From: Haris Okanovic <haris.okanovic@ni.com>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Anna-Maria Gleixner <anna-maria@linutronix.de>,
Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
linux-rt-users@vger.kernel.org, linux-kernel@vger.kernel.org,
julia.cartwright@ni.com, gratian.crisan@ni.com,
harisokn@gmail.com
Subject: Re: [PATCH] Revert "timers: Don't wake ktimersoftd on every tick"
Date: Mon, 17 Jul 2017 16:58:42 -0500 [thread overview]
Message-ID: <b8cec915-f6f5-8249-d72a-ecf9566ce6ee@ni.com> (raw)
In-Reply-To: <alpine.DEB.2.20.1706041558150.2813@nanos>
On 06/04/2017 09:17 AM, Thomas Gleixner wrote:
> On Fri, 2 Jun 2017, Haris Okanovic wrote:
>> On 05/26/2017 03:50 PM, Thomas Gleixner wrote:
>>>>> static void expire_timers(struct timer_base *base)
>>>>> {
>>>>> struct hlist_head *head;
>>>>> + int expCount = base->expired_count;
>>>
>>> No camel case for heavens sake!
>>>
>>> And this requires:
>>>
>>> cnt = READ_ONCE(base->expired_count);
>>>
>>>>> - while (base->expired_count--) {
>>>>> - head = base->expired_lists + base->expired_count;
>>>>> + while (expCount--) {
>>>>> + head = base->expired_lists + expCount;
>>>>> __expire_timers(base, head);
>>>>> }
>>>
>>> Plus a comment.
>>
>> Fixed, thanks.
>>
>> Are your recommending READ_ONCE() purely for documentation purposes?
>
> Yes.
>
>>> The other thing I noticed was this weird condition which does not do the
>>> look ahead when base->clk is back for some time.
>>
>> The soft interrupt fires unconditionally if base->clk hasn't advanced in some
>> time to limit how long cpu spends in hard interrupt context.
>
> That makes no sense.
>
I wrote this part out of an abundance of caution: I didn't want a
potentially unbounded operation to run in hardirq context. This logic
forces both the update to timer bases & firing of timers into softirq
context if the clock advances by a lot (some arbitrary large number of
ticks, HZ in this case).
However, I think you're right that this is unneeded since
run_local_timers() is called per tick, and thus would never exercise
this case.
Removed this case.
>>> Why don't you use the
>>> existing optimization which uses the bitmap for fast forward?
>>>
>>
>> Are you referring to forward_timer_base()/base->next_expiry? I think it's only
>> updated in the nohz case. Can you share function name/line number(s) if you're
>> thinking of something else.
>
> I think just using collect_expired_timers() should be enough. In the !NOHZ
> case the base shouldn't be that far back, right?
>
Refactored.
>>> The other issue I have is that this can race at all. If you raised the
>>> softirq in the look ahead then you should not go into that function until
>>> the softirq has actually completed. There is no point in wasting time in
>>> the hrtimer interrupt if the softirq is running anyway.
>>>
>>
>> Makes sense. Skipping the large `if` block in run_local_timers() when
>> `local_softirq_pending() & TIMER_SOFTIRQ`.
>
> No. You need your own state tracking. The TIMER_SOFTIRQ bit is cleared when
> the softirq is invoked, but that does not mean that it finished running.
>
> run_local_timers()
> {
> lock(base->lock);
> if (!base->softirq_activated)
> if (base_has_timers_to_expire()) {
> base->softirq_activated = true;
> raise_softirq(TIMER_SOFTIRQ);
> }
> }
> unlock(base->lock);
> }
>
> timer_softirq()
> {
> lock(base->lock);
> expire_timers();
> base->softirq_activated = false;
> unlock(base->lock);
> }
>
> That way you avoid any operation in the tick interrupt as long as the soft
> interrupt processing has not completed.
Added a per-cpu block_softirq boolean.
>
> Thanks,
>
> tglx
>
I'll post a v2 patch shortly.
Thanks,
Haris
next prev parent reply other threads:[~2017-07-17 21:58 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-12-13 21:44 [RFC v2] timers: Don't wake ktimersoftd on every tick Haris Okanovic
2016-12-23 17:28 ` Sebastian Andrzej Siewior
2016-12-28 18:06 ` Haris Okanovic
2017-02-03 16:51 ` Sebastian Andrzej Siewior
2017-02-03 18:21 ` [PATCH] " Haris Okanovic
2017-02-10 17:02 ` Sebastian Andrzej Siewior
2017-05-26 17:16 ` [PATCH] Revert "timers: Don't wake ktimersoftd on every tick" Anna-Maria Gleixner
2017-05-26 19:27 ` Haris Okanovic
2017-05-26 19:49 ` Thomas Gleixner
2017-05-26 20:25 ` Haris Okanovic
2017-05-26 20:50 ` Thomas Gleixner
2017-06-02 14:37 ` Haris Okanovic
2017-06-04 14:17 ` Thomas Gleixner
2017-07-17 21:58 ` Haris Okanovic [this message]
2017-07-17 22:04 ` [PATCH v2] timers: Don't wake ktimersoftd on every tick Haris Okanovic
2017-07-18 21:33 ` Thomas Gleixner
2017-08-03 21:04 ` Haris Okanovic
2017-08-03 21:06 ` [PATCH v3 1/2] " Haris Okanovic
2017-08-03 21:06 ` [PATCH v3 2/2] timers: Don't search for expired timers while TIMER_SOFTIRQ is scheduled Haris Okanovic
2018-01-05 19:37 ` Haris Okanovic
2018-03-01 15:49 ` Haris Okanovic
2018-03-01 15:54 ` Thomas Gleixner
2018-03-01 16:35 ` Haris Okanovic
2018-03-01 16:47 ` Sebastian Andrzej Siewior
2018-03-01 18:37 ` Haris Okanovic
2018-03-01 19:06 ` Sebastian Andrzej Siewior
2018-03-02 14:52 ` Sebastian Andrzej Siewior
2018-03-02 16:29 ` Haris Okanovic
2018-03-02 16:39 ` Sebastian Andrzej Siewior
2018-03-02 17:19 ` Haris Okanovic
2018-03-06 23:39 ` [PATCH v4 1/2] timers: Don't wake ktimersoftd on every tick Haris Okanovic
2018-03-06 23:39 ` [PATCH v4 2/2] timers: Don't search for expired timers while TIMER_SOFTIRQ is scheduled Haris Okanovic
[not found] ` <1523332961.4827.1.camel@gmx.de>
2018-04-12 15:00 ` [PATCH v4 1/2] timers: Don't wake ktimersoftd on every tick Haris Okanovic
2018-06-19 12:43 ` Daniel Bristot de Oliveira
2018-06-20 14:24 ` Haris Okanovic
2018-06-28 16:36 ` Haris Okanovic
2018-06-28 16:40 ` [PATCH v5 " Haris Okanovic
2018-06-28 16:40 ` [PATCH v5 2/2] timers: Don't search for expired timers while TIMER_SOFTIRQ is scheduled Haris Okanovic
2018-07-13 12:01 ` [PATCH v5 1/2] timers: Don't wake ktimersoftd on every tick Anna-Maria Gleixner
2018-07-13 14:37 ` Haris Okanovic
2017-05-27 7:47 ` [PATCH] Revert "timers: Don't wake ktimersoftd on every tick" Sebastian Andrzej Siewior
2017-02-03 18:27 ` [RFC v2] timers: Don't wake ktimersoftd on every tick Haris Okanovic
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=b8cec915-f6f5-8249-d72a-ecf9566ce6ee@ni.com \
--to=haris.okanovic@ni.com \
--cc=anna-maria@linutronix.de \
--cc=bigeasy@linutronix.de \
--cc=gratian.crisan@ni.com \
--cc=harisokn@gmail.com \
--cc=julia.cartwright@ni.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rt-users@vger.kernel.org \
--cc=tglx@linutronix.de \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®