From: Jay Lan <jlan@sgi.com>
To: Shailabh Nagar <nagar@watson.ibm.com>
Cc: Andrew Morton <akpm@osdl.org>,
jlan@engr.sgi.com, balbir@in.ibm.com, csturtiv@sgi.com,
linux-kernel@vger.kernel.org
Subject: Re: [Patch][RFC] Disabling per-tgid stats on task exit in taskstats
Date: Mon, 26 Jun 2006 10:33:04 -0700 [thread overview]
Message-ID: <44A01A50.1050403@sgi.com> (raw)
In-Reply-To: <449CD4B3.8020300@watson.ibm.com>
Shailabh Nagar wrote:
> Andrew Morton wrote:
>
>> On Fri, 23 Jun 2006 22:59:04 -0400
>> Shailabh Nagar <nagar@watson.ibm.com> wrote:
>>
>>
>>
>>>>> It was due to a loop in fill_tgid() when per-TG stats
>>>>> data are assembled for netlink:
>>>>> do {
>>>>> rc = delayacct_add_tsk(stats, tsk);
>>>>> if (rc)
>>>>> break;
>>>>>
>>>>> } while_each_thread(first, tsk);
>>>>>
>>>>> and it is executed inside a lock.
>>>>> Fortunately single threaded appls do not hit this code.
>>>>>
>>>>>
>>>>
>>>> Am I reading this right? We do that loop when each thread within the
>>>> thread group exits?
>>>>
>>>>
>>>
>>> Yes.
>>>
>>>
>>>
>>>> How come?
>>>>
>>>>
>>>>
>>>
>>> To get the sum of all per-tid data for threads that are currently alive.
>>> This is returned to userspace with each thread exit.
>>>
>>
>>
>> I realise that. How about we stop doing it?
>>
>> When a thread exits it only makes sense to send up the stats for that
>> thread.
>
>
>> Why does the kernel assume that userspace is also interested in
>> the accumulated stats of its siblings? And if userspace _is_
>> interested in
>> that info, it's still present in-kernel and can be queried for.
>>
>>
> The reason for sending out sum of siblings's stats was as follows:
> I didn't maintain a per-tgid data structure in-kernel where the exiting
> threads taskstats could be accumalated
> , erroneously thinking that this would require such a structure to be
> *additionally* updated each time a statististic
> was being collected and that would be way too much overhead. Also to
> save on space. Thus if userspace wants to get the per-tgid stats for the
> thread group when the last thread exits, then it cannot
> do so by querying since such a query only returns the sum of currently
> live threads (data from exited threads is lost).
>
> So, the current design chooses to return the sum of all siblings + self
> when each thread exits. Using this userspace
> can maintain the per-tgid data for all currently living threads of the
> group + previously exited threads.
>
> But as pointed out in an earlier mail, it looks like this is
> unnecessarily elaborate way of trying to avoid maintaining
> a separate per-tgid data structure in the kernel (in addition to the
> per-tid ones we already have).
>
> What can be done is to create a taskstats structure for a thread group
> the moment the *second* thread gets created.
> Then each exiting thread can accumalate its stats to this struct. If
> userspace queries for per-tgid data, the sum of all
> live threads + value in this struct can be returned. And when the last
> thread of the thread group exits, the struct's
> value can be output.
>
> While this will mean an extra taskstats structure hanging around for the
> lifetime of a multithreaded app (not single threaded
> ones), it should cut down on the overhead of running through all threads
> that we see in the current design.
> More importantly, it will reduce the frequency of per-tgid data send to
> once for each thread group exit instead of once
> per thread exit.
>
> Will that work for everyone ?
As long as the per-pid delayacct struct has a pointer to the per-tgid
data struct and deoes not need to go through the loop on every exit.
>
>>>> Is there some better lock we can use in there? It only has to be
>>>> threadgroup-wide rather than kernel-wide.
>>>>
>>>>
>>>>
>>>
>>> The lock we're holding is the tasklist_lock. To go through all the
>>> threads of a thread group
>>> thats the only lock that can protect integrity of while_each_thread
>>> afaics.
>>>
>>
>>
>> At present, yes. That's persumably not impossible to fix.
>>
>>
> In the above design, if a userspace query for per-tgid data arrives,
> then I'll still need to run through
> all the threads of a thread group (to return their sum + that of already
> exited threads accumalated in the
> extra per-tgid taskstats struct).
But, this query-reply logic can be separated from that executed at
exit.
