#include <linux/kernel_stat.h>
#include <linux/perf_event.h>
#include <linux/ftrace_event.h>
-#include <linux/hw_breakpoint.h>
#include <asm/irq_regs.h>
*/
static DEFINE_SPINLOCK(perf_resource_lock);
-/*
- * Architecture provided APIs - weak aliases:
- */
-extern __weak const struct pmu *hw_perf_event_init(struct perf_event *event)
-{
- return NULL;
-}
-
void __weak hw_perf_disable(void) { barrier(); }
void __weak hw_perf_enable(void) { barrier(); }
struct perf_event_context *ctx;
rcu_read_lock();
- retry:
+retry:
ctx = rcu_dereference(task->perf_event_ctxp);
if (ctx) {
/*
static inline u64 perf_clock(void)
{
- return cpu_clock(raw_smp_processor_id());
+ return local_clock();
}
/*
}
}
+static inline int
+event_filter_match(struct perf_event *event)
+{
+ return event->cpu == -1 || event->cpu == smp_processor_id();
+}
+
static void
event_sched_out(struct perf_event *event,
struct perf_cpu_context *cpuctx,
struct perf_event_context *ctx)
{
+ 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 = ctx->time - event->tstamp_stopped;
+ event->tstamp_running += delta;
+ event->tstamp_stopped = ctx->time;
+ }
+
if (event->state != PERF_EVENT_STATE_ACTIVE)
return;
struct perf_event_context *ctx)
{
struct perf_event *event;
-
- if (group_event->state != PERF_EVENT_STATE_ACTIVE)
- return;
+ int state = group_event->state;
event_sched_out(group_event, cpuctx, ctx);
list_for_each_entry(event, &group_event->sibling_list, group_entry)
event_sched_out(event, cpuctx, ctx);
- if (group_event->attr.exclusive)
+ if (state == PERF_EVENT_STATE_ACTIVE && group_event->attr.exclusive)
cpuctx->exclusive = 0;
}
return;
raw_spin_lock(&ctx->lock);
- /*
- * Protect the list operation against NMI by disabling the
- * events on a global level.
- */
- perf_disable();
event_sched_out(event, cpuctx, ctx);
perf_max_events - perf_reserved_percpu);
}
- perf_enable();
raw_spin_unlock(&ctx->lock);
}
return;
}
- retry:
+retry:
task_oncpu_function_call(task, __perf_event_disable, event);
raw_spin_lock_irq(&ctx->lock);
struct perf_event_context *ctx)
{
struct perf_event *event, *partial_group = NULL;
- const struct pmu *pmu = group_event->pmu;
+ struct pmu *pmu = group_event->pmu;
bool txn = false;
- int ret;
if (group_event->state == PERF_EVENT_STATE_OFF)
return 0;
}
}
- if (!txn)
+ if (!txn || !pmu->commit_txn(pmu))
return 0;
- ret = pmu->commit_txn(pmu);
- if (!ret) {
- pmu->cancel_txn(pmu);
- return 0;
- }
-
group_error:
/*
* Groups can be scheduled in as one unit only, so undo any
ctx->is_active = 1;
update_context_time(ctx);
- /*
- * Protect the list operation against NMI by disabling the
- * events on a global level. NOP for non NMI based events.
- */
- perf_disable();
-
add_event_to_ctx(event, ctx);
if (event->cpu != -1 && event->cpu != smp_processor_id())
if (!err && !ctx->task && cpuctx->max_pertask)
cpuctx->max_pertask--;
- unlock:
- perf_enable();
-
+unlock:
raw_spin_unlock(&ctx->lock);
}
event->state = PERF_EVENT_STATE_INACTIVE;
event->tstamp_enabled = ctx->time - event->total_time_enabled;
- list_for_each_entry(sub, &event->sibling_list, group_entry)
- if (sub->state >= PERF_EVENT_STATE_INACTIVE)
+ list_for_each_entry(sub, &event->sibling_list, group_entry) {
+ if (sub->state >= PERF_EVENT_STATE_INACTIVE) {
sub->tstamp_enabled =
ctx->time - sub->total_time_enabled;
+ }
+ }
}
/*
if (!group_can_go_on(event, cpuctx, 1)) {
err = -EEXIST;
} else {
- perf_disable();
if (event == leader)
err = group_sched_in(event, cpuctx, ctx);
else
err = event_sched_in(event, cpuctx, ctx);
- perf_enable();
}
if (err) {
}
}
- unlock:
+unlock:
raw_spin_unlock(&ctx->lock);
}
if (event->state == PERF_EVENT_STATE_ERROR)
event->state = PERF_EVENT_STATE_OFF;
- retry:
+retry:
raw_spin_unlock_irq(&ctx->lock);
task_oncpu_function_call(task, __perf_event_enable, event);
if (event->state == PERF_EVENT_STATE_OFF)
__perf_event_mark_enabled(event, ctx);
- out:
+out:
raw_spin_unlock_irq(&ctx->lock);
}
goto out;
update_context_time(ctx);
- perf_disable();
if (!ctx->nr_active)
- goto out_enable;
+ goto out;
- if (event_type & EVENT_PINNED)
+ 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)
+ 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:
+ }
+out:
raw_spin_unlock(&ctx->lock);
}
* In order to keep per-task stats reliable we need to flip the event
* values when we flip the contexts.
