#include <linux/percpu.h>
#include <linux/ptrace.h>
#include <linux/vmstat.h>
+#include <linux/vmalloc.h>
#include <linux/hardirq.h>
#include <linux/rculist.h>
#include <linux/uaccess.h>
#include <linux/anon_inodes.h>
#include <linux/kernel_stat.h>
#include <linux/perf_event.h>
+#include <linux/ftrace_event.h>
+#include <linux/hw_breakpoint.h>
#include <asm/irq_regs.h>
put_ctx(ctx);
}
+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_event_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 event.
+ */
+static void update_event_times(struct perf_event *event)
+{
+ struct perf_event_context *ctx = event->ctx;
+ u64 run_end;
+
+ if (event->state < PERF_EVENT_STATE_INACTIVE ||
+ event->group_leader->state < PERF_EVENT_STATE_INACTIVE)
+ return;
+
+ if (ctx->is_active)
+ run_end = ctx->time;
+ else
+ run_end = event->tstamp_stopped;
+
+ event->total_time_enabled = run_end - event->tstamp_enabled;
+
+ if (event->state == PERF_EVENT_STATE_INACTIVE)
+ run_end = event->tstamp_stopped;
+ else
+ run_end = ctx->time;
+
+ event->total_time_running = run_end - event->tstamp_running;
+}
+
/*
* Add a event from the lists for its context.
* Must be called with ctx->mutex and ctx->lock held.
if (event->group_leader != event)
event->group_leader->nr_siblings--;
+ update_event_times(event);
+
+ /*
+ * If event was in error state, then keep it
+ * that way, otherwise bogus counts will be
+ * returned on read(). The only way to get out
+ * of error state is by explicit re-enabling
+ * of the event
+ */
+ if (event->state > PERF_EVENT_STATE_OFF)
+ event->state = PERF_EVENT_STATE_OFF;
+
/*
* If this was a group event with sibling events then
* upgrade the siblings to singleton events by adding them
* can remove the event safely, if the call above did not
* succeed.
*/
- if (!list_empty(&event->group_entry)) {
+ if (!list_empty(&event->group_entry))
list_del_event(event, 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_event_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 event.
- */
-static void update_event_times(struct perf_event *event)
-{
- struct perf_event_context *ctx = event->ctx;
- u64 run_end;
-
- if (event->state < PERF_EVENT_STATE_INACTIVE ||
- event->group_leader->state < PERF_EVENT_STATE_INACTIVE)
- return;
-
- event->total_time_enabled = ctx->time - event->tstamp_enabled;
-
- if (event->state == PERF_EVENT_STATE_INACTIVE)
- run_end = event->tstamp_stopped;
- else
- run_end = ctx->time;
-
- event->total_time_running = run_end - event->tstamp_running;
-}
-
/*
* Update total_time_enabled and total_time_running for all events in a group.
*/
update_context_time(ctx);
perf_disable();
- if (ctx->nr_active)
+ if (ctx->nr_active) {
list_for_each_entry(event, &ctx->group_list, group_entry)
group_sched_out(event, cpuctx, ctx);
-
+ }
perf_enable();
out:
spin_unlock(&ctx->lock);
&& !ctx1->pin_count && !ctx2->pin_count;
}
-static void __perf_event_read(void *event);
-
static void __perf_event_sync_stat(struct perf_event *event,
struct perf_event *next_event)
{
*/
switch (event->state) {
case PERF_EVENT_STATE_ACTIVE:
- __perf_event_read(event);
- break;
+ event->pmu->read(event);
+ /* fall-through */
case PERF_EVENT_STATE_INACTIVE:
update_event_times(event);
if (!ctx->nr_stat)
return;
+ update_context_time(ctx);
+
event = list_first_entry(&ctx->event_list,
struct perf_event, event_entry);
if (likely(!ctx || !cpuctx->task_ctx))
return;
- update_context_time(ctx);
-
rcu_read_lock();
parent = rcu_dereference(ctx->parent_ctx);
next_ctx = next->perf_event_ctxp;
u64 interrupts, freq;
spin_lock(&ctx->lock);
- list_for_each_entry(event, &ctx->group_list, group_entry) {
+ list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
if (event->state != PERF_EVENT_STATE_ACTIVE)
continue;
struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
struct perf_event *event = info;
struct perf_event_context *ctx = event->ctx;
- unsigned long flags;
/*
* If this is a task context, we need to check whether it is
if (ctx->task && cpuctx->task_ctx != ctx)
return;
- local_irq_save(flags);
- if (ctx->is_active)
- update_context_time(ctx);
- event->pmu->read(event);
+ spin_lock(&ctx->lock);
+ update_context_time(ctx);
update_event_times(event);
- local_irq_restore(flags);
+ spin_unlock(&ctx->lock);
+
+ event->pmu->read(event);
}
static u64 perf_event_read(struct perf_event *event)
smp_call_function_single(event->oncpu,
__perf_event_read, event, 1);
} else if (event->state == PERF_EVENT_STATE_INACTIVE) {
+ struct perf_event_context *ctx = event->ctx;
+ unsigned long flags;
+
+ spin_lock_irqsave(&ctx->lock, flags);
+ update_context_time(ctx);
update_event_times(event);
+ spin_unlock_irqrestore(&ctx->lock, flags);
}
return atomic64_read(&event->count);
return ERR_PTR(err);
}
+static void perf_event_free_filter(struct perf_event *event);
+
static void free_event_rcu(struct rcu_head *head)
{
struct perf_event *event;
event = container_of(head, struct perf_event, rcu_head);
if (event->ns)
put_pid_ns(event->ns);
+ perf_event_free_filter(event);
kfree(event);
}
call_rcu(&event->rcu_head, free_event_rcu);
}
-/*
- * Called when the last reference to the file is gone.
