtimer, struct perf_event, hw.hrtimer);
+ event = container_of(hrtimer, struct perf_event, hw.hrtimer);
event->pmu->read(event);
- data.addr = 0;
- data.raw = NULL;
+ perf_sample_data_init(&data, 0);
data.period = event->hw.last_period;
regs = get_irq_regs();
- /*
- * In case we exclude kernel IPs or are somehow not in interrupt
- * context, provide the next best thing, the user IP.
- */
- if ((event->attr.exclude_kernel || !regs) &&
- !event->attr.exclude_user)
- regs = task_pt_regs(current);
- if (regs) {
+ if (regs && !perf_exclude_event(event, regs)) {
if (!(event->attr.exclude_idle && current->pid == 0))
if (perf_event_overflow(event, 0, &data, regs))
ret = HRTIMER_NORESTART;
.read = task_clock_perf_event_read,
};
+static void swevent_hlist_release_rcu(struct rcu_head *rcu_head)
+{
+ struct swevent_hlist *hlist;
+
+ hlist = container_of(rcu_head, struct swevent_hlist, rcu_head);
+ kfree(hlist);
+}
+
+static void swevent_hlist_release(struct perf_cpu_context *cpuctx)
+{
+ struct swevent_hlist *hlist;
+
+ if (!cpuctx->swevent_hlist)
+ return;
+
+ hlist = cpuctx->swevent_hlist;
+ rcu_assign_pointer(cpuctx->swevent_hlist, NULL);
+ call_rcu(&hlist->rcu_head, swevent_hlist_release_rcu);
+}
+
+static void swevent_hlist_put_cpu(struct perf_event *event, int cpu)
+{
+ struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu);
+
+ mutex_lock(&cpuctx->hlist_mutex);
+
+ if (!--cpuctx->hlist_refcount)
+ swevent_hlist_release(cpuctx);
+
+ mutex_unlock(&cpuctx->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 perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu);
+ int err = 0;
+
+ mutex_lock(&cpuctx->hlist_mutex);
+
+ if (!cpuctx->swevent_hlist && cpu_online(cpu)) {
+ struct swevent_hlist *hlist;
+
+ hlist = kzalloc(sizeof(*hlist), GFP_KERNEL);
+ if (!hlist) {
+ err = -ENOMEM;
+ goto exit;
+ }
+ rcu_assign_pointer(cpuctx->swevent_hlist, hlist);
+ }
+ cpuctx->hlist_refcount++;
+ exit:
+ mutex_unlock(&cpuctx->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;
+}
+
#ifdef CONFIG_EVENT_TRACING
void perf_tp_event(int event_id, u64 addr, u64 count, void *record,
- int entry_size)
+ int entry_size, struct pt_regs *regs)
{
+ struct perf_sample_data data;
struct perf_raw_record raw = {
.size = entry_size,
.data = record,
};
- struct perf_sample_data data = {
- .addr = addr,
- .raw = &raw,
- };
-
- struct pt_regs *regs = get_irq_regs();
-
- if (!regs)
- regs = task_pt_regs(current);
+ perf_sample_data_init(&data, addr);
+ data.raw = &raw;
/* Trace events already protected against recursion */
do_perf_sw_event(PERF_TYPE_TRACEPOINT, event_id, count, 1,
- &data, regs);
+ &data, regs);
}
EXPORT_SYMBOL_GPL(perf_tp_event);
static void tp_perf_event_destroy(struct perf_event *event)
{
- ftrace_profile_disable(event->attr.config);
+ perf_trace_disable(event->attr.config);
+ swevent_hlist_put(event);
}
static const struct pmu *tp_perf_event_init(struct perf_event *event)
{
+ int err;
+
/*
* Raw tracepoint data is a severe data leak, only allow root to
* have these.