*/
- value = atomic64_read(&next_event->count);
- value = atomic64_xchg(&event->count, value);
- atomic64_set(&next_event->count, value);
+ 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);
if (event->cpu != -1 && event->cpu != smp_processor_id())
continue;
- if (group_can_go_on(event, cpuctx, can_add_hw))
+ if (group_can_go_on(event, cpuctx, can_add_hw)) {
if (group_sched_in(event, cpuctx, ctx))
can_add_hw = 0;
+ }
}
}
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_FLEXIBLE)
ctx_flexible_sched_in(ctx, cpuctx);
- perf_enable();
- out:
+out:
raw_spin_unlock(&ctx->lock);
}
if (cpuctx->task_ctx == ctx)
return;
- perf_disable();
-
/*
* We want to keep the following priority order:
* cpu pinned (that don't need to move), task pinned,
ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE);
cpuctx->task_ctx = ctx;
-
- perf_enable();
}
#define MAX_INTERRUPTS (~0ULL)
hwc->sample_period = sample_period;
- if (atomic64_read(&hwc->period_left) > 8*sample_period) {
- perf_disable();
+ if (local64_read(&hwc->period_left) > 8*sample_period) {
perf_event_stop(event);
- atomic64_set(&hwc->period_left, 0);
+ local64_set(&hwc->period_left, 0);
perf_event_start(event);
- perf_enable();
}
}
*/
if (interrupts == MAX_INTERRUPTS) {
perf_log_throttle(event, 1);
- perf_disable();
event->pmu->unthrottle(event);
- perf_enable();
}
if (!event->attr.freq || !event->attr.sample_freq)
continue;
- perf_disable();
event->pmu->read(event);
- now = atomic64_read(&event->count);
+ now = local64_read(&event->count);
delta = now - hwc->freq_count_stamp;
hwc->freq_count_stamp = now;
if (delta > 0)
perf_adjust_period(event, TICK_NSEC, delta);
- perf_enable();
}
raw_spin_unlock(&ctx->lock);
}
if (!rotate)
return;
- perf_disable();
cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE);
if (ctx)
task_ctx_sched_out(ctx, EVENT_FLEXIBLE);
cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE);
if (ctx)
task_ctx_sched_in(curr, EVENT_FLEXIBLE);
- perf_enable();
}
static int event_enable_on_exec(struct perf_event *event,
raw_spin_unlock(&ctx->lock);
perf_event_task_sched_in(task);
- out:
+out:
local_irq_restore(flags);
}
event->pmu->read(event);
}
+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)
{
/*
raw_spin_unlock_irqrestore(&ctx->lock, flags);
}
- return atomic64_read(&event->count);
+ 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 = sizeof(*entries) + sizeof(struct perf_callchain_entry *) *
+ num_possible_cpus();
+
+ 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;
}
/*
if (!ptrace_may_access(task, PTRACE_MODE_READ))
goto errout;
- retry:
+retry:
ctx = perf_lock_task_context(task, &flags);
if (ctx) {
unclone_ctx(ctx);
put_task_struct(task);
return ctx;
- errout:
+errout:
put_task_struct(task);
return ERR_PTR(err);
}
}
static void perf_pending_sync(struct perf_event *event);
-static void perf_mmap_data_put(struct perf_mmap_data *data);
+static void perf_buffer_put(struct perf_buffer *buffer);
static void free_event(struct perf_event *event)
{
if (!event->parent) {
atomic_dec(&nr_events);
- if (event->attr.mmap)
+ 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 (event->data) {
- perf_mmap_data_put(event->data);
- event->data = NULL;
+ if (event->buffer) {
+ perf_buffer_put(event->buffer);
+ event->buffer = NULL;
}
if (event->destroy)
static unsigned int perf_poll(struct file *file, poll_table *wait)
{
struct perf_event *event = file->private_data;
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
unsigned int events = POLL_HUP;
rcu_read_lock();
- data = rcu_dereference(event->data);
- if (data)
- events = atomic_xchg(&data->poll, 0);
+ buffer = rcu_dereference(event->buffer);
+ if (buffer)
+ events = atomic_xchg(&buffer->poll, 0);
rcu_read_unlock();
poll_wait(file, &event->waitq, wait);
static void perf_event_reset(struct perf_event *event)
{
(void)perf_event_read(event);
- atomic64_set(&event->count, 0);
+ local64_set(&event->count, 0);
perf_event_update_userpage(event);
}
void perf_event_update_userpage(struct perf_event *event)
{
struct perf_event_mmap_page *userpg;
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
rcu_read_lock();
- data = rcu_dereference(event->data);
- if (!data)
+ buffer = rcu_dereference(event->buffer);
+ if (!buffer)
goto unlock;
- userpg = data->user_page;
+ userpg = buffer->user_page;
/*
* Disable preemption so as to not let the corresponding user-space
++userpg->lock;
barrier();
userpg->index = perf_event_index(event);
- userpg->offset = atomic64_read(&event->count);
+ userpg->offset = perf_event_count(event);
if (event->state == PERF_EVENT_STATE_ACTIVE)
- userpg->offset -= atomic64_read(&event->hw.prev_count);
+ userpg->offset -= local64_read(&event->hw.prev_count);
userpg->time_enabled = event->total_time_enabled +
atomic64_read(&event->child_total_time_enabled);
rcu_read_unlock();
}
+static unsigned long perf_data_size(struct perf_buffer *buffer);
+
+static void
+perf_buffer_init(struct perf_buffer *buffer, long watermark, int flags)
+{
+ long max_size = perf_data_size(buffer);
+
+ if (watermark)
+ buffer->watermark = min(max_size, watermark);
+
+ if (!buffer->watermark)
+ buffer->watermark = max_size / 2;
+
+ if (flags & PERF_BUFFER_WRITABLE)
+ buffer->writable = 1;
+
+ atomic_set(&buffer->refcount, 1);
+}
+
#ifndef CONFIG_PERF_USE_VMALLOC
/*
*/
static struct page *
-perf_mmap_to_page(struct perf_mmap_data *data, unsigned long pgoff)
+perf_mmap_to_page(struct perf_buffer *buffer, unsigned long pgoff)
{
- if (pgoff > data->nr_pages)
+ if (pgoff > buffer->nr_pages)
return NULL;
if (pgoff == 0)
- return virt_to_page(data->user_page);
+ return virt_to_page(buffer->user_page);
- return virt_to_page(data->data_pages[pgoff - 1]);
+ return virt_to_page(buffer->data_pages[pgoff - 1]);
}
static void *perf_mmap_alloc_page(int cpu)
return page_address(page);
}
-static struct perf_mmap_data *
-perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
+static struct perf_buffer *
+perf_buffer_alloc(int nr_pages, long watermark, int cpu, int flags)
{
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
unsigned long size;
int i;
- size = sizeof(struct perf_mmap_data);
+ size = sizeof(struct perf_buffer);
size += nr_pages * sizeof(void *);
- data = kzalloc(size, GFP_KERNEL);
- if (!data)
+ buffer = kzalloc(size, GFP_KERNEL);
+ if (!buffer)
goto fail;
- data->user_page = perf_mmap_alloc_page(event->cpu);
- if (!data->user_page)
+ buffer->user_page = perf_mmap_alloc_page(cpu);
+ if (!buffer->user_page)
goto fail_user_page;
for (i = 0; i < nr_pages; i++) {
- data->data_pages[i] = perf_mmap_alloc_page(event->cpu);
- if (!data->data_pages[i])
+ buffer->data_pages[i] = perf_mmap_alloc_page(cpu);
+ if (!buffer->data_pages[i])
goto fail_data_pages;
}
- data->nr_pages = nr_pages;
+ buffer->nr_pages = nr_pages;
+
+ perf_buffer_init(buffer, watermark, flags);
- return data;
+ return buffer;
fail_data_pages:
for (i--; i >= 0; i--)
- free_page((unsigned long)data->data_pages[i]);
+ free_page((unsigned long)buffer->data_pages[i]);
- free_page((unsigned long)data->user_page);
+ free_page((unsigned long)buffer->user_page);
fail_user_page:
- kfree(data);
+ kfree(buffer);
fail:
return NULL;
__free_page(page);
}
-static void perf_mmap_data_free(struct perf_mmap_data *data)
+static void perf_buffer_free(struct perf_buffer *buffer)
{
int i;
- perf_mmap_free_page((unsigned long)data->user_page);
- for (i = 0; i < data->nr_pages; i++)
- perf_mmap_free_page((unsigned long)data->data_pages[i]);
- kfree(data);
+ perf_mmap_free_page((unsigned long)buffer->user_page);
+ for (i = 0; i < buffer->nr_pages; i++)
+ perf_mmap_free_page((unsigned long)buffer->data_pages[i]);
+ kfree(buffer);
}
-static inline int page_order(struct perf_mmap_data *data)
+static inline int page_order(struct perf_buffer *buffer)
{
return 0;
}
* Required for architectures that have d-cache aliasing issues.