- */
-static int perf_release(struct inode *inode, struct file *file)
+int perf_event_release_kernel(struct perf_event *event)
{
- struct perf_event *event = file->private_data;
struct perf_event_context *ctx = event->ctx;
- file->private_data = NULL;
-
WARN_ON_ONCE(ctx->parent_ctx);
mutex_lock(&ctx->mutex);
perf_event_remove_from_context(event);
return 0;
}
+EXPORT_SYMBOL_GPL(perf_event_release_kernel);
+
+/*
+ * Called when the last reference to the file is gone.
+ */
+static int perf_release(struct inode *inode, struct file *file)
+{
+ struct perf_event *event = file->private_data;
+
+ file->private_data = NULL;
+
+ return perf_event_release_kernel(event);
+}
static int perf_event_read_size(struct perf_event *event)
{
return size;
}
-static u64 perf_event_read_value(struct perf_event *event)
+u64 perf_event_read_value(struct perf_event *event, u64 *enabled, u64 *running)
{
struct perf_event *child;
u64 total = 0;
+ *enabled = 0;
+ *running = 0;
+
+ mutex_lock(&event->child_mutex);
total += perf_event_read(event);
- list_for_each_entry(child, &event->child_list, child_list)
+ *enabled += event->total_time_enabled +
+ atomic64_read(&event->child_total_time_enabled);
+ *running += event->total_time_running +
+ atomic64_read(&event->child_total_time_running);
+
+ list_for_each_entry(child, &event->child_list, child_list) {
total += perf_event_read(child);
+ *enabled += child->total_time_enabled;
+ *running += child->total_time_running;
+ }
+ mutex_unlock(&event->child_mutex);
return total;
}
-
-static int perf_event_read_entry(struct perf_event *event,
- u64 read_format, char __user *buf)
-{
- int n = 0, count = 0;
- u64 values[2];
-
- values[n++] = perf_event_read_value(event);
- if (read_format & PERF_FORMAT_ID)
- values[n++] = primary_event_id(event);
-
- count = n * sizeof(u64);
-
- if (copy_to_user(buf, values, count))
- return -EFAULT;
-
- return count;
-}
+EXPORT_SYMBOL_GPL(perf_event_read_value);
static int perf_event_read_group(struct perf_event *event,
u64 read_format, char __user *buf)
{
struct perf_event *leader = event->group_leader, *sub;
- int n = 0, size = 0, err = -EFAULT;
- u64 values[3];
+ int n = 0, size = 0, ret = -EFAULT;
+ struct perf_event_context *ctx = leader->ctx;
+ u64 values[5];
+ u64 count, enabled, running;
+
+ mutex_lock(&ctx->mutex);
+ count = perf_event_read_value(leader, &enabled, &running);
values[n++] = 1 + leader->nr_siblings;
- if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) {
- values[n++] = leader->total_time_enabled +
- atomic64_read(&leader->child_total_time_enabled);
- }
- if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) {
- values[n++] = leader->total_time_running +
- atomic64_read(&leader->child_total_time_running);
- }
+ if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
+ values[n++] = enabled;
+ if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
+ values[n++] = running;
+ values[n++] = count;
+ if (read_format & PERF_FORMAT_ID)
+ values[n++] = primary_event_id(leader);
size = n * sizeof(u64);
if (copy_to_user(buf, values, size))
- return -EFAULT;
-
- err = perf_event_read_entry(leader, read_format, buf + size);
- if (err < 0)
- return err;
+ goto unlock;
- size += err;
+ ret = size;
list_for_each_entry(sub, &leader->sibling_list, group_entry) {
- err = perf_event_read_entry(sub, read_format,
- buf + size);
- if (err < 0)
- return err;
+ n = 0;
+
+ values[n++] = perf_event_read_value(sub, &enabled, &running);
+ if (read_format & PERF_FORMAT_ID)
+ values[n++] = primary_event_id(sub);
+
+ size = n * sizeof(u64);
- size += err;
+ if (copy_to_user(buf + ret, values, size)) {
+ ret = -EFAULT;
+ goto unlock;
+ }
+
+ ret += size;
}
+unlock:
+ mutex_unlock(&ctx->mutex);
- return size;
+ return ret;
}
static int perf_event_read_one(struct perf_event *event,
u64 read_format, char __user *buf)
{
+ u64 enabled, running;
u64 values[4];
int n = 0;
- values[n++] = perf_event_read_value(event);