Thanks,
- jay
>
> So that could still benefit from such a thread group specific lock.
> Scope of change is a bit more of course
> so will need to take a closer look.
>
> --Shailabh
next prev parent reply other threads:[~2006-06-26 17:33 UTC|newest]
Thread overview: 151+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-06-09 7:41 [Patch][RFC] Disabling per-tgid stats on task exit in taskstats Shailabh Nagar
2006-06-09 8:00 ` Andrew Morton
2006-06-09 10:51 ` Balbir Singh
2006-06-09 11:21 ` Andrew Morton
2006-06-09 13:20 ` Shailabh Nagar
2006-06-09 18:25 ` Jay Lan
2006-06-09 19:12 ` Shailabh Nagar
2006-06-09 15:36 ` Balbir Singh
2006-06-09 18:35 ` Jay Lan
2006-06-09 19:31 ` Shailabh Nagar
2006-06-09 21:56 ` Shailabh Nagar
2006-06-09 22:42 ` Jay Lan
2006-06-09 23:22 ` Andrew Morton
2006-06-09 23:47 ` Jay Lan
2006-06-09 23:56 ` Andrew Morton
2006-06-10 12:21 ` Shailabh Nagar
2006-06-12 18:31 ` Jay Lan
2006-06-12 21:57 ` Shailabh Nagar
2006-06-10 13:05 ` Shailabh Nagar
2006-06-12 18:54 ` Jay Lan
2006-06-21 19:11 ` Jay Lan
2006-06-21 19:14 ` Jay Lan
2006-06-21 19:34 ` Shailabh Nagar
2006-06-21 23:35 ` Jay Lan
2006-06-21 23:45 ` Shailabh Nagar
2006-06-23 17:14 ` Shailabh Nagar
2006-06-23 18:19 ` Jay Lan
2006-06-23 18:53 ` Shailabh Nagar
2006-06-23 20:00 ` Jay Lan
2006-06-23 20:16 ` Shailabh Nagar
2006-06-23 20:36 ` Jay Lan
2006-06-23 21:19 ` Andrew Morton
2006-06-23 22:07 ` Jay Lan
2006-06-23 23:47 ` Andrew Morton
2006-06-24 2:59 ` Shailabh Nagar
2006-06-24 4:39 ` Andrew Morton
2006-06-24 5:59 ` Shailabh Nagar
2006-06-26 17:33 ` Jay Lan [this message]
2006-06-26 17:52 ` Shailabh Nagar
2006-06-26 17:55 ` Andrew Morton
2006-06-26 18:00 ` Shailabh Nagar
2006-06-26 18:12 ` Andrew Morton
2006-06-26 18:26 ` Jay Lan
2006-06-26 18:39 ` Andrew Morton
2006-06-26 18:49 ` Shailabh Nagar
2006-06-26 19:00 ` Jay Lan
2006-06-28 21:30 ` Jay Lan
2006-06-28 21:53 ` Andrew Morton
2006-06-28 22:02 ` Jay Lan
2006-06-29 8:40 ` Paul Jackson
2006-06-29 12:30 ` Valdis.Kletnieks
2006-06-29 16:44 ` Paul Jackson
2006-06-29 18:01 ` Andrew Morton
2006-06-29 18:07 ` Paul Jackson
2006-06-29 18:26 ` Paul Jackson
2006-06-29 19:15 ` Shailabh Nagar
2006-06-29 19:41 ` Paul Jackson
2006-06-29 21:42 ` Shailabh Nagar
2006-06-29 21:54 ` Jay Lan
2006-06-29 22:09 ` Shailabh Nagar
2006-06-29 22:23 ` Paul Jackson
2006-06-30 0:15 ` Shailabh Nagar
2006-06-30 0:40 ` Paul Jackson
2006-06-30 1:00 ` Shailabh Nagar
2006-06-30 1:05 ` Paul Jackson
[not found] ` <44A46C6C.1090405@watson.ibm.com>
2006-06-30 0:38 ` Paul Jackson
2006-06-30 2:21 ` Paul Jackson
2006-06-30 2:46 ` Shailabh Nagar
2006-06-30 2:54 ` Paul Jackson
2006-06-30 3:02 ` Paul Jackson
2006-06-29 19:22 ` Shailabh Nagar
2006-06-29 19:10 ` Shailabh Nagar
2006-06-29 19:10 ` Shailabh Nagar
2006-06-29 19:23 ` Paul Jackson