*/
-static inline int page_order(struct perf_mmap_data *data)
+static inline int page_order(struct perf_buffer *buffer)
{
- return data->page_order;
+ return buffer->page_order;
}
static struct page *
-perf_mmap_to_page(struct perf_mmap_data *data, unsigned long pgoff)
+perf_mmap_to_page(struct perf_buffer *buffer, unsigned long pgoff)
{
- if (pgoff > (1UL << page_order(data)))
+ if (pgoff > (1UL << page_order(buffer)))
return NULL;
- return vmalloc_to_page((void *)data->user_page + pgoff * PAGE_SIZE);
+ return vmalloc_to_page((void *)buffer->user_page + pgoff * PAGE_SIZE);
}
static void perf_mmap_unmark_page(void *addr)
page->mapping = NULL;
}
-static void perf_mmap_data_free_work(struct work_struct *work)
+static void perf_buffer_free_work(struct work_struct *work)
{
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
void *base;
int i, nr;
- data = container_of(work, struct perf_mmap_data, work);
- nr = 1 << page_order(data);
+ buffer = container_of(work, struct perf_buffer, work);
+ nr = 1 << page_order(buffer);
- base = data->user_page;
+ base = buffer->user_page;
for (i = 0; i < nr + 1; i++)
perf_mmap_unmark_page(base + (i * PAGE_SIZE));
vfree(base);
- kfree(data);
+ kfree(buffer);
}
-static void perf_mmap_data_free(struct perf_mmap_data *data)
+static void perf_buffer_free(struct perf_buffer *buffer)
{
- schedule_work(&data->work);
+ schedule_work(&buffer->work);
}
-static struct perf_mmap_data *
-perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
+static struct perf_buffer *
+perf_buffer_alloc(int nr_pages, long watermark, int cpu, int flags)
{
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
unsigned long size;
void *all_buf;
- size = sizeof(struct perf_mmap_data);
+ size = sizeof(struct perf_buffer);
size += sizeof(void *);
- data = kzalloc(size, GFP_KERNEL);
- if (!data)
+ buffer = kzalloc(size, GFP_KERNEL);
+ if (!buffer)
goto fail;
- INIT_WORK(&data->work, perf_mmap_data_free_work);
+ INIT_WORK(&buffer->work, perf_buffer_free_work);
all_buf = vmalloc_user((nr_pages + 1) * PAGE_SIZE);
if (!all_buf)
goto fail_all_buf;
- data->user_page = all_buf;
- data->data_pages[0] = all_buf + PAGE_SIZE;
- data->page_order = ilog2(nr_pages);
- data->nr_pages = 1;
+ buffer->user_page = all_buf;
+ buffer->data_pages[0] = all_buf + PAGE_SIZE;
+ buffer->page_order = ilog2(nr_pages);
+ buffer->nr_pages = 1;
- return data;
+ perf_buffer_init(buffer, watermark, flags);
+
+ return buffer;
fail_all_buf:
- kfree(data);
+ kfree(buffer);
fail:
return NULL;
#endif
-static unsigned long perf_data_size(struct perf_mmap_data *data)
+static unsigned long perf_data_size(struct perf_buffer *buffer)
{
- return data->nr_pages << (PAGE_SHIFT + page_order(data));
+ return buffer->nr_pages << (PAGE_SHIFT + page_order(buffer));
}
static int perf_mmap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
struct perf_event *event = vma->vm_file->private_data;
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
int ret = VM_FAULT_SIGBUS;
if (vmf->flags & FAULT_FLAG_MKWRITE) {
}
rcu_read_lock();
- data = rcu_dereference(event->data);
- if (!data)
+ buffer = rcu_dereference(event->buffer);
+ if (!buffer)
goto unlock;
if (vmf->pgoff && (vmf->flags & FAULT_FLAG_WRITE))
goto unlock;
- vmf->page = perf_mmap_to_page(data, vmf->pgoff);
+ vmf->page = perf_mmap_to_page(buffer, vmf->pgoff);
if (!vmf->page)
goto unlock;
return ret;
}
-static void
-perf_mmap_data_init(struct perf_event *event, struct perf_mmap_data *data)
-{
- long max_size = perf_data_size(data);
-
- if (event->attr.watermark) {
- data->watermark = min_t(long, max_size,
- event->attr.wakeup_watermark);
- }
-
- if (!data->watermark)
- data->watermark = max_size / 2;
-
- atomic_set(&data->refcount, 1);
- rcu_assign_pointer(event->data, data);
-}
-
-static void perf_mmap_data_free_rcu(struct rcu_head *rcu_head)
+static void perf_buffer_free_rcu(struct rcu_head *rcu_head)
{
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
- data = container_of(rcu_head, struct perf_mmap_data, rcu_head);
- perf_mmap_data_free(data);
+ buffer = container_of(rcu_head, struct perf_buffer, rcu_head);
+ perf_buffer_free(buffer);
}
-static struct perf_mmap_data *perf_mmap_data_get(struct perf_event *event)
+static struct perf_buffer *perf_buffer_get(struct perf_event *event)
{
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
rcu_read_lock();
- data = rcu_dereference(event->data);
- if (data) {
- if (!atomic_inc_not_zero(&data->refcount))
- data = NULL;
+ buffer = rcu_dereference(event->buffer);
+ if (buffer) {
+ if (!atomic_inc_not_zero(&buffer->refcount))
+ buffer = NULL;
}
rcu_read_unlock();
- return data;
+ return buffer;
}
-static void perf_mmap_data_put(struct perf_mmap_data *data)
+static void perf_buffer_put(struct perf_buffer *buffer)
{
- if (!atomic_dec_and_test(&data->refcount))
+ if (!atomic_dec_and_test(&buffer->refcount))
return;
- call_rcu(&data->rcu_head, perf_mmap_data_free_rcu);
+ call_rcu(&buffer->rcu_head, perf_buffer_free_rcu);
}
static void perf_mmap_open(struct vm_area_struct *vma)
struct perf_event *event = vma->vm_file->private_data;
if (atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex)) {
- unsigned long size = perf_data_size(event->data);
+ unsigned long size = perf_data_size(event->buffer);
struct user_struct *user = event->mmap_user;
- struct perf_mmap_data *data = event->data;