- if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) {
- values[n++] = event->total_time_enabled +
- atomic64_read(&event->child_total_time_enabled);
- }
- if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) {
- values[n++] = event->total_time_running +
- atomic64_read(&event->child_total_time_running);
- }
+ values[n++] = perf_event_read_value(event, &enabled, &running);
+ if (read_format & PERF_FORMAT_TOTAL_TIME_ENABLED)
+ values[n++] = enabled;
+ if (read_format & PERF_FORMAT_TOTAL_TIME_RUNNING)
+ values[n++] = running;
if (read_format & PERF_FORMAT_ID)
values[n++] = primary_event_id(event);
return -ENOSPC;
WARN_ON_ONCE(event->ctx->parent_ctx);
- mutex_lock(&event->child_mutex);
if (read_format & PERF_FORMAT_GROUP)
ret = perf_event_read_group(event, read_format, buf);
else
ret = perf_event_read_one(event, read_format, buf);
- mutex_unlock(&event->child_mutex);
return ret;
}
return ret;
}
-int perf_event_set_output(struct perf_event *event, int output_fd);
+static int perf_event_set_output(struct perf_event *event, int output_fd);
+static int perf_event_set_filter(struct perf_event *event, void __user *arg);
static long perf_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
case PERF_EVENT_IOC_SET_OUTPUT:
return perf_event_set_output(event, arg);
+ case PERF_EVENT_IOC_SET_FILTER:
+ return perf_event_set_filter(event, (void __user *)arg);
+
default:
return -ENOTTY;
}
rcu_read_unlock();
}
-static int perf_mmap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
+static unsigned long perf_data_size(struct perf_mmap_data *data)
{
- struct perf_event *event = vma->vm_file->private_data;
- struct perf_mmap_data *data;
- int ret = VM_FAULT_SIGBUS;
-
- if (vmf->flags & FAULT_FLAG_MKWRITE) {
- if (vmf->pgoff == 0)
- ret = 0;
- return ret;
- }
-
- rcu_read_lock();
- data = rcu_dereference(event->data);
- if (!data)
- goto unlock;
-
- if (vmf->pgoff == 0) {
- vmf->page = virt_to_page(data->user_page);
- } else {
- int nr = vmf->pgoff - 1;
-
- if ((unsigned)nr > data->nr_pages)
- goto unlock;
+ return data->nr_pages << (PAGE_SHIFT + data->data_order);
+}
- if (vmf->flags & FAULT_FLAG_WRITE)
- goto unlock;
+#ifndef CONFIG_PERF_USE_VMALLOC
- vmf->page = virt_to_page(data->data_pages[nr]);
- }
+/*
+ * Back perf_mmap() with regular GFP_KERNEL-0 pages.
+ */
- get_page(vmf->page);
- vmf->page->mapping = vma->vm_file->f_mapping;
- vmf->page->index = vmf->pgoff;
+static struct page *
+perf_mmap_to_page(struct perf_mmap_data *data, unsigned long pgoff)
+{
+ if (pgoff > data->nr_pages)
+ return NULL;
- ret = 0;
-unlock:
- rcu_read_unlock();
+ if (pgoff == 0)
+ return virt_to_page(data->user_page);
- return ret;
+ return virt_to_page(data->data_pages[pgoff - 1]);
}
-static int perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
+static struct perf_mmap_data *
+perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
{
struct perf_mmap_data *data;
unsigned long size;
goto fail_data_pages;
}
+ data->data_order = 0;
data->nr_pages = nr_pages;
- atomic_set(&data->lock, -1);
-
- if (event->attr.watermark) {
- data->watermark = min_t(long, PAGE_SIZE * nr_pages,
- event->attr.wakeup_watermark);
- }
- if (!data->watermark)
- data->watermark = max(PAGE_SIZE, PAGE_SIZE * nr_pages / 4);
-
- rcu_assign_pointer(event->data, data);
- return 0;
+ return data;
fail_data_pages:
for (i--; i >= 0; i--)
free_page((unsigned long)data->user_page);
-fail_user_page:
- kfree(data);
+fail_user_page:
+ kfree(data);
+
+fail:
+ return NULL;
+}
+
+static void perf_mmap_free_page(unsigned long addr)
+{
+ struct page *page = virt_to_page((void *)addr);
+
+ page->mapping = NULL;
+ __free_page(page);
+}
+
+static void perf_mmap_data_free(struct perf_mmap_data *data)
+{
+ 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);
+}
+
+#else
+
+/*
+ * Back perf_mmap() with vmalloc memory.
+ *
+ * Required for architectures that have d-cache aliasing issues.