!capable(CAP_SYS_ADMIN))
return ERR_PTR(-EPERM);
- if (ftrace_profile_enable(event->attr.config))
+ if (perf_trace_enable(event->attr.config))
return NULL;
event->destroy = tp_perf_event_destroy;
+ err = swevent_hlist_get(event);
+ if (err) {
+ perf_trace_disable(event->attr.config);
+ return ERR_PTR(err);
+ }
return &perf_ops_generic;
}
struct perf_sample_data sample;
struct pt_regs *regs = data;
- sample.raw = NULL;
- sample.addr = bp->attr.bp_addr;
+ perf_sample_data_init(&sample, bp->attr.bp_addr);
if (!perf_exclude_event(bp, regs))
perf_swevent_add(bp, 1, 1, &sample, regs);
WARN_ON(event->parent);
atomic_dec(&perf_swevent_enabled[event_id]);
+ swevent_hlist_put(event);
}
static const struct pmu *sw_perf_event_init(struct perf_event *event)
case PERF_COUNT_SW_ALIGNMENT_FAULTS:
case PERF_COUNT_SW_EMULATION_FAULTS:
if (!event->parent) {
+ int err;
+
+ err = swevent_hlist_get(event);
+ if (err)
+ return ERR_PTR(err);
+
atomic_inc(&perf_swevent_enabled[event_id]);
event->destroy = sw_perf_event_destroy;
}
if (attr->type >= PERF_TYPE_MAX)
return -EINVAL;
- if (attr->__reserved_1 || attr->__reserved_2)
+ if (attr->__reserved_1)
return -EINVAL;
if (attr->sample_type & ~(PERF_SAMPLE_MAX-1))
else
child_event->state = PERF_EVENT_STATE_OFF;
- if (parent_event->attr.freq)
- child_event->hw.sample_period = parent_event->hw.sample_period;
+ 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;
+
+ atomic64_set(&hwc->period_left, sample_period);
+ }
child_event->overflow_handler = parent_event->overflow_handler;
return ret;
}
+static void __init perf_event_init_all_cpus(void)
+{
+ int cpu;
+ struct perf_cpu_context *cpuctx;
+
+ for_each_possible_cpu(cpu) {
+ cpuctx = &per_cpu(perf_cpu_context, cpu);
+ mutex_init(&cpuctx->hlist_mutex);
+ __perf_event_init_context(&cpuctx->ctx, NULL);
+ }
+}
+
static void __cpuinit perf_event_init_cpu(int cpu)
{
struct perf_cpu_context *cpuctx;
cpuctx = &per_cpu(perf_cpu_context, cpu);
- __perf_event_init_context(&cpuctx->ctx, NULL);
spin_lock(&perf_resource_lock);
cpuctx->max_pertask = perf_max_events - perf_reserved_percpu;
spin_unlock(&perf_resource_lock);
- hw_perf_event_setup(cpu);
+ mutex_lock(&cpuctx->hlist_mutex);
+ if (cpuctx->hlist_refcount > 0) {
+ struct swevent_hlist *hlist;
+
+ hlist = kzalloc(sizeof(*hlist), GFP_KERNEL);
+ WARN_ON_ONCE(!hlist);
+ rcu_assign_pointer(cpuctx->swevent_hlist, hlist);
+ }
+ mutex_unlock(&cpuctx->hlist_mutex);
}
#ifdef CONFIG_HOTPLUG_CPU
struct perf_cpu_context *cpuctx = &per_cpu(perf_cpu_context, cpu);
struct perf_event_context *ctx = &cpuctx->ctx;
+ mutex_lock(&cpuctx->hlist_mutex);
+ swevent_hlist_release(cpuctx);
+ mutex_unlock(&cpuctx->hlist_mutex);
+
mutex_lock(&ctx->mutex);
smp_call_function_single(cpu, __perf_event_exit_cpu, NULL, 1);
mutex_unlock(&ctx->mutex);
perf_event_init_cpu(cpu);
break;
- case CPU_ONLINE:
- case CPU_ONLINE_FROZEN:
- hw_perf_event_setup_online(cpu);
- break;
-
case CPU_DOWN_PREPARE:
case CPU_DOWN_PREPARE_FROZEN:
perf_event_exit_cpu(cpu);
void __init perf_event_init(void)
{
+ perf_event_init_all_cpus();
perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_UP_PREPARE,
(void *)(long)smp_processor_id());
perf_cpu_notify(&perf_cpu_nb, (unsigned long)CPU_ONLINE,
register_cpu_notifier(&perf_cpu_nb);
}
-static ssize_t perf_show_reserve_percpu(struct sysdev_class *class, char *buf)
+static ssize_t perf_show_reserve_percpu(struct sysdev_class *class,
+ struct sysdev_class_attribute *attr,
+ char *buf)
{
return sprintf(buf, "%d\n", perf_reserved_percpu);
}
static ssize_t
perf_set_reserve_percpu(struct sysdev_class *class,
+ struct sysdev_class_attribute *attr,
const char *buf,
size_t count)
{
return count;
}
-static ssize_t perf_show_overcommit(struct sysdev_class *class, char *buf)
+static ssize_t perf_show_overcommit(struct sysdev_class *class,
+ struct sysdev_class_attribute *attr,
+ char *buf)
{
return sprintf(buf, "%d\n", perf_overcommit);
}
static ssize_t
-perf_set_overcommit(struct sysdev_class *class, const char *buf, size_t count)
+perf_set_overcommit(struct sysdev_class *class,
+ struct sysdev_class_attribute *attr,
+ const char *buf, size_t count)
{
unsigned long val;
int err;