From: Thomas Charbonnel <thomas@undata.org>
To: Ingo Molnar <mingo@elte.hu>
Cc: linux-kernel@vger.kernel.org, "K.R. Foley" <kr@cybsft.com>,
Felipe Alfaro Solana <lkml@felipe-alfaro.com>,
Daniel Schmitt <pnambic@unu.nu>,
Lee Revell <rlrevell@joe-job.com>,
Mark_H_Johnson@raytheon.com
Subject: Re: [patch] voluntary-preempt-2.6.9-rc1-bk4-Q9
Date: Thu, 02 Sep 2004 16:38:33 +0200 [thread overview]
Message-ID: <1094135913.5573.2.camel@localhost> (raw)
In-Reply-To: <20040902132342.GA8677@elte.hu>
Ingo Molnar wrote :
> do you have the NMI watchdog enabled? That could help us debugging this.
> Enabling the APIC/IO_APIC and using nmi_watchdog=1 would be the ideal
> solution - if that doesnt work then nmi_watchdog=2 would be fine too,
> but remove this code from arch/i386/kernel/nmi.c:
>
> if (nmi_watchdog == NMI_LOCAL_APIC)
> nmi_hz = 1;
>
> (otherwise nmi_watchdog=2 would result in one NMI per second, not enough
> to shed light on the above 2.8 msec latency.)
>
> Ingo
nmi_watchdog=1 worked fine.
Here's a trace (there seem to be a counter overflow) :
preemption latency trace v1.0.5 on 2.6.9-rc1-VP-Q9
--------------------------------------------------
latency: 539 us, entries: 38 (38)
-----------------
| task: swapper/0, uid:0 nice:0 policy:0 rt_prio:0
-----------------
=> started at: do_IRQ+0x19/0x190
=> ended at: do_IRQ+0x14a/0x190
=======>
00010000 0.000ms (+0.000ms): do_IRQ (common_interrupt)
00010000 0.000ms (+0.000ms): do_IRQ (default_idle)
00010000 0.000ms (+0.000ms): do_IRQ (<00000000>)
00010001 0.000ms (+0.002ms): mask_and_ack_8259A (do_IRQ)
00010001 0.002ms (+0.000ms): generic_redirect_hardirq (do_IRQ)
00010000 0.002ms (+0.000ms): generic_handle_IRQ_event (do_IRQ)
00010000 0.002ms (+0.000ms): timer_interrupt (generic_handle_IRQ_event)
00010001 0.003ms (+0.005ms): mark_offset_tsc (timer_interrupt)
00010001 0.008ms (+0.000ms): do_timer (timer_interrupt)
00010001 0.008ms (+0.000ms): update_process_times (do_timer)
00010001 0.008ms (+0.000ms): update_one_process (update_process_times)
00010001 0.009ms (+0.000ms): run_local_timers (update_process_times)
00010001 0.009ms (+0.000ms): raise_softirq (update_process_times)
00010001 0.009ms (+0.000ms): scheduler_tick (update_process_times)
00010001 0.009ms (+0.000ms): sched_clock (scheduler_tick)
00010001 0.009ms (+0.000ms): update_wall_time (do_timer)
00010001 0.009ms (+0.000ms): update_wall_time_one_tick
(update_wall_time)
00010001 0.010ms (+0.000ms): generic_note_interrupt (do_IRQ)
00010001 0.010ms (+0.000ms): end_8259A_irq (do_IRQ)
00010001 0.010ms (+0.000ms): enable_8259A_irq (do_IRQ)
00000001 0.011ms (+0.524ms): do_softirq (do_IRQ)
00000001 0.535ms (+0.000ms): do_nmi (mcount)
00010001 0.536ms (+0.000ms): profile_tick (nmi_watchdog_tick)
00010001 0.536ms (+0.000ms): profile_hook (profile_tick)
00010002 0.536ms (+0.000ms): notifier_call_chain (profile_hook)
00010001 0.537ms (+689953.062ms): profile_hit (nmi_watchdog_tick)
00000002 689953.599ms (+1.523ms): sched_clock (activate_task)
00000001 0.537ms (+0.000ms): wake_up_process (do_softirq)
00000001 0.537ms (+0.000ms): try_to_wake_up (wake_up_process)
00000001 0.537ms (+0.000ms): task_rq_lock (try_to_wake_up)
00000002 0.537ms (+0.000ms): activate_task (try_to_wake_up)
00000002 0.538ms (+0.000ms): sched_clock (activate_task)
00000002 0.538ms (+0.000ms): recalc_task_prio (activate_task)
00000002 0.538ms (+0.000ms): effective_prio (recalc_task_prio)
00000002 0.538ms (+0.000ms): enqueue_task (activate_task)
00000001 0.538ms (+0.000ms): preempt_schedule (try_to_wake_up)
00000001 0.539ms (+0.000ms): sub_preempt_count (do_IRQ)
00000001 0.539ms (+0.000ms): update_max_trace (check_preempt_timing)
Thomas
next prev parent reply other threads:[~2004-09-02 14:39 UTC|newest]
Thread overview: 85+ messages / expand[flat|nested] mbox.gz Atom feed top