+ */
+
+static struct page *
+perf_mmap_to_page(struct perf_mmap_data *data, unsigned long pgoff)
+{
+ if (pgoff > (1UL << data->data_order))
+ return NULL;
+
+ return vmalloc_to_page((void *)data->user_page + pgoff * PAGE_SIZE);
+}
+
+static void perf_mmap_unmark_page(void *addr)
+{
+ struct page *page = vmalloc_to_page(addr);
+
+ page->mapping = NULL;
+}
+
+static void perf_mmap_data_free_work(struct work_struct *work)
+{
+ struct perf_mmap_data *data;
+ void *base;
+ int i, nr;
+
+ data = container_of(work, struct perf_mmap_data, work);
+ nr = 1 << data->data_order;
+
+ base = data->user_page;
+ for (i = 0; i < nr + 1; i++)
+ perf_mmap_unmark_page(base + (i * PAGE_SIZE));
+
+ vfree(base);
+ kfree(data);
+}
+
+static void perf_mmap_data_free(struct perf_mmap_data *data)
+{
+ schedule_work(&data->work);
+}
+
+static struct perf_mmap_data *
+perf_mmap_data_alloc(struct perf_event *event, int nr_pages)
+{
+ struct perf_mmap_data *data;
+ unsigned long size;
+ void *all_buf;
+
+ WARN_ON(atomic_read(&event->mmap_count));
+
+ size = sizeof(struct perf_mmap_data);
+ size += sizeof(void *);
+
+ data = kzalloc(size, GFP_KERNEL);
+ if (!data)
+ goto fail;
+
+ INIT_WORK(&data->work, perf_mmap_data_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->data_order = ilog2(nr_pages);
+ data->nr_pages = 1;
+
+ return data;
+
+fail_all_buf:
+ kfree(data);
+
+fail:
+ return NULL;
+}
+
+#endif
+
+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;
+ int ret = VM_FAULT_SIGBUS;
+
+ if (vmf->flags & FAULT_FLAG_MKWRITE) {
+ if (vmf->pgoff == 0)
+ ret = 0;
+ return ret;
+ }
+
+ rcu_read_lock();
+ data = rcu_dereference(event->data);
+ if (!data)
+ goto unlock;
+
+ if (vmf->pgoff && (vmf->flags & FAULT_FLAG_WRITE))
+ goto unlock;
+
+ vmf->page = perf_mmap_to_page(data, vmf->pgoff);
+ if (!vmf->page)
+ goto unlock;
+
+ get_page(vmf->page);
+ vmf->page->mapping = vma->vm_file->f_mapping;
+ vmf->page->index = vmf->pgoff;
+
+ ret = 0;
+unlock:
+ rcu_read_unlock();
+
+ return ret;
+}
+
+static void
+perf_mmap_data_init(struct perf_event *event, struct perf_mmap_data *data)
+{
+ long max_size = perf_data_size(data);
+
+ atomic_set(&data->lock, -1);
+
+ if (event->attr.watermark) {
+ data->watermark = min_t(long, max_size,
+ event->attr.wakeup_watermark);
+ }
-fail:
- return -ENOMEM;
-}
+ if (!data->watermark)
+ data->watermark = max_size / 2;
-static void perf_mmap_free_page(unsigned long addr)
-{
- struct page *page = virt_to_page((void *)addr);
- page->mapping = NULL;
- __free_page(page);
+ rcu_assign_pointer(event->data, data);
}
-static void __perf_mmap_data_free(struct rcu_head *rcu_head)
+static void perf_mmap_data_free_rcu(struct rcu_head *rcu_head)
{
struct perf_mmap_data *data;
- int i;
data = container_of(rcu_head, struct perf_mmap_data, rcu_head);
-
- 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_data_free(data);
}
-static void perf_mmap_data_free(struct perf_event *event)
+static void perf_mmap_data_release(struct perf_event *event)
{
struct perf_mmap_data *data = event->data;
WARN_ON(atomic_read(&event->mmap_count));
rcu_assign_pointer(event->data, NULL);
- call_rcu(&data->rcu_head, __perf_mmap_data_free);
+ call_rcu(&data->rcu_head, perf_mmap_data_free_rcu);
}
static void perf_mmap_open(struct vm_area_struct *vma)
WARN_ON_ONCE(event->ctx->parent_ctx);
if (atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex)) {
+ unsigned long size = perf_data_size(event->data);
struct user_struct *user = current_user();
- atomic_long_sub(event->data->nr_pages + 1, &user->locked_vm);
+ atomic_long_sub((size >> PAGE_SHIFT) + 1, &user->locked_vm);
vma->vm_mm->locked_vm -= event->data->nr_locked;
- perf_mmap_data_free(event);
+ perf_mmap_data_release(event);
mutex_unlock(&event->mmap_mutex);
}
}
unsigned long user_locked, user_lock_limit;
struct user_struct *user = current_user();
unsigned long locked, lock_limit;
+ struct perf_mmap_data *data;
unsigned long vma_size;
unsigned long nr_pages;
long user_extra, extra;
}
WARN_ON(event->data);
- ret = perf_mmap_data_alloc(event, nr_pages);
- if (ret)
+
