From owner-freebsd-stable  Sat Jun  5  5:56:53 1999
Delivered-To: freebsd-stable@freebsd.org
Received: from storm.twcol.com (storm.twcol.com [204.210.251.99])
	by hub.freebsd.org (Postfix) with ESMTP id 1629614D68
	for <freebsd-stable@freebsd.org>; Sat,  5 Jun 1999 05:56:50 -0700 (PDT)
	(envelope-from dzerkel@columbus.rr.com)
Received: from columbus.rr.com (dhcp219142.columbus.rr.com [208.169.219.142])
	by storm.twcol.com (8.9.1/8.9.1) with ESMTP id JAA26117
	for <freebsd-stable@freebsd.org>; Sat, 5 Jun 1999 09:02:02 -0400 (EDT)
Message-ID: <37591E90.2FD85193@columbus.rr.com>
Date: Sat, 05 Jun 1999 08:56:48 -0400
From: "Danny J. Zerkel" <dzerkel@columbus.rr.com>
Organization: Zerkular Software
X-Mailer: Mozilla 4.6 [en] (X11; I; FreeBSD 3.2-STABLE i386)
X-Accept-Language: Russian, ru, en
MIME-Version: 1.0
To: freebsd-stable@freebsd.org
Subject: Re: kern/10411: top, vmstat, iostat show 0% cpu idle & usage on SMP 
 system
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-stable@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

Joe,

Do you have apm configured in your kernel?  If so, try:

device          apm0    at isa? flags 0x20

So far, this seems to correct it.

When I started digging around, I found that the stat clock had been
disabled.
The only thing that can disable it is apm, but the test looks backwards
to me:

*** /usr/src/sys/i386/isa/clock.c:
line 946:
cpu_initclocks()
{
        int diag;
#ifdef APIC_IO
        int apic_8254_trial;
#endif /* APIC_IO */
        
        if (statclock_disable) {  
                /*
                 * The stat interrupt mask is different without the
                 * statistics clock.  Also, don't set the interrupt
                 * flag which would normally cause the RTC to generate
                 * interrupts.
                 */
                stat_imask = HWI_MASK | SWI_MASK;
                rtc_statusb = RTCSB_24HR;
        } else {
                /* Setting stathz to nonzero early helps avoid races. */
                stathz = RTC_NOPROFRATE;
                profhz = RTC_PROFRATE;
        }

The problem seems to be that the statclock gets disabled if
statclock_disable
ISN'T set.  My stathz was set to RTC_NOPROFRATE, which disables
collection
of stats.

> > After installing a MP kernel on my system I have noticed that the various
> > system utilities that display cpu usage all report 0.0% for everything.
> > The UP kernel does not have this problem and the MP kernel will work ok
> > briefly after rebooting, until I apply a load (such as starting X or
> > building a kernel).
> 
> I am seeing this problem as well.
> 
> ASUS P2B-DS, 2 x PII-400/512, disks moved from an HP Vectra XU/200 that
> worked just fine.  I was running a mildly modified 3.0R on the Vectra
> which reported statistics correctly.  Moving these to the P2B-DS broke,
> for reasons unknown, since I have other P2B-DS's on 3.0R which report
> statistics fine.
> 
> I then upgraded the box to 3.1R since I thought maybe my mods were causing
> the problem, but it is still broken.
> 
> Oddly, it appears to work fine in single-user, although I didn't play with
> it for more than a few minutes.
> 
> More oddly, load average appears to report properly.
> 
> If any bug whacker would like access to the box in question, it can be 
> arranged easily enough (its a games machine).  I'd sure like to see a fix
> for this.
> 
> ... Joe

-- Danny J. Zerkel
dzerkel@columbus.rr.com
"Sursum ad Absurdum"


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message