linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Rich Felker <dalias@libc.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>,
	devicetree@vger.kernel.org, linux-kernel@vger.kernel.org,
	linux-sh@vger.kernel.org, Rob Herring <robh+dt@kernel.org>,
	Mark Rutland <mark.rutland@arm.com>,
	"Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Subject: Re: [PATCH v7 2/2] clocksource: add J-Core timer/clocksource driver
Date: Sat, 8 Oct 2016 12:26:14 -0400	[thread overview]
Message-ID: <20161008162614.GY19318@brightrain.aerifal.cx> (raw)
In-Reply-To: <alpine.DEB.2.20.1610081312190.5222@nanos>

On Sat, Oct 08, 2016 at 01:32:06PM +0200, Thomas Gleixner wrote:
> On Fri, 7 Oct 2016, Rich Felker wrote:
> > 
> > If I'm not mistaken, the bug is in tick_nohz_restart. According to the
> 
> I think you are mistaken. Let's look at CPU0 only:

OK.

>           <idle>-0     [000] d...   150.829698: __tick_nohz_idle_enter: can stop idle tick
>           <idle>-0     [000] d...   150.829774: __tick_nohz_idle_enter: entering stop_sched_tick
>           <idle>-0     [000] d...   150.830001: tick_stop: success=1 dependency=NONE
>           <idle>-0     [000] d...   150.830127: hrtimer_cancel: hrtimer=109e949c
>           <idle>-0     [000] d...   150.830380: jcore_pit_set_next_event: pit 109eccb0 0 set delta 82124660 en=4410000 cntr=75 throt=82124660
>           <idle>-0     [000] d...   150.830498: hrtimer_start: hrtimer=109e949c function=tick_sched_timer expires=150840000000 softexpires=150840000000
>           <idle>-0     [000] d...   150.830720: jcore_pit_set_next_event: pit 109eccb0 0 set delta 539823 en=4410000 cntr=34 throt=539823
> 
> Tick is programmed to fire at 150.84
> 
>           <idle>-0     [000] d...   150.830822: __tick_nohz_idle_enter: returning from stop_sched_tick
>           <idle>-0     [000] d...   150.830940: rcu_dyntick: Start 140000000000000 0
>           <idle>-0     [000] d...   150.831072: rcu_dyntick: End 0 1
>           <idle>-0     [000] d...   150.831165: cpu_idle: state=0 cpu_id=0
>           <idle>-0     [000] d...   150.831269: rcu_dyntick: Start 1 0
>           <idle>-0     [000] ....   150.831368: cpu_startup_entry: enabled irqs, spinning
> 
> CPU spins and waits for an interrupt to happen
> 
> 
>           <idle>-0     [000] d...   150.841530: rcu_dyntick: End 0 1
> 
> Dropping out of the spin about the time we expect the PIT interrupt to
> fire. And an interrupt is the reason why we drop out because the 'need
> resched' flag is not set and we end up in:

OK, I missed that.

>           <idle>-0     [000] d.s.   150.841724: tick_irq_enter: update jiffies via irq
> 
> which is called from irq_enter()
> 
>           <idle>-0     [000] d.s.   150.841918: tick_do_update_jiffies64: finished do_timer(1)
>           <idle>-0     [000] d.s.   150.842348: tick_do_update_jiffies64: finished updating jiffies
> 
> 
> So here we would expect:
> 
>     	 irq_handler_entry: irq=16 name=jcore_pit
> 	 ...
> 	 irq_handler_exit .....
>     	 
> 
> or any other interrupt being invoked. But we see this here:

According to the trace the 'irq' is soft ('s'). I couldn't find the
code path from the idle loop to softirq but so maybe this is a bug. Is
it possible that in some cases the arch irq entry does not get
identified as a hard irq or traced and then the subsequent tick code
thinks it's running in a softirq and behaves differently? I could add
more tracing around irq entry.

>           <idle>-0     [000] d...   150.842603: __tick_nohz_idle_enter: can stop idle tick
> 
> And that's just wrong. 

Can you explain?

> Now looking at CPU1 at the same time:
> 
>           <idle>-0     [001] d.h.   150.841829: irq_handler_entry: irq=16 name=jcore_
>           <idle>-0     [001] d.h.   150.842054: hrtimer_cancel: hrtimer=109f449c
>           <idle>-0     [001] d.h.   150.842218: hrtimer_expire_entry: hrtimer=109f449c function=tick_sched_timer now=150840579940
> 
> So there the PIT interrupt enters and runs the interrupt handler.
> 
> Now looking deeper at the PIT interrupts:
> 
>      kworker/0:1-208   [000] d.h.   150.641822: irq_handler_entry: irq=16 name=jcore_pit
>      kworker/0:1-208   [000] d.h.   150.644928: irq_handler_exit: irq=16 ret=handled
> 
> 
>           <idle>-0     [001] dnh.   150.645370: irq_handler_entry: irq=16 name=jcore_pit
>           <idle>-0     [001] dnh.   150.647444: irq_handler_exit: irq=16 ret=handled
> 
> Both CPUs have the same interrupt number for their per cpu PIT interrupt.
> 
> IIRC you said earlier when the percpu interrupt discussion happened, that
> your per cpu PITs have distinct interrupt numbers, but that does not seem
> the case here.

No, I said the opposite. They use the same irq number but they're each
wired to their own cpu, and there's no way for them to fire on the
wrong cpu.

Rich

  reply	other threads:[~2016-10-08 16:26 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-09-24  5:07 [PATCH v7 0/2] J-Core timer support Rich Felker
2016-09-24  5:07 ` [PATCH v7 2/2] clocksource: add J-Core timer/clocksource driver Rich Felker
2016-09-26 21:07   ` Rich Felker
2016-09-26 21:27     ` Daniel Lezcano
2016-09-26 22:23       ` Rich Felker
2016-09-26 23:55         ` Thomas Gleixner
2016-09-27  0:42           ` Rich Felker
2016-09-27 22:08             ` Rich Felker
2016-09-30 13:15               ` Thomas Gleixner
2016-09-30 13:48                 ` Paul E. McKenney
2016-10-01 17:05                   ` Rich Felker
2016-10-01 17:58                     ` Paul E. McKenney
2016-10-02  0:00                       ` Rich Felker
2016-10-02  3:59                         ` Rich Felker
2016-10-02  5:59                           ` Paul E. McKenney
2016-10-02  6:30                         ` Paul E. McKenney
2016-10-08  1:32                 ` Rich Felker
2016-10-08 11:32                   ` Thomas Gleixner
2016-10-08 16:26                     ` Rich Felker [this message]
2016-10-08 17:03                       ` Thomas Gleixner
2016-10-09  1:28                         ` Rich Felker
2016-10-09  9:14                           ` Thomas Gleixner
2016-10-09 14:35                             ` Rich Felker
2016-10-03 22:10       ` Rich Felker
2016-10-04  7:06         ` Paul E. McKenney
2016-10-04 20:58           ` Rich Felker
2016-10-04 21:14             ` Paul E. McKenney
2016-10-04 21:48               ` Rich Felker
2016-10-05 12:41                 ` Paul E. McKenney
2016-09-24  5:07 ` [PATCH v7 1/2] of: add J-Core timer bindings Rich Felker

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=20161008162614.GY19318@brightrain.aerifal.cx \
    --to=dalias@libc.org \
    --cc=daniel.lezcano@linaro.org \
    --cc=devicetree@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-sh@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=paulmck@linux.vnet.ibm.com \
    --cc=robh+dt@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;
as well as URLs for NNTP newsgroup(s).