+ data = perf_mmap_data_alloc(event, nr_pages);
+ ret = -ENOMEM;
+ if (!data)
goto unlock;
+ ret = 0;
+ perf_mmap_data_init(event, data);
+
atomic_set(&event->mmap_count, 1);
atomic_long_add(user_extra, &user->locked_vm);
vma->vm_mm->locked_vm += extra;
if (!data->writable)
return true;
- mask = (data->nr_pages << PAGE_SHIFT) - 1;
+ mask = perf_data_size(data) - 1;
offset = (offset - tail) & mask;
head = (head - tail) & mask;
static void perf_output_lock(struct perf_output_handle *handle)
{
struct perf_mmap_data *data = handle->data;
- int cpu;
+ int cur, cpu = get_cpu();
handle->locked = 0;
- local_irq_save(handle->flags);
- cpu = smp_processor_id();
-
- if (in_nmi() && atomic_read(&data->lock) == cpu)
- return;
+ for (;;) {
+ cur = atomic_cmpxchg(&data->lock, -1, cpu);
+ if (cur == -1) {
+ handle->locked = 1;
+ break;
+ }
+ if (cur == cpu)
+ break;
- while (atomic_cmpxchg(&data->lock, -1, cpu) != -1)
cpu_relax();
-
- handle->locked = 1;
+ }
}
static void perf_output_unlock(struct perf_output_handle *handle)
if (atomic_xchg(&data->wakeup, 0))
perf_output_wakeup(handle);
out:
- local_irq_restore(handle->flags);
+ put_cpu();
}
void perf_output_copy(struct perf_output_handle *handle,
const void *buf, unsigned int len)
{
unsigned int pages_mask;
- unsigned int offset;
+ unsigned long offset;
unsigned int size;
void **pages;
pages = handle->data->data_pages;
do {
- unsigned int page_offset;
+ unsigned long page_offset;
+ unsigned long page_size;
int nr;
nr = (offset >> PAGE_SHIFT) & pages_mask;
- page_offset = offset & (PAGE_SIZE - 1);
- size = min_t(unsigned int, PAGE_SIZE - page_offset, len);
+ page_size = 1UL << (handle->data->data_order + PAGE_SHIFT);
+ page_offset = offset & (page_size - 1);
+ size = min_t(unsigned int, page_size - page_offset, len);
memcpy(pages[nr] + page_offset, buf, size);
{
struct perf_event *event;
- if (system_state != SYSTEM_RUNNING || list_empty(&ctx->event_list))
- return;
-
- rcu_read_lock();
list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
if (perf_event_task_match(event))
perf_event_task_output(event, task_event);
}
- rcu_read_unlock();
}
static void perf_event_task_event(struct perf_task_event *task_event)
struct perf_cpu_context *cpuctx;
struct perf_event_context *ctx = task_event->task_ctx;
+ rcu_read_lock();
cpuctx = &get_cpu_var(perf_cpu_context);
perf_event_task_ctx(&cpuctx->ctx, task_event);
put_cpu_var(perf_cpu_context);
- rcu_read_lock();
if (!ctx)
ctx = rcu_dereference(task_event->task->perf_event_ctxp);
if (ctx)
{
struct perf_event *event;
- if (system_state != SYSTEM_RUNNING || list_empty(&ctx->event_list))
- return;
-
- rcu_read_lock();
list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
if (perf_event_comm_match(event))
perf_event_comm_output(event, comm_event);
}
- rcu_read_unlock();
}
static void perf_event_comm_event(struct perf_comm_event *comm_event)
char comm[TASK_COMM_LEN];
memset(comm, 0, sizeof(comm));
- strncpy(comm, comm_event->task->comm, sizeof(comm));
+ strlcpy(comm, comm_event->task->comm, sizeof(comm));
size = ALIGN(strlen(comm)+1, sizeof(u64));
comm_event->comm = comm;
comm_event->event_id.header.size = sizeof(comm_event->event_id) + size;
+ rcu_read_lock();
cpuctx = &get_cpu_var(perf_cpu_context);
perf_event_comm_ctx(&cpuctx->ctx, comm_event);
put_cpu_var(perf_cpu_context);
- rcu_read_lock();
/*
* doesn't really matter which of the child contexts the
* events ends up in.
{
struct perf_event *event;
- if (system_state != SYSTEM_RUNNING || list_empty(&ctx->event_list))
- return;
-
- rcu_read_lock();
list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
if (perf_event_mmap_match(event, mmap_event))
perf_event_mmap_output(event, mmap_event);
}
- rcu_read_unlock();
}
static void perf_event_mmap_event(struct perf_mmap_event *mmap_event)
mmap_event->event_id.header.size = sizeof(mmap_event->event_id) + size;
+ rcu_read_lock();
cpuctx = &get_cpu_var(perf_cpu_context);
perf_event_mmap_ctx(&cpuctx->ctx, mmap_event);
put_cpu_var(perf_cpu_context);
- rcu_read_lock();
/*
* doesn't really matter which of the child contexts the
* events ends up in.
