Re: [PATCH 0/4] ftrace, workqueuetrace: Add worklet informationame

[Frederic Weisbecker]

On Mon, Apr 20, 2009 at 02:42:11PM +0800, Zhaolei wrote:
> Hello,
> 
> To make workqueuetrace support per-worklet's time mesurement(max execution,
> latecy), workqueue trace need to support per-worklet's stat information first.
> This patchset add them.
> 
> Before patch:
> # cat [debugfs]/tracing/trace_stat/workqueues
> # CPU  INSERTED  EXECUTED   NAME
> # |      |         |          |
> 
>   0    589        589       events/0
>   0      0          0       cpuset
>   0    656        656       khelper
>   0      0          0       netns
>   0    809        809       kblockd/0
>   0      0          0       kacpid
>   0      0          0       kacpi_notify
>   0      0          0       ata/0
>   0      0          0       ata_aux
>   0      0          0       ksuspend_usbd
>   0      0          0       aio/0
>   0      0          0       nfsiod
>   0      0          0       crypto/0
>   0      0          0       kpsmoused
>   0      0          0       kstriped
>   0      0          0       kondemand/0
>   0      0          0       usbhid_resumer
>   0     21         21       rpciod/0
> 
> After patch:
> # cat [debugfs]/tracing/trace_stat/workqueues
> # CPU  INSERTED  EXECUTED   TASKNAME:PID
> # |      |         |        `-WORKFUNC
> # |      |         |          |
>   0    273        273       events/0:5
>   0      1          1        |-reg_todo+0x0/0x648
>   0      4          4        |-check_corruption+0x0/0x2c
>   0      1          1        |-power_supply_changed_work+0x0/0x3c
>   0      6          6        |-console_callback+0x0/0x145
>   0    229        229        |-vmstat_update+0x0/0x33
>   0     12         12        |-do_cache_clean+0x0/0x6a
>   0      1          1        |-wq_barrier_func+0x0/0x12
>   0     15         15        |-mce_work_fn+0x0/0x31
>   0      3          3        |-rt_worker_func+0x0/0x2de
>   0      1          1        `-grace_ender+0x0/0x14
>   0      0          0       cpuset:6
>   0    659        659       khelper:7
>   0    659        659        `-__call_usermodehelper+0x0/0xa0
>   0      0          0       netns:11
>   0   1818       1818       kblockd/0:13
>   0   1115       1115        |-blk_unplug_work+0x0/0x1e
>   0    703        703        `-cfq_kick_queue+0x0/0x2f
>   0      0          0       kacpid:412
>   0      0          0       kacpi_notify:415
>   0      0          0       ata/0:463
>   0      0          0       ata_aux:509
>   0      0          0       ksuspend_usbd:515
>   0      0          0       aio/0:717
>   0      0          0       nfsiod:764
>   0      0          0       crypto/0:785
>   0      0          0       kpsmoused:1141
>   0      0          0       kstriped:1248
>   0      0          0       kondemand/0:1259
>   0      0          0       usbhid_resumer:1264
>   0     20         20       rpciod/0:1333
>   0      1          1        |-xs_tcp_connect_worker4+0x0/0x1b
>   0     19         19        `-xs_udp_connect_worker4+0x0/0x130



Oh, Nice!
This is an interesting view of worklets percentage/frequency of execution.
We could also have a file per workqueue, but anyway it seems to me a very good
first step!


> 
> After this step, we'll begin to add worklet time mesurement support to
> workqueuetrace.


Great!
Thanks.


> Thanks
> Zhaolei
> 

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/