+ struct perf_buffer *buffer = event->buffer;
atomic_long_sub((size >> PAGE_SHIFT) + 1, &user->locked_vm);
vma->vm_mm->locked_vm -= event->mmap_locked;
- rcu_assign_pointer(event->data, NULL);
+ rcu_assign_pointer(event->buffer, NULL);
mutex_unlock(&event->mmap_mutex);
- perf_mmap_data_put(data);
+ perf_buffer_put(buffer);
free_uid(user);
}
}
unsigned long user_locked, user_lock_limit;
struct user_struct *user = current_user();
unsigned long locked, lock_limit;
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
unsigned long vma_size;
unsigned long nr_pages;
long user_extra, extra;
- int ret = 0;
+ int ret = 0, flags = 0;
/*
* Don't allow mmap() of inherited per-task counters. This would
nr_pages = (vma_size / PAGE_SIZE) - 1;
/*
- * If we have data pages ensure they're a power-of-two number, so we
+ * If we have buffer 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))
WARN_ON_ONCE(event->ctx->parent_ctx);
mutex_lock(&event->mmap_mutex);
- if (event->data) {
- if (event->data->nr_pages == nr_pages)
- atomic_inc(&event->data->refcount);
+ if (event->buffer) {
+ if (event->buffer->nr_pages == nr_pages)
+ atomic_inc(&event->buffer->refcount);
else
ret = -EINVAL;
goto unlock;
goto unlock;
}
- WARN_ON(event->data);
+ WARN_ON(event->buffer);
+
+ if (vma->vm_flags & VM_WRITE)
+ flags |= PERF_BUFFER_WRITABLE;
- data = perf_mmap_data_alloc(event, nr_pages);
- if (!data) {
+ buffer = perf_buffer_alloc(nr_pages, event->attr.wakeup_watermark,
+ event->cpu, flags);
+ if (!buffer) {
ret = -ENOMEM;
goto unlock;
}
-
- perf_mmap_data_init(event, data);
- if (vma->vm_flags & VM_WRITE)
- event->data->writable = 1;
+ rcu_assign_pointer(event->buffer, buffer);
atomic_long_add(user_extra, &user->locked_vm);
event->mmap_locked = extra;
__perf_pending_run();
}
-/*
- * Callchain support -- arch specific
- */
-
-__weak struct perf_callchain_entry *perf_callchain(struct pt_regs *regs)
-{
- 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
/*
* Output
*/
-static bool perf_output_space(struct perf_mmap_data *data, unsigned long tail,
+static bool perf_output_space(struct perf_buffer *buffer, unsigned long tail,
unsigned long offset, unsigned long head)
{
unsigned long mask;
- if (!data->writable)
+ if (!buffer->writable)
return true;
- mask = perf_data_size(data) - 1;
+ mask = perf_data_size(buffer) - 1;
offset = (offset - tail) & mask;
head = (head - tail) & mask;
static void perf_output_wakeup(struct perf_output_handle *handle)
{
- atomic_set(&handle->data->poll, POLL_IN);
+ atomic_set(&handle->buffer->poll, POLL_IN);
if (handle->nmi) {
handle->event->pending_wakeup = 1;
*/
static void perf_output_get_handle(struct perf_output_handle *handle)
{
- struct perf_mmap_data *data = handle->data;
+ struct perf_buffer *buffer = handle->buffer;
preempt_disable();
- local_inc(&data->nest);
- handle->wakeup = local_read(&data->wakeup);
+ local_inc(&buffer->nest);
+ handle->wakeup = local_read(&buffer->wakeup);
}
static void perf_output_put_handle(struct perf_output_handle *handle)
{
- struct perf_mmap_data *data = handle->data;
+ struct perf_buffer *buffer = handle->buffer;
unsigned long head;
again:
- head = local_read(&data->head);
+ head = local_read(&buffer->head);
/*
* IRQ/NMI can happen here, which means we can miss a head update.
*/
- if (!local_dec_and_test(&data->nest))
+ if (!local_dec_and_test(&buffer->nest))
goto out;
/*
* Publish the known good head. Rely on the full barrier implied
- * by atomic_dec_and_test() order the data->head read and this
+ * by atomic_dec_and_test() order the buffer->head read and this
* write.
*/
- data->user_page->data_head = head;
+ buffer->user_page->data_head = head;
/*
* Now check if we missed an update, rely on the (compiler)
- * barrier in atomic_dec_and_test() to re-read data->head.
+ * barrier in atomic_dec_and_test() to re-read buffer->head.
*/
- if (unlikely(head != local_read(&data->head))) {
- local_inc(&data->nest);
+ if (unlikely(head != local_read(&buffer->head))) {
+ local_inc(&buffer->nest);
goto again;
}
- if (handle->wakeup != local_read(&data->wakeup))
+ if (handle->wakeup != local_read(&buffer->wakeup))
perf_output_wakeup(handle);
- out:
+out:
preempt_enable();
}
buf += size;
handle->size -= size;
if (!handle->size) {
- struct perf_mmap_data *data = handle->data;
+ struct perf_buffer *buffer = handle->buffer;
handle->page++;
- handle->page &= data->nr_pages - 1;
- handle->addr = data->data_pages[handle->page];
- handle->size = PAGE_SIZE << page_order(data);
+ handle->page &= buffer->nr_pages - 1;
+ handle->addr = buffer->data_pages[handle->page];
+ handle->size = PAGE_SIZE << page_order(buffer);
}
} while (len);
}
struct perf_event *event, unsigned int size,
int nmi, int sample)
{
- struct perf_mmap_data *data;
+ struct perf_buffer *buffer;
unsigned long tail, offset, head;
int have_lost;
struct {
if (event->parent)
event = event->parent;
- data = rcu_dereference(event->data);
- if (!data)
+ buffer = rcu_dereference(event->buffer);
+ if (!buffer)
goto out;
- handle->data = data;
+ handle->buffer = buffer;
handle->event = event;
handle->nmi = nmi;
handle->sample = sample;
- if (!data->nr_pages)
+ if (!buffer->nr_pages)
goto out;
- have_lost = local_read(&data->lost);
+ have_lost = local_read(&buffer->lost);
if (have_lost)
size += sizeof(lost_event);
* tail pointer. So that all reads will be completed before the
* write is issued.
