Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 03 Jul 2006 08:34:34 -0400
From:      Chuck Swiger <cswiger@mac.com>
To:        Nikos Vassiliadis <nvass@teledomenet.gr>
Cc:        nocturnal <nocturnal@swehack.se>, freebsd-questions@freebsd.org
Subject:   Re: Interrupt using all the CPU
Message-ID:  <44A90EDA.6040707@mac.com>
In-Reply-To: <200607031504.39332.nvass@teledomenet.gr>
References:  <44A8D770.5060808@swehack.se> <44A8F0D6.4060307@swehack.se> <200607031306.12477.Danovitsch@vitsch.net> <200607031504.39332.nvass@teledomenet.gr>

next in thread | previous in thread | raw e-mail | index | archive | help
Nikos Vassiliadis wrote:
> On Monday 03 July 2006 14:06, Daan Vreeken [PA4DAN] wrote:
[ ... ]
>>> interrupt                          total       rate
>>> irq0: clk                       25130235         99
>>> irq1: atkbd0                           4          0
>>> irq6: fdc0                             1          0
>>> irq7: ppc0                             1          0
>>> irq8: rtc                         288300          1
>>> irq11: atapci1                    637852          2
>>> irq12: vr0 uhci0+                3890833         15
>>> irq13: npx0                            1          0
>>> irq14: ata0                           54          0
>>> Total                           29947281        119
>> Hmm.. There seems to be nothing wrong with these numbers. If you were
>> suffering from an interrupt storm I would have expected much larger numbers
>> in the "rate" column. The column "rate" shows the average number of
>> interrupts / second. In total your system handles about 120 interrupts /
>> second, so that shouldn't be a problem.
> 
> I have the feeling that rate is calculated like this:
> number of interrupts since boottime / seconds of uptime

OK.

> so, this won't help if his is having now an interrupt storm
> while most of the time the system is interrupted ~ 120 times
> per second

An interrupt storm on what, though?

The ~100 interrupts per second from the clock on IRQ0 are absolutely normal 
with HZ=100; the only other candidate seems to be the combination of vr0 & USB 
controller on IRQ 12.

If it's possible to disable the USB controller for a bit in the BIOS, it would 
be interesting to see whether that makes any difference.  You can probably 
kill the parallel port, too.

It's also interesting to note that IRQ 8, the RTC, seems to be mostly idle-- 
normally that fires at stathz=128...check "sysctl kern.clockrate".

> systat -vmstat $time shows interrupts per second regarding the last $time 
> seconds

Right, although the output from "vmstat 1" would survive being pasted into 
email better, I suspect.

-- 
-Chuck



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44A90EDA.6040707>