APMD-List:
Archives

  From: Jont Allen <jba@research.att.com>
  To  : Avery Pennarun <apenwarr@worldvisions.ca>
  Date: Fri, 03 Sep 1999 23:00:07 -0400

Re: Another CPU slows to a crawl after resume from suspend

Avery Pennarun wrote:
> 
> On Thu, Sep 02, 1999 at 12:04:32AM -0700, John Pezaris wrote:
> 
> > Akin to previous messages sent some time ago to this list, my computer
> > becomes incredibly slow after a suspend-resume cycle.  I suspect that
> > what's happening is the CPU speed is being set to slow while suspended, and
> > that value never gets restored to the original tear-em-up frequency.
> 
> I've often wondered whether such problems are because the CPU is running too
> slowly, or because the system timer was accidentally reset, or because the
> cache memory got disabled.
> 
> Have you tried running the "bogomips" userspace program to re-check your
> system speed?  Also, try running 'date' a lot of times and see if seconds
> are counting at the right rate.
> 
> I don't know of any quick cpu-speed-reset programs.  Maybe "apm --standby"
> will kick things and make them better?
> 
> Good luck.
> 
> Avery

Dear List,

There is a `well known' documented kernel bug (according to David Hinds) where
the CPU can boot with a random clock speed. There is no known fix
at present. In some cases, the CPU can start at 8 mHz! 

My IBM 560X (notebook) randomly boots at 137 bogomips, while the normal speed
is close to 463. (i.e., cat /proc/cpuinfo|grep bogo)

This causes the IP address of my NIC card to be wrong. So the only
way I know, is that when I plugin the PCMCIA NIC, it doesn't get the
correct MAC address!

The only way to recover from this is to reboot. I have had it happen
twice, or even three times, in a row. It is a real `drag' as you can
imagine. Most of the time this is not a problem. It occurs maybe
once every couple of months. Maybe even less!

Ask D. Hinds about it, if you want more information.

> David,
>
> Sometimes when I boot, as you pointed out, the bogomips is wrong.
> This causes problems.
>       Was this problem ever fixed? If so, approx what/kernel rev.
>       Where can I read about the problem (on the web?)

It hasn't been fixed: it is a basic [Linux] design flaw.  The CPU clock 
speed on some systems can change from moment to moment, so the BogoMIPS
concept (of basing delay loops on the CPU clock) is broken.

-- Dave
                                                                             

Jont
-- 
Jont B. Allen
AT&T Labs-Research, Shannon Laboratory
180 Park Ave., Room E161, Florham Park NJ, 07932-0971
973/360-8545voice, x7111fax, http://www.research.att.com/~jba
 -My favorite URL: www-history.mcs.st-and.ac.uk/history
 -Imagination is more important than knowledge.  --Albert Einstein
 -Microsoft: The company with negative imagination.
 -In a world without fences, who needs Gates? -- www.linuxjournal.com