Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 19 Jul 2006 17:06:43 +0300
From:      "victor cruceru" <victor.cruceru@gmail.com>
To:        freebsd-hackers@freebsd.org
Subject:   Re: "swiN: clock sio" process taking 75% CPU
Message-ID:  <49402550607190706q6548ebe7lb9c4dd1ff66b3f1@mail.gmail.com>

next in thread | raw e-mail | index | archive | help
Hi Gareth,
Did you try to disable the console screensaver?
Sometimes this helps.

> Message: 1
> Date: Tue, 18 Jul 2006 13:17:32 +0100
> From: Gareth McCaughan <gmccaughan@synaptics-uk.com>
> Subject: "swiN: clock sio" process taking 75% CPU
> To: freebsd-hackers@freebsd.org
> Message-ID: <200607181317.33416.gmccaughan@synaptics-uk.com>
> Content-Type: text/plain;  charset="us-ascii"
>
> (I've asked this question on -questions and -stable, with no success;
> hence I'm taking it to the assembled wizardry of -hackers. A bit of
> googling suggests that I'm far from the first person to have had a
> similar problem, though it seems to be worse for me than for the
> others I've found.)
>
> I have a box running 6-STABLE, cvsupped last week.
> Until recently it was running 5.something and showed
> the same peculiar behaviour as I'm about to describe.
> Further back, it used to run 4.x, and I don't recall
> anything like this happening then.
>
> About 6 minutes after booting (on three occasions, but I
> don't guarantee this doesn't vary), a process (well, a
> kernel interrupt thread, I guess) that appears in the
> output of "ps" as "[swi4: clock sio]" begins to use
> about 3/4 of the machine's CPU. I think it does so
> more or less instantaneously. It continues to do so
> indefinitely, so far as I can tell.
>
> I'm not aware of anything specific that triggers this,
> though I suppose there must *be* something. It happens
> apparently spontaneously, on a lightly loaded machine.
>
> Those cycles are genuinely being consumed; other processes
> run much more slowly than they "should", and take much more
> wall time than CPU time.
>
> I've tried diddling my kernel's HZ value; the behaviour
> with HZ=100 and with HZ=1000 is the same, so far as I'm
> able to tell. I've no idea whether it might be relevant,
> but I have option DEVICE_POLLING turned on; toggling
> sysctl kern.polling.enable doesn't seem to make any
> difference.
>
> The machine is a very uninteresting single-CPU Athlon box,
> clocked at 1.6GHz, several years old. Here's its dmesg output,
> with a few uninteresting bits of information leakage elided.
>

-- 
victor cruceru
------------------------------------------------
Non est respondendum ad omnia.
( Cicero, Pro Murena Oratio )
------------------------------------------------



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?49402550607190706q6548ebe7lb9c4dd1ff66b3f1>