2004-08-30 19:13 [patch] voluntary-preempt-2.6.9-rc1-bk4-Q5 Mark_H_Johnson
2004-08-30 19:21 ` Ingo Molnar
2004-09-01 12:31 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-Q5 - netdev_max_back_log is too small P.O. Gaillard
2004-09-01 13:05 ` Ingo Molnar
2004-09-02 11:24 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-Q6 - network is no longer smooth P.O. Gaillard
2004-09-02 11:28 ` Ingo Molnar
2004-09-02 15:26 ` P.O. Gaillard
2004-08-31 8:49 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-Q5 Ingo Molnar
2004-09-02 6:33 ` Ingo Molnar
2004-09-02 6:55 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-Q8 Ingo Molnar
2004-09-02 7:04 ` Lee Revell
2004-09-02 7:15 ` Ingo Molnar
2004-09-02 7:31 ` Lee Revell
2004-09-02 7:46 ` Ingo Molnar
2004-09-03 1:10 ` Rusty Russell
2004-09-03 1:10 ` Rusty Russell
2004-09-02 23:25 ` Lee Revell
2004-09-02 23:28 ` Ingo Molnar
2004-09-02 23:32 ` Lee Revell
2004-09-02 7:17 ` Ingo Molnar
2004-09-02 8:23 ` Ingo Molnar
2004-09-02 11:10 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-Q9 Ingo Molnar
2004-09-02 12:14 ` Thomas Charbonnel
2004-09-02 13:16 ` Thomas Charbonnel
2004-09-02 13:23 ` Ingo Molnar
2004-09-02 14:38 ` Thomas Charbonnel [this message]
2004-09-02 21:57 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-R0 Ingo Molnar
2004-09-02 22:06 ` Lee Revell
2004-09-02 22:14 ` Ingo Molnar
2004-09-02 22:15 ` Lee Revell
2004-09-03 0:24 ` Lee Revell
2004-09-03 3:17 ` Eric St-Laurent
2004-09-03 6:26 ` Lee Revell
2004-09-03 6:36 ` Ingo Molnar
2004-09-03 6:49 ` Lee Revell
2004-09-03 7:01 ` Ingo Molnar
2004-09-03 7:05 ` Ingo Molnar
2004-09-03 7:40 ` Lee Revell
2004-09-03 7:50 ` Free Ekanayaka
2004-09-03 8:05 ` Lee Revell
2004-09-03 9:05 ` Free Ekanayaka
2004-09-03 9:25 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-R1 Ingo Molnar
2004-09-03 9:50 ` Luke Yelavich
2004-09-03 10:29 ` Ingo Molnar
2004-09-03 10:43 ` Luke Yelavich
2004-09-03 11:33 ` Thomas Charbonnel
2004-09-03 11:49 ` Ingo Molnar
2004-09-03 12:05 ` Thomas Charbonnel
2004-09-03 16:14 ` Thomas Charbonnel
2004-09-03 17:36 ` Thomas Charbonnel
2004-09-03 11:36 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-R2 Ingo Molnar
2004-09-03 8:09 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-R0 Luke Yelavich
2004-09-03 8:13 ` Lee Revell
2004-09-03 8:21 ` Luke Yelavich
2004-09-03 12:52 ` Luke Yelavich
2004-09-03 18:09 ` K.R. Foley
2004-09-03 11:04 ` K.R. Foley
2004-09-03 17:02 ` K.R. Foley
2004-09-03 20:40 ` Lee Revell
2004-09-03 17:10 ` K.R. Foley
2004-09-03 18:17 ` Ingo Molnar
2004-09-03 18:36 ` K.R. Foley
2004-09-03 19:30 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-R3 Ingo Molnar
2004-09-03 19:49 ` K.R. Foley
2004-09-04 3:39 ` K.R. Foley
2004-09-04 3:43 ` K.R. Foley
2004-09-04 6:41 ` Ingo Molnar
2004-09-04 12:28 ` K.R. Foley
2004-09-04 8:57 ` Ingo Molnar
2004-09-04 10:16 ` Lee Revell
2004-09-04 14:35 ` K.R. Foley
2004-09-04 20:05 ` Ingo Molnar
2004-09-03 18:39 ` [patch] voluntary-preempt-2.6.9-rc1-bk4-R0 Ingo Molnar
2004-09-03 18:41 ` K.R. Foley
-- strict thread matches above, loose matches on Subject: below --
2004-09-02 20:21 [patch] voluntary-preempt-2.6.9-rc1-bk4-Q9 Mark_H_Johnson
2004-09-02 20:38 ` Ingo Molnar
2004-09-02 22:43 ` Ingo Molnar
2004-09-03 9:09 ` Ingo Molnar
2004-09-03 13:42 Mark_H_Johnson
2004-09-03 15:33 Mark_H_Johnson
2004-09-04 0:04 ` Lee Revell
2004-09-04 16:52 ` Alan Cox
2004-09-04 18:05 ` Lee Revell
2004-09-03 18:28 Mark_H_Johnson
2004-09-03 18:33 ` Ingo Molnar
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=1094135913.5573.2.camel@localhost \
--to=thomas@undata.org \
--cc=Mark_H_Johnson@raytheon.com \
--cc=kr@cybsft.com \
--cc=linux-kernel@vger.kernel.org \
--cc=lkml@felipe-alfaro.com \
--cc=mingo@elte.hu \
--cc=pnambic@unu.nu \
--cc=rlrevell@joe-job.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.