提交 7f8b4e4e 编写于 作者: P Peter Zijlstra 提交者: Ingo Molnar

perf_counter: Add scale information to the mmap control page

Add the needed time scale to the self-profile mmap information.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: NIngo Molnar <mingo@elte.hu>
上级 41f95331
......@@ -232,12 +232,14 @@ struct perf_counter_mmap_page {
__u32 lock; /* seqlock for synchronization */
__u32 index; /* hardware counter identifier */
__s64 offset; /* add to hardware counter value */
__u64 time_enabled; /* time counter active */
__u64 time_running; /* time counter on cpu */
/*
* Hole for extension of the self monitor capabilities
*/
__u64 __reserved[125]; /* align to 1k */
__u64 __reserved[123]; /* align to 1k */
/*
* Control data for the mmap() data buffer.
......
......@@ -1782,6 +1782,12 @@ void perf_counter_update_userpage(struct perf_counter *counter)
if (counter->state == PERF_COUNTER_STATE_ACTIVE)
userpg->offset -= atomic64_read(&counter->hw.prev_count);
userpg->time_enabled = counter->total_time_enabled +
atomic64_read(&counter->child_total_time_enabled);
userpg->time_running = counter->total_time_running +
atomic64_read(&counter->child_total_time_running);
barrier();
++userpg->lock;
preempt_enable();
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册