perf_event_disable(event);
}
- perf_event_output(event, nmi, data, regs);
+ if (event->overflow_handler)
+ event->overflow_handler(event, nmi, data, regs);
+ else
+ perf_event_output(event, nmi, data, regs);
+
return ret;
}
return nr;
}
-static void perf_swevent_overflow(struct perf_event *event,
+static void perf_swevent_overflow(struct perf_event *event, u64 overflow,
int nmi, struct perf_sample_data *data,
struct pt_regs *regs)
{
struct hw_perf_event *hwc = &event->hw;
int throttle = 0;
- u64 overflow;
data->period = event->hw.last_period;
- overflow = perf_swevent_set_period(event);
+ if (!overflow)
+ overflow = perf_swevent_set_period(event);
if (hwc->interrupts == MAX_INTERRUPTS)
return;
atomic64_add(nr, &event->count);
+ if (!regs)
+ return;
+
if (!hwc->sample_period)
return;
- if (!regs)
+ 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))
return;
- if (!atomic64_add_negative(nr, &hwc->period_left))
- perf_swevent_overflow(event, nmi, data, regs);
+ perf_swevent_overflow(event, 0, nmi, data, regs);
}
static int perf_swevent_is_counting(struct perf_event *event)
return 1;
}
+static int perf_tp_event_match(struct perf_event *event,
+ struct perf_sample_data *data);
+
+static int perf_exclude_event(struct perf_event *event,
+ struct pt_regs *regs)
+{
+ if (regs) {
+ if (event->attr.exclude_user && user_mode(regs))
+ return 1;
+
+ if (event->attr.exclude_kernel && !user_mode(regs))
+ return 1;
+ }
+
+ return 0;
+}
+
static int perf_swevent_match(struct perf_event *event,
enum perf_type_id type,
- u32 event_id, struct pt_regs *regs)
+ u32 event_id,
+ struct perf_sample_data *data,
+ struct pt_regs *regs)
{
if (!perf_swevent_is_counting(event))
return 0;
if (event->attr.type != type)
return 0;
+
if (event->attr.config != event_id)
return 0;
- if (regs) {
- if (event->attr.exclude_user && user_mode(regs))
- return 0;
+ if (perf_exclude_event(event, regs))
+ return 0;
- if (event->attr.exclude_kernel && !user_mode(regs))
- return 0;
- }
+ if (event->attr.type == PERF_TYPE_TRACEPOINT &&
+ !perf_tp_event_match(event, data))
+ return 0;
return 1;
}
{
struct perf_event *event;
- if (system_state != SYSTEM_RUNNING || list_empty(&ctx->event_list))
- return;
-
- rcu_read_lock();
list_for_each_entry_rcu(event, &ctx->event_list, event_entry) {
- if (perf_swevent_match(event, type, event_id, regs))
+ if (perf_swevent_match(event, type, event_id, data, regs))
perf_swevent_add(event, nr, nmi, data, regs);
}
- rcu_read_unlock();
}
-static int *perf_swevent_recursion_context(struct perf_cpu_context *cpuctx)
+int perf_swevent_get_recursion_context(void)
{
+ struct perf_cpu_context *cpuctx = &get_cpu_var(perf_cpu_context);
+ int rctx;
+
if (in_nmi())
- return &cpuctx->recursion[3];
+ rctx = 3;
+ else if (in_irq())
+ rctx = 2;
+ else if (in_softirq())
+ rctx = 1;
+ else
+ rctx = 0;
+
+ if (cpuctx->recursion[rctx]) {
+ put_cpu_var(perf_cpu_context);
+ return -1;
+ }
- if (in_irq())
- return &cpuctx->recursion[2];
+ cpuctx->recursion[rctx]++;
+ barrier();
- if (in_softirq())
- return &cpuctx->recursion[1];
+ return rctx;
+}
+EXPORT_SYMBOL_GPL(perf_swevent_get_recursion_context);
- return &cpuctx->recursion[0];
+void perf_swevent_put_recursion_context(int rctx)
+{
+ struct perf_cpu_context *cpuctx = &__get_cpu_var(perf_cpu_context);
+ barrier();
+ cpuctx->recursion[rctx]--;
+ put_cpu_var(perf_cpu_context);
}
+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 = &get_cpu_var(perf_cpu_context);
- int *recursion = perf_swevent_recursion_context(cpuctx);
+ struct perf_cpu_context *cpuctx;
struct perf_event_context *ctx;
- if (*recursion)
- goto out;
-
- (*recursion)++;
- barrier();
-
+ cpuctx = &__get_cpu_var(perf_cpu_context);
+ rcu_read_lock();
perf_swevent_ctx_event(&cpuctx->ctx, type, event_id,
nr, nmi, data, regs);
- rcu_read_lock();
/*
* doesn't really matter which of the child contexts the
* events ends up in.
