[As-users] v1.8.11 + toshiba 320CT = 4second freeze every 1020

Tim Moore (afterstep@nsr500.net)
Mon, 28 Jun 2004 13:16:39 -0700

I know this release is not current but would like to fix it since my laptop 
has natural limitations (64MB max, