Re: [RFC 00/13] perf: Add perf kwork

[Yang Jihong <yangjihong1@huawei.com>]

Hello,

On 2022/6/23 22:25, Paul A. Clarke wrote:
> On Thu, Jun 23, 2022 at 09:02:45AM +0800, Yang Jihong wrote:
>> On 2022/6/22 22:59, Paul A. Clarke wrote:
>>> On Mon, Jun 13, 2022 at 05:45:52PM +0800, Yang Jihong wrote:
>>>> Sometimes, we need to analyze time properties of kernel work such as irq,
>>>> softirq, and workqueue, such as delay and running time of specific interrupts.
>>>> Currently, these events have kernel tracepoints, but perf tool does not
>>>> directly analyze the delay of these events
>>>>
>>>> The perf-kwork tool is used to trace time properties of kernel work
>>>> (such as irq, softirq, and workqueue), including runtime, latency,
>>>> and timehist, using the infrastructure in the perf tools to allow
>>>> tracing extra targets
>>>>
>>>> test case:
>>>>
>>>>     # perf kwork report
>>>>
>>>>       Kwork Name                | Cpu  | Total Runtime | Frequency | Max runtime   | Max runtime start   | Max runtime end     |
>>>>      ---------------------------------------------------------------------------------------------------------------------------
>>>>       (s)RCU:9                  | 0007 |      3.488 ms |      1258 |      0.145 ms |    3398384.220013 s |    3398384.220157 s |
>>>>       (s)NET_RX:3               | 0003 |      1.866 ms |       156 |      0.042 ms |    3398385.629764 s |    3398385.629806 s |
>>>>       (s)TIMER:1                | 0000 |      1.799 ms |       117 |      0.055 ms |    3398385.568033 s |    3398385.568088 s |
>>>>       (w)0xffff9c66e563ee98     | 0006 |      1.561 ms |         5 |      0.351 ms |    3398384.060021 s |    3398384.060371 s |
>>>
>>> What units are used for "Frequency"? It would be helpful to include somewhere.
> 
>> This refers to the number of event in the trace period.
> 
> I see. I suggest changing the column heading to say that.
> "Count", "Number of Events", "Occurrences", something like that.
> 
> "Frequency" is a count-per-time-unit, like hertz, which doesn't
> match well here.

OK, I'll fix in next version.

Thanks,
Jihong