*/
- tail = ACCESS_ONCE(data->user_page->data_tail);
+ tail = ACCESS_ONCE(buffer->user_page->data_tail);
smp_rmb();
- offset = head = local_read(&data->head);
+ offset = head = local_read(&buffer->head);
head += size;
- if (unlikely(!perf_output_space(data, tail, offset, head)))
+ if (unlikely(!perf_output_space(buffer, tail, offset, head)))
goto fail;
- } while (local_cmpxchg(&data->head, offset, head) != offset);
+ } while (local_cmpxchg(&buffer->head, offset, head) != offset);
- if (head - local_read(&data->wakeup) > data->watermark)
- local_add(data->watermark, &data->wakeup);
+ if (head - local_read(&buffer->wakeup) > buffer->watermark)
+ local_add(buffer->watermark, &buffer->wakeup);
- handle->page = offset >> (PAGE_SHIFT + page_order(data));
- handle->page &= data->nr_pages - 1;
- handle->size = offset & ((PAGE_SIZE << page_order(data)) - 1);
- handle->addr = data->data_pages[handle->page];
+ handle->page = offset >> (PAGE_SHIFT + page_order(buffer));
+ handle->page &= buffer->nr_pages - 1;
+ handle->size = offset & ((PAGE_SIZE << page_order(buffer)) - 1);
+ handle->addr = buffer->data_pages[handle->page];
handle->addr += handle->size;
- handle->size = (PAGE_SIZE << page_order(data)) - handle->size;
+ handle->size = (PAGE_SIZE << page_order(buffer)) - handle->size;
if (have_lost) {
lost_event.header.type = PERF_RECORD_LOST;
lost_event.header.misc = 0;
lost_event.header.size = sizeof(lost_event);
lost_event.id = event->id;
- lost_event.lost = local_xchg(&data->lost, 0);
+ lost_event.lost = local_xchg(&buffer->lost, 0);
perf_output_put(handle, lost_event);
}
return 0;
fail:
- local_inc(&data->lost);
+ local_inc(&buffer->lost);
perf_output_put_handle(handle);
out:
rcu_read_unlock();
void perf_output_end(struct perf_output_handle *handle)
{
struct perf_event *event = handle->event;
- struct perf_mmap_data *data = handle->data;
+ struct perf_buffer *buffer = handle->buffer;
int wakeup_events = event->attr.wakeup_events;
if (handle->sample && wakeup_events) {
- int events = local_inc_return(&data->events);
+ int events = local_inc_return(&buffer->events);
if (events >= wakeup_events) {
- local_sub(wakeup_events, &data->events);
- local_inc(&data->wakeup);
+ local_sub(wakeup_events, &buffer->events);
+ local_inc(&buffer->wakeup);
}
}
u64 values[4];
int n = 0;
- values[n++] = atomic64_read(&event->count);
+ values[n++] = perf_event_count(event);
if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) {
values[n++] = event->total_time_enabled +
atomic64_read(&event->child_total_time_enabled);
if (leader != event)
leader->pmu->read(leader);
- values[n++] = atomic64_read(&leader->count);
+ values[n++] = perf_event_count(leader);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(leader);
if (sub != event)
sub->pmu->read(sub);
- values[n++] = atomic64_read(&sub->count);
+ values[n++] = perf_event_count(sub);
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(sub);
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, nmi, 1))
- return;
+ goto exit;
perf_output_sample(&handle, &header, data, event);
perf_output_end(&handle);
+
+exit:
+ rcu_read_unlock();
}
/*
/*
* task tracking -- fork/exit
*
- * enabled by: attr.comm | attr.mmap | attr.task
+ * enabled by: attr.comm | attr.mmap | attr.mmap_data | attr.task
*/
struct perf_task_event {
if (event->cpu != -1 && event->cpu != smp_processor_id())
return 0;
- if (event->attr.comm || event->attr.mmap || event->attr.task)
+ if (event->attr.comm || event->attr.mmap ||
+ event->attr.mmap_data || event->attr.task)
return 1;
return 0;
}
static int perf_event_mmap_match(struct perf_event *event,
- struct perf_mmap_event *mmap_event)
+ struct perf_mmap_event *mmap_event,
+ int executable)
{
if (event->state < PERF_EVENT_STATE_INACTIVE)
return 0;
if (event->cpu != -1 && event->cpu != smp_processor_id())
return 0;
- if (event->attr.mmap)
+ 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)
+ 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))
+ if (perf_event_mmap_match(event, mmap_event, executable))
perf_event_mmap_output(event, mmap_event);
}
}
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));
rcu_read_lock();
cpuctx = &get_cpu_var(perf_cpu_context);
- perf_event_mmap_ctx(&cpuctx->ctx, mmap_event);
+ perf_event_mmap_ctx(&cpuctx->ctx, mmap_event, vma->vm_flags & VM_EXEC);
ctx = rcu_dereference(current->perf_event_ctxp);
if (ctx)
- perf_event_mmap_ctx(ctx, mmap_event);
+ perf_event_mmap_ctx(ctx, mmap_event, vma->vm_flags & VM_EXEC);
put_cpu_var(perf_cpu_context);
rcu_read_unlock();
kfree(buf);
}
-void __perf_event_mmap(struct vm_area_struct *vma)
+void perf_event_mmap(struct vm_area_struct *vma)
{
struct perf_mmap_event mmap_event;
hwc->last_period = hwc->sample_period;
again:
- old = val = atomic64_read(&hwc->period_left);
+ old = val = local64_read(&hwc->period_left);
if (val < 0)
return 0;
nr = div64_u64(period + val, period);
offset = nr * period;
val -= offset;
- if (atomic64_cmpxchg(&hwc->period_left, old, val) != old)
+ if (local64_cmpxchg(&hwc->period_left, old, val) != old)
goto again;
return nr;
{
struct hw_perf_event *hwc = &event->hw;
- atomic64_add(nr, &event->count);
+ local64_add(nr, &event->count);
if (!regs)
return;
if (nr == 1 && hwc->sample_period == 1 && !event->attr.freq)
return perf_swevent_overflow(event, 1, nmi, data, regs);
- if (atomic64_add_negative(nr, &hwc->period_left))
+ if (local64_add_negative(nr, &hwc->period_left))
return;
perf_swevent_overflow(event, 0, nmi, data, regs);
int perf_swevent_get_recursion_context(void)
{
struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
- int rctx;
-
- if (in_nmi())
- rctx = 3;
- else if (in_irq())
- rctx = 2;
- else if (in_softirq())
- rctx = 1;
- else
- rctx = 0;
-
- if (cpuctx->recursion[rctx])
- return -1;
-
- cpuctx->recursion[rctx]++;
- barrier();
- return rctx;
+ return get_recursion_context(cpuctx->recursion);
}
EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context);
-void perf_swevent_put_recursion_context(int rctx)
+void inline perf_swevent_put_recursion_context(int rctx)
{
struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
- barrier();
- cpuctx->recursion[rctx]--;
-}
-EXPORT_SYMBOL_GPL(perf_swevent_put_recursion_context);
+ put_recursion_context(cpuctx->recursion, rctx);
+}
void __perf_sw_event(u32 event_id, u64 nr, int nmi,