2006-06-29 19:33 ` Andrew Morton
2006-06-29 19:43 ` Shailabh Nagar
2006-06-29 19:43 ` Shailabh Nagar
2006-06-29 20:00 ` Andrew Morton
2006-06-29 22:13 ` Shailabh Nagar
2006-06-29 22:13 ` Shailabh Nagar
2006-06-29 23:00 ` jamal
2006-06-29 20:01 ` Shailabh Nagar
2006-06-29 20:01 ` Shailabh Nagar
2006-06-29 21:22 ` Paul Jackson
2006-06-29 22:54 ` jamal
2006-06-30 0:38 ` Shailabh Nagar
2006-06-30 0:38 ` Shailabh Nagar
2006-06-30 1:05 ` Andrew Morton
2006-06-30 1:11 ` Shailabh Nagar
2006-06-30 1:11 ` Shailabh Nagar
2006-06-30 1:30 ` jamal
2006-06-30 3:01 ` Shailabh Nagar
2006-06-30 3:01 ` Shailabh Nagar
2006-06-30 12:45 ` jamal
2006-06-30 2:25 ` Paul Jackson
2006-06-30 2:35 ` Andrew Morton
2006-06-30 2:43 ` Paul Jackson
2006-06-29 19:33 ` Jay Lan
2006-06-30 18:53 ` Shailabh Nagar
2006-06-30 18:53 ` Shailabh Nagar
2006-06-30 19:10 ` Shailabh Nagar
2006-06-30 19:10 ` Shailabh Nagar
2006-06-30 19:19 ` Shailabh Nagar
2006-06-30 19:19 ` Shailabh Nagar
2006-06-30 20:19 ` jamal
2006-06-30 22:50 ` Andrew Morton
2006-07-01 2:20 ` Shailabh Nagar
2006-07-01 2:20 ` Shailabh Nagar
2006-07-01 2:43 ` Andrew Morton
2006-07-01 3:37 ` Shailabh Nagar
2006-07-01 3:37 ` Shailabh Nagar
2006-07-01 3:51 ` Andrew Morton
2006-07-03 21:11 ` Shailabh Nagar
2006-07-03 21:41 ` Andrew Morton
2006-07-04 0:13 ` Shailabh Nagar
2006-07-04 0:38 ` Andrew Morton
2006-07-04 20:19 ` Paul Jackson
2006-07-04 20:22 ` Paul Jackson
2006-07-04 0:54 ` Shailabh Nagar
2006-07-04 1:01 ` Andrew Morton
2006-07-04 13:05 ` jamal
2006-07-04 15:18 ` Shailabh Nagar
2006-07-04 16:37 ` Shailabh Nagar
2006-07-04 19:24 ` jamal
2006-07-05 14:09 ` Shailabh Nagar
2006-07-05 14:09 ` Shailabh Nagar
2006-07-05 20:25 ` Chris Sturtivant
2006-07-05 20:25 ` Chris Sturtivant
2006-07-05 20:32 ` Shailabh Nagar
2006-07-05 20:32 ` Shailabh Nagar
2006-07-03 4:53 ` Paul Jackson
2006-07-03 15:02 ` Shailabh Nagar
2006-07-03 15:55 ` Paul Jackson
2006-07-03 16:31 ` Paul Jackson
2006-07-04 0:09 ` Shailabh Nagar
2006-07-04 19:59 ` Paul Jackson
2006-07-05 17:20 ` Jay Lan
2006-07-05 17:20 ` Jay Lan
2006-07-05 18:18 ` Shailabh Nagar
2006-07-05 18:18 ` Shailabh Nagar
2006-06-30 22:56 ` Andrew Morton
2006-06-29 18:05 ` Nick Piggin
2006-06-29 12:42 ` Shailabh Nagar
2006-06-24 3:08 ` Shailabh Nagar
2006-06-21 20:38 ` Andrew Morton
2006-06-21 21:31 ` Shailabh Nagar
2006-06-21 21:45 ` Jay Lan
2006-06-21 21:54 ` Andrew Morton
2006-06-21 22:19 ` Jay Lan
2006-06-21 21:59 ` Shailabh Nagar
2006-06-09 15:55 ` Chris Sturtivant
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=44A01A50.1050403@sgi.com \
--to=jlan@sgi.com \
--cc=akpm@osdl.org \
--cc=balbir@in.ibm.com \
--cc=csturtiv@sgi.com \
--cc=jlan@engr.sgi.com \
--cc=linux-kernel@vger.kernel.org \
--cc=nagar@watson.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.