if (ctx)
perf_swevent_ctx_event(ctx, type, event_id, nr, nmi, data, regs);
rcu_read_unlock();
-
- barrier();
- (*recursion)--;
-
-out:
- put_cpu_var(perf_cpu_context);
}
void __perf_sw_event(u32 event_id, u64 nr, int nmi,
struct pt_regs *regs, u64 addr)
{
- struct perf_sample_data data = {
- .addr = addr,
- };
+ struct perf_sample_data data;
+ int rctx;
- do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, nmi,
- &data, regs);
+ rctx = perf_swevent_get_recursion_context();
+ if (rctx < 0)
+ return;
+
+ data.addr = addr;
+ data.raw = NULL;
+
+ do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, nmi, &data, regs);
+
+ perf_swevent_put_recursion_context(rctx);
}
static void perf_swevent_read(struct perf_event *event)
event->pmu->read(event);
data.addr = 0;
+ data.period = event->hw.last_period;
regs = get_irq_regs();
/*
* In case we exclude kernel IPs or are somehow not in interrupt
regs = task_pt_regs(current);
if (regs) {
- if (perf_event_overflow(event, 0, &data, regs))
- ret = HRTIMER_NORESTART;
+ 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);
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
*/
int cpu = raw_smp_processor_id();
atomic64_set(&hwc->prev_count, cpu_clock(cpu));
- hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
- hwc->hrtimer.function = perf_swevent_hrtimer;
- if (hwc->sample_period) {
- u64 period = max_t(u64, 10000, hwc->sample_period);
- __hrtimer_start_range_ns(&hwc->hrtimer,
- ns_to_ktime(period), 0,
- HRTIMER_MODE_REL, 0);
- }
+ perf_swevent_start_hrtimer(event);
return 0;
}
static void cpu_clock_perf_event_disable(struct perf_event *event)
{
- if (event->hw.sample_period)
- hrtimer_cancel(&event->hw.hrtimer);
+ perf_swevent_cancel_hrtimer(event);
cpu_clock_perf_event_update(event);
}
now = event->ctx->time;
atomic64_set(&hwc->prev_count, now);
- hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
- hwc->hrtimer.function = perf_swevent_hrtimer;
- if (hwc->sample_period) {
- u64 period = max_t(u64, 10000, hwc->sample_period);
- __hrtimer_start_range_ns(&hwc->hrtimer,
- ns_to_ktime(period), 0,
- HRTIMER_MODE_REL, 0);
- }
+
+ perf_swevent_start_hrtimer(event);
return 0;
}
static void task_clock_perf_event_disable(struct perf_event *event)
{
- if (event->hw.sample_period)
- hrtimer_cancel(&event->hw.hrtimer);
+ perf_swevent_cancel_hrtimer(event);
task_clock_perf_event_update(event, event->ctx->time);
}
};
#ifdef CONFIG_EVENT_PROFILE
+
void perf_tp_event(int event_id, u64 addr, u64 count, void *record,
int entry_size)
{
if (!regs)
regs = task_pt_regs(current);
+ /* Trace events already protected against recursion */
do_perf_sw_event(PERF_TYPE_TRACEPOINT, event_id, count, 1,
&data, regs);
}
EXPORT_SYMBOL_GPL(perf_tp_event);
-extern int ftrace_profile_enable(int);
-extern void ftrace_profile_disable(int);
+static int perf_tp_event_match(struct perf_event *event,
+ struct perf_sample_data *data)
+{
+ void *record = data->raw->data;
+
+ if (likely(!event->filter) || filter_match_preds(event->filter, record))
+ return 1;
+ return 0;
+}
static void tp_perf_event_destroy(struct perf_event *event)
{
return &perf_ops_generic;
}
+
+static int perf_event_set_filter(struct perf_event *event, void __user *arg)
+{
+ char *filter_str;
+ int ret;
+
+ if (event->attr.type != PERF_TYPE_TRACEPOINT)
+ return -EINVAL;
+
+ filter_str = strndup_user(arg, PAGE_SIZE);
+ if (IS_ERR(filter_str))
+ return PTR_ERR(filter_str);
+
+ ret = ftrace_profile_set_filter(event, event->attr.config, filter_str);
+
+ kfree(filter_str);
+ return ret;
+}
+
+static void perf_event_free_filter(struct perf_event *event)
+{
+ ftrace_profile_free_filter(event);
+}
+
#else
+
+static int perf_tp_event_match(struct perf_event *event,
+ struct perf_sample_data *data)
+{
+ return 1;
+}
+
static const struct pmu *tp_perf_event_init(struct perf_event *event)
{
return NULL;
}
+
+static int perf_event_set_filter(struct perf_event *event, void __user *arg)
+{
+ return -ENOENT;
+}
+
+static void perf_event_free_filter(struct perf_event *event)
+{
+}
+
+#endif /* CONFIG_EVENT_PROFILE */
+
+#ifdef CONFIG_HAVE_HW_BREAKPOINT
+static void bp_perf_event_destroy(struct perf_event *event)
+{
+ release_bp_slot(event);
+}
+
+static const struct pmu *bp_perf_event_init(struct perf_event *bp)
+{