struct pt_regs *regs, u64 addr)
return 0;
}
-static const struct pmu perf_ops_generic = {
- .enable = perf_swevent_enable,
- .disable = perf_swevent_disable,
- .start = perf_swevent_int,
- .stop = perf_swevent_void,
- .read = perf_swevent_read,
- .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */
-};
-
-/*
- * hrtimer based swevent callback
- */
-
-static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer)
+/* Deref the hlist from the update side */
+static inline struct swevent_hlist *
+swevent_hlist_deref(struct perf_cpu_context *cpuctx)
{
- 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);
- 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, 0, &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;
-
- hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
- hwc->hrtimer.function = perf_swevent_hrtimer;
- if (hwc->sample_period) {
- u64 period;
-
- if (hwc->remaining) {
- if (hwc->remaining < 0)
- period = 10000;
- else
- period = hwc->remaining;
- hwc->remaining = 0;
- } else {
- period = max_t(u64, 10000, hwc->sample_period);
- }
- __hrtimer_start_range_ns(&hwc->hrtimer,
- ns_to_ktime(period), 0,
- HRTIMER_MODE_REL, 0);
- }
-}
-
-static void perf_swevent_cancel_hrtimer(struct perf_event *event)
-{
- struct hw_perf_event *hwc = &event->hw;
-
- if (hwc->sample_period) {
- ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer);
- hwc->remaining = ktime_to_ns(remaining);
-
- hrtimer_cancel(&hwc->hrtimer);
- }
-}
-
-/*
- * Software event: cpu wall time clock
- */
-
-static void cpu_clock_perf_event_update(struct perf_event *event)
-{
- int cpu = raw_smp_processor_id();
- s64 prev;
- u64 now;
-
- now = cpu_clock(cpu);
- prev = atomic64_xchg(&event->hw.prev_count, now);
- atomic64_add(now - prev, &event->count);
-}
-
-static int cpu_clock_perf_event_enable(struct perf_event *event)
-{
- struct hw_perf_event *hwc = &event->hw;
- int cpu = raw_smp_processor_id();
-
- atomic64_set(&hwc->prev_count, cpu_clock(cpu));
- perf_swevent_start_hrtimer(event);
-
- return 0;
-}
-
-static void cpu_clock_perf_event_disable(struct perf_event *event)
-{
- perf_swevent_cancel_hrtimer(event);
- cpu_clock_perf_event_update(event);
-}
-
-static void cpu_clock_perf_event_read(struct perf_event *event)
-{
- cpu_clock_perf_event_update(event);
-}
-
-static const struct pmu perf_ops_cpu_clock = {
- .enable = cpu_clock_perf_event_enable,
- .disable = cpu_clock_perf_event_disable,
- .read = cpu_clock_perf_event_read,
-};
-
-/*
- * Software event: task time clock
- */
-
-static void task_clock_perf_event_update(struct perf_event *event, u64 now)
-{
- u64 prev;
- s64 delta;
-
- prev = atomic64_xchg(&event->hw.prev_count, now);
- delta = now - prev;
- atomic64_add(delta, &event->count);
-}
-
-static int task_clock_perf_event_enable(struct perf_event *event)
-{
- struct hw_perf_event *hwc = &event->hw;
- u64 now;
-
- now = event->ctx->time;
-
- atomic64_set(&hwc->prev_count, now);
-
- perf_swevent_start_hrtimer(event);
-
- return 0;
-}
-
-static void task_clock_perf_event_disable(struct perf_event *event)
-{
- perf_swevent_cancel_hrtimer(event);
- task_clock_perf_event_update(event, event->ctx->time);
-
-}
-
-static void task_clock_perf_event_read(struct perf_event *event)
-{
- u64 time;
-
- if (!in_nmi()) {
- update_context_time(event->ctx);
- time = event->ctx->time;
- } else {
- u64 now = perf_clock();
- u64 delta = now - event->ctx->timestamp;
- time = event->ctx->time + delta;
- }
-
- task_clock_perf_event_update(event, time);
-}
-
-static const struct pmu perf_ops_task_clock = {
- .enable = task_clock_perf_event_enable,
- .disable = task_clock_perf_event_disable,
- .read = task_clock_perf_event_read,
-};
-
-/* Deref the hlist from the update side */
-static inline struct swevent_hlist *
-swevent_hlist_deref(struct perf_cpu_context *cpuctx)
-{
- return rcu_dereference_protected(cpuctx->swevent_hlist,
- lockdep_is_held(&cpuctx->hlist_mutex));
-}
+ return rcu_dereference_protected(cpuctx->swevent_hlist,
+ lockdep_is_held(&cpuctx->hlist_mutex));
+}
static void swevent_hlist_release_rcu(struct rcu_head *rcu_head)
{
rcu_assign_pointer(cpuctx->swevent_hlist, hlist);
}
cpuctx->hlist_refcount++;
- exit:
+exit:
mutex_unlock(&cpuctx->hlist_mutex);
return err;
put_online_cpus();
return 0;
- fail:
+fail:
for_each_possible_cpu(cpu) {
if (cpu == failed_cpu)
break;
return err;
}
-#ifdef CONFIG_EVENT_TRACING
+atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX];
-static const struct pmu perf_ops_tracepoint = {
- .enable = perf_trace_enable,
- .disable = perf_trace_disable,
+static void sw_perf_event_destroy(struct perf_event *event)
+{
+ u64 event_id = event->attr.config;
+
+ WARN_ON(event->parent);
+
+ atomic_dec(&perf_swevent_enabled[event_id]);
+ swevent_hlist_put(event);
+}
+
+static int perf_swevent_init(struct perf_event *event)
+{
+ int 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;
+
+ atomic_inc(&perf_swevent_enabled[event_id]);
+ event->destroy = sw_perf_event_destroy;
+ }
+
+ return 0;
+}
+
+static struct pmu perf_swevent = {
+ .event_init = perf_swevent_init,
+ .enable = perf_swevent_enable,
+ .disable = perf_swevent_disable,
.start = perf_swevent_int,
.stop = perf_swevent_void,
.read = perf_swevent_read,
- .unthrottle = perf_swevent_void,
+ .unthrottle = perf_swevent_void, /* hwc->interrupts already reset */
};
+#ifdef CONFIG_EVENT_TRACING
+
static int perf_tp_filter_match(struct perf_event *event,
struct perf_sample_data *data)
{
}
void perf_tp_event(u64 addr, u64 count, void *record, int entry_size,
- struct pt_regs *regs, struct hlist_head *head)
+ struct pt_regs *regs, struct hlist_head *head, int rctx)
{
struct perf_sample_data data;
struct perf_event *event;
perf_sample_data_init(&data, addr);
data.raw = &raw;
- rcu_read_lock();
hlist_for_each_entry_rcu(event, node, head, hlist_entry) {
if (perf_tp_event_match(event, &data, regs))
perf_swevent_add(event, count, 1, &data, regs);
}
- rcu_read_unlock();
+
+ perf_swevent_put_recursion_context(rctx);
}
EXPORT_SYMBOL_GPL(perf_tp_event);
perf_trace_destroy(event);
}
-static const struct pmu *tp_perf_event_init(struct perf_event *event)
+static int perf_tp_event_init(struct perf_event *event)
{
int err;
+ if (event->attr.type != PERF_TYPE_TRACEPOINT)
+ return -ENOENT;
+
/*
* Raw tracepoint data is a severe data leak, only allow root to
* have these.
if ((event->attr.sample_type & PERF_SAMPLE_RAW) &&
perf_paranoid_tracepoint_raw() &&
!capable(CAP_SYS_ADMIN))
- return ERR_PTR(-EPERM);
+ return -EPERM;
err = perf_trace_init(event);
if (err)
- return NULL;
+ return err;
event->destroy = tp_perf_event_destroy;
- return &perf_ops_tracepoint;
+ return 0;
+}
+
+static struct pmu perf_tracepoint = {
+ .event_init = perf_tp_event_init,
+ .enable = perf_trace_enable,
+ .disable = perf_trace_disable,
+ .start = perf_swevent_int,
+ .stop = perf_swevent_void,
+ .read = perf_swevent_read,
+ .unthrottle = perf_swevent_void,
+};
+
+static inline void perf_tp_register(void)
+{
+ perf_pmu_register(&perf_tracepoint);
}
static int perf_event_set_filter(struct perf_event *event, void __user *arg)
#else
-static const struct pmu *tp_perf_event_init(struct perf_event *event)
+static inline void perf_tp_register(void)
{
- return NULL;
}
static int perf_event_set_filter(struct perf_event *event, void __user *arg)
#endif /* CONFIG_EVENT_TRACING */
#ifdef CONFIG_HAVE_HW_BREAKPOINT
-static void bp_perf_event_destroy(struct perf_event *event)
+void perf_bp_event(struct perf_event *bp, void *data)
{
- release_bp_slot(event);
+ struct perf_sample_data sample;
+ struct pt_regs *regs = data;
+
+ perf_sample_data_init(&sample, bp->attr.bp_addr);
+
+ if (!perf_exclude_event(bp, regs))
+ perf_swevent_add(bp, 1, 1, &sample, regs);
}
+#endif
+
+/*
+ * hrtimer based swevent callback
+ */
-static const struct pmu *bp_perf_event_init(struct perf_event *bp)
+static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer)
{
- int err;
+ enum hrtimer_restart ret = HRTIMER_RESTART;
+ struct perf_sample_data data;
+ struct pt_regs *regs;
+ struct perf_event *event;
+ u64 period;
- err = register_perf_hw_breakpoint(bp);
- if (err)
- return ERR_PTR(err);
+ event = container_of(hrtimer, struct perf_event, hw.hrtimer);
+ event->pmu->read(event);
+
+ perf_sample_data_init(&data, 0);
+ data.period = event->hw.last_period;
+ regs = get_irq_regs();
- bp->destroy = bp_perf_event_destroy;
+ 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;
+ }
+
+ period = max_t(u64, 10000, event->hw.sample_period);
+ hrtimer_forward_now(hrtimer, ns_to_ktime(period));
- return &perf_ops_bp;
+ return ret;
}
-void perf_bp_event(struct perf_event *bp, void *data)
+static void perf_swevent_start_hrtimer(struct perf_event *event)
{
- struct perf_sample_data sample;
- struct pt_regs *regs = data;
+ struct hw_perf_event *hwc = &event->hw;
- perf_sample_data_init(&sample, bp->attr.bp_addr);
+ hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+ hwc->hrtimer.function = perf_swevent_hrtimer;
+ if (hwc->sample_period) {
+ u64 period;
- if (!perf_exclude_event(bp, regs))
- perf_swevent_add(bp, 1, 1, &sample, regs);
+ if (hwc->remaining) {
+ if (hwc->remaining < 0)
+ period = 10000;
+ else
+ period = hwc->remaining;
+ hwc->remaining = 0;
+ } else {
+ period = max_t(u64, 10000, hwc->sample_period);
+ }
+ __hrtimer_start_range_ns(&hwc->hrtimer,
+ ns_to_ktime(period), 0,
+ HRTIMER_MODE_REL, 0);
+ }
}
-#else
-static const struct pmu *bp_perf_event_init(struct perf_event *bp)
+
+static void perf_swevent_cancel_hrtimer(struct perf_event *event)
{
- return NULL;
+ struct hw_perf_event *hwc = &event->hw;
+
+ if (hwc->sample_period) {
+ ktime_t remaining = hrtimer_get_remaining(&hwc->hrtimer);
+ hwc->remaining = ktime_to_ns(remaining);
+
+ hrtimer_cancel(&hwc->hrtimer);
+ }
}
-void perf_bp_event(struct perf_event *bp, void *regs)
+/*
+ * Software event: cpu wall time clock
+ */
+
+static void cpu_clock_event_update(struct perf_event *event)
{
+ int cpu = raw_smp_processor_id();
+ s64 prev;
+ u64 now;
+
+ now = cpu_clock(cpu);
+ prev = local64_xchg(&event->hw.prev_count, now);
+ local64_add(now - prev, &event->count);
}
-#endif
-atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX];
+static int cpu_clock_event_enable(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ int cpu = raw_smp_processor_id();
-static void sw_perf_event_destroy(struct perf_event *event)
+ local64_set(&hwc->prev_count, cpu_clock(cpu));
+ perf_swevent_start_hrtimer(event);
+
+ return 0;
+}
+
+static void cpu_clock_event_disable(struct perf_event *event)
{
- u64 event_id = event->attr.config;
+ perf_swevent_cancel_hrtimer(event);
+ cpu_clock_event_update(event);
+}
- WARN_ON(event->parent);
+static void cpu_clock_event_read(struct perf_event *event)
+{
+ cpu_clock_event_update(event);
+}
- atomic_dec(&perf_swevent_enabled[event_id]);
- swevent_hlist_put(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;
+
+ return 0;
}
-static const struct pmu *sw_perf_event_init(struct perf_event *event)
+static struct pmu perf_cpu_clock = {
+ .event_init = cpu_clock_event_init,
+ .enable = cpu_clock_event_enable,
+ .disable = cpu_clock_event_disable,
+ .read = cpu_clock_event_read,
+};
+
+/*
+ * Software event: task time clock
+ */
+
+static void task_clock_event_update(struct perf_event *event, u64 now)
{
- const struct pmu *pmu = NULL;
- u64 event_id = event->attr.config;
+ u64 prev;
+ s64 delta;
- /*
- * Software events (currently) can't in general distinguish
- * between user, kernel and hypervisor events.
- * However, context switches and cpu migrations are considered
- * to be kernel events, and page faults are never hypervisor
- * events.
- */
- switch (event_id) {
- case PERF_COUNT_SW_CPU_CLOCK:
- pmu = &perf_ops_cpu_clock;
+ prev = local64_xchg(&event->hw.prev_count, now);
+ delta = now - prev;
+ local64_add(delta, &event->count);
+}
- break;
- case PERF_COUNT_SW_TASK_CLOCK:
- /*
- * If the user instantiates this as a per-cpu event,
- * use the cpu_clock event instead.