+ int err;
+ /*
+ * The breakpoint is already filled if we haven't created the counter
+ * through perf syscall
+ * FIXME: manage to get trigerred to NULL if it comes from syscalls
+ */
+ if (!bp->callback)
+ err = register_perf_hw_breakpoint(bp);
+ else
+ err = __register_perf_hw_breakpoint(bp);
+ if (err)
+ return ERR_PTR(err);
+
+ bp->destroy = bp_perf_event_destroy;
+
+ return &perf_ops_bp;
+}
+
+void perf_bp_event(struct perf_event *bp, void *data)
+{
+ struct perf_sample_data sample;
+ struct pt_regs *regs = data;
+
+ sample.addr = bp->attr.bp_addr;
+
+ if (!perf_exclude_event(bp, regs))
+ perf_swevent_add(bp, 1, 1, &sample, regs);
+}
+#else
+static const struct pmu *bp_perf_event_init(struct perf_event *bp)
+{
+ return NULL;
+}
+
+void perf_bp_event(struct perf_event *bp, void *regs)
+{
+}
#endif
atomic_t perf_swevent_enabled[PERF_COUNT_SW_MAX];
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) {
atomic_inc(&perf_swevent_enabled[event_id]);
event->destroy = sw_perf_event_destroy;
struct perf_event_context *ctx,
struct perf_event *group_leader,
struct perf_event *parent_event,
+ perf_callback_t callback,
gfp_t gfpflags)
{
const struct pmu *pmu;
event->state = PERF_EVENT_STATE_INACTIVE;
+ if (!callback && parent_event)
+ callback = parent_event->callback;
+
+ event->callback = callback;
+
if (attr->disabled)
event->state = PERF_EVENT_STATE_OFF;
pmu = tp_perf_event_init(event);
break;
+ case PERF_TYPE_BREAKPOINT:
+ pmu = bp_perf_event_init(event);
+ break;
+
+
default:
break;
}
goto out;
}
-int perf_event_set_output(struct perf_event *event, int output_fd)
+static int perf_event_set_output(struct perf_event *event, int output_fd)
{
struct perf_event *output_event = NULL;
struct file *output_file = NULL;
}
event = perf_event_alloc(&attr, cpu, ctx, group_leader,
- NULL, GFP_KERNEL);
+ NULL, NULL, GFP_KERNEL);
err = PTR_ERR(event);
if (IS_ERR(event))
goto err_put_context;
return err;
}
+/**
+ * perf_event_create_kernel_counter
+ *
+ * @attr: attributes of the counter to create
+ * @cpu: cpu in which the counter is bound
+ * @pid: task to profile
+ */
+struct perf_event *
+perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu,
+ pid_t pid, perf_callback_t callback)
+{
+ struct perf_event *event;
+ struct perf_event_context *ctx;
+ int err;
+
+ /*
+ * Get the target context (task or percpu):
+ */
+
+ ctx = find_get_context(pid, cpu);
+ if (IS_ERR(ctx)) {
+ err = PTR_ERR(ctx);
+ goto err_exit;
+ }
+
+ event = perf_event_alloc(attr, cpu, ctx, NULL,
+ NULL, callback, GFP_KERNEL);
+ if (IS_ERR(event)) {
+ err = PTR_ERR(event);
+ goto err_put_context;
+ }
+
+ event->filp = NULL;
+ WARN_ON_ONCE(ctx->parent_ctx);
+ mutex_lock(&ctx->mutex);
+ perf_install_in_context(ctx, event, cpu);
+ ++ctx->generation;
+ mutex_unlock(&ctx->mutex);
+
+ event->owner = current;
+ get_task_struct(current);
+ mutex_lock(¤t->perf_event_mutex);
+ list_add_tail(&event->owner_entry, ¤t->perf_event_list);
+ mutex_unlock(¤t->perf_event_mutex);
+
+ return event;
+
+ err_put_context:
+ put_ctx(ctx);
+ err_exit:
+ return ERR_PTR(err);
+}
+EXPORT_SYMBOL_GPL(perf_event_create_kernel_counter);
+
/*
* inherit a event from parent task to child task:
*/
child_event = perf_event_alloc(&parent_event->attr,
parent_event->cpu, child_ctx,
group_leader, parent_event,
- GFP_KERNEL);
+ NULL, GFP_KERNEL);
if (IS_ERR(child_event))
return child_event;
get_ctx(child_ctx);
if (parent_event->attr.freq)
child_event->hw.sample_period = parent_event->hw.sample_period;
+ child_event->overflow_handler = parent_event->overflow_handler;
+
/*
* Link it up in the child's context:
*/
{
struct perf_event *parent_event;
- update_event_times(child_event);
perf_event_remove_from_context(child_event);
parent_event = child_event->parent;
* the events from it.
*/
unclone_ctx(child_ctx);
+ update_context_time(child_ctx);
spin_unlock_irqrestore(&child_ctx->lock, flags);
/*
* We dont have to disable NMIs - we are only looking at
* the list, not manipulating it:
*/
- list_for_each_entry_rcu(event, &parent_ctx->event_list, event_entry) {
- if (event != event->group_leader)
- continue;
+ list_for_each_entry(event, &parent_ctx->group_list, group_entry) {
if (!event->attr.inherit) {
inherited_all = 0;
git://ftp.safe.ca / safe / jmp / linux-2.6 / blobdiff