- */
- if (event->ctx->task)
- pmu = &perf_ops_task_clock;
- else
- pmu = &perf_ops_cpu_clock;
+static int task_clock_event_enable(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ u64 now;
- break;
- case PERF_COUNT_SW_PAGE_FAULTS:
- case PERF_COUNT_SW_PAGE_FAULTS_MIN:
- case PERF_COUNT_SW_PAGE_FAULTS_MAJ:
- case PERF_COUNT_SW_CONTEXT_SWITCHES:
- case PERF_COUNT_SW_CPU_MIGRATIONS:
- 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);
+ now = event->ctx->time;
+
+ local64_set(&hwc->prev_count, now);
+
+ perf_swevent_start_hrtimer(event);
- atomic_inc(&perf_swevent_enabled[event_id]);
- event->destroy = sw_perf_event_destroy;
+ return 0;
+}
+
+static void task_clock_event_disable(struct perf_event *event)
+{
+ perf_swevent_cancel_hrtimer(event);
+ task_clock_event_update(event, event->ctx->time);
+
+}
+
+static void task_clock_event_read(struct perf_event *event)
+{
+ u64 time;
+
+ if (!in_nmi()) {
+ update_context_time(event->ctx);
+ time = event->ctx->time;
+ } else {
+ u64 now = perf_clock();
+ u64 delta = now - event->ctx->timestamp;
+ 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;
+
+ return 0;
+}
+
+static struct pmu perf_task_clock = {
+ .event_init = task_clock_event_init,
+ .enable = task_clock_event_enable,
+ .disable = task_clock_event_disable,
+ .read = task_clock_event_read,
+};
+
+static LIST_HEAD(pmus);
+static DEFINE_MUTEX(pmus_lock);
+static struct srcu_struct pmus_srcu;
+
+int perf_pmu_register(struct pmu *pmu)
+{
+ mutex_lock(&pmus_lock);
+ list_add_rcu(&pmu->entry, &pmus);
+ mutex_unlock(&pmus_lock);
+
+ return 0;
+}
+
+void perf_pmu_unregister(struct pmu *pmu)
+{
+ mutex_lock(&pmus_lock);
+ list_del_rcu(&pmu->entry);
+ mutex_unlock(&pmus_lock);
+
+ synchronize_srcu(&pmus_srcu);
+}
+
+struct pmu *perf_init_event(struct perf_event *event)
+{
+ struct pmu *pmu = NULL;
+ int idx;
+
+ idx = srcu_read_lock(&pmus_srcu);
+ list_for_each_entry_rcu(pmu, &pmus, entry) {
+ int ret = pmu->event_init(event);
+ if (!ret)
+ break;
+ if (ret != -ENOENT) {
+ pmu = ERR_PTR(ret);
+ break;
}
- pmu = &perf_ops_generic;
- break;
}
+ srcu_read_unlock(&pmus_srcu, idx);
return pmu;
}
perf_overflow_handler_t overflow_handler,
gfp_t gfpflags)
{
- const struct pmu *pmu;
+ struct pmu *pmu;
struct perf_event *event;
struct hw_perf_event *hwc;
long err;
hwc->sample_period = 1;
hwc->last_period = hwc->sample_period;
- atomic64_set(&hwc->period_left, hwc->sample_period);
+ local64_set(&hwc->period_left, hwc->sample_period);
/*
* we currently do not support PERF_FORMAT_GROUP on inherited events
if (attr->inherit && (attr->read_format & PERF_FORMAT_GROUP))
goto done;
- switch (attr->type) {
- case PERF_TYPE_RAW:
- case PERF_TYPE_HARDWARE:
- case PERF_TYPE_HW_CACHE:
- pmu = hw_perf_event_init(event);
- break;
-
- case PERF_TYPE_SOFTWARE:
- pmu = sw_perf_event_init(event);
- break;
-
- case PERF_TYPE_TRACEPOINT:
- pmu = tp_perf_event_init(event);
- break;
-
- case PERF_TYPE_BREAKPOINT:
- pmu = bp_perf_event_init(event);
- break;
-
+ pmu = perf_init_event(event);
- default:
- break;
- }
done:
err = 0;
if (!pmu)
if (!event->parent) {
atomic_inc(&nr_events);
- if (event->attr.mmap)
+ 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_event_set_output(struct perf_event *event, struct perf_event *output_event)
{
- struct perf_mmap_data *data = NULL, *old_data = NULL;
+ struct perf_buffer *buffer = NULL, *old_buffer = NULL;
int ret = -EINVAL;
if (!output_event)
if (output_event) {
/* get the buffer we want to redirect to */
- data = perf_mmap_data_get(output_event);
- if (!data)
+ buffer = perf_buffer_get(output_event);
+ if (!buffer)
goto unlock;
}
- old_data = event->data;
- rcu_assign_pointer(event->data, data);
+ old_buffer = event->buffer;
+ rcu_assign_pointer(event->buffer, buffer);
ret = 0;
unlock:
mutex_unlock(&event->mmap_mutex);
- if (old_data)
- perf_mmap_data_put(old_data);
+ if (old_buffer)
+ perf_buffer_put(old_buffer);
out:
return ret;
}
hwc->sample_period = sample_period;
hwc->last_period = sample_period;
- atomic64_set(&hwc->period_left, sample_period);
+ local64_set(&hwc->period_left, sample_period);
}
child_event->overflow_handler = parent_event->overflow_handler;
if (child_event->attr.inherit_stat)
perf_event_read_event(child_event, child);
- child_val = atomic64_read(&child_event->count);
+ child_val = perf_event_count(child_event);
/*
* Add back the child's count to the parent's count:
*/
- atomic64_add(child_val, &parent_event->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,
{
unsigned int cpu = (long)hcpu;
- switch (action) {
+ switch (action & ~CPU_TASKS_FROZEN) {
case CPU_UP_PREPARE:
- case CPU_UP_PREPARE_FROZEN:
+ case CPU_DOWN_FAILED:
perf_event_init_cpu(cpu);
break;
+ case CPU_UP_CANCELED:
case CPU_DOWN_PREPARE:
- case CPU_DOWN_PREPARE_FROZEN:
perf_event_exit_cpu(cpu);
break;
return NOTIFY_OK;
}
-/*
- * This has to have a higher priority than migration_notifier in sched.c.
- */
-static struct notifier_block __cpuinitdata perf_cpu_nb = {
- .notifier_call = perf_cpu_notify,
- .priority = 20,
-};
-
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,
- (void *)(long)smp_processor_id());
- register_cpu_notifier(&perf_cpu_nb);
+ init_srcu_struct(&pmus_srcu);
+ perf_pmu_register(&perf_swevent);
+ perf_pmu_register(&perf_cpu_clock);
+ perf_pmu_register(&perf_task_clock);
+ perf_tp_register();
+ perf_cpu_notifier(perf_cpu_notify);
}
static ssize_t perf_show_reserve_percpu(struct sysdev_class *class,
projects / mv-sheeva.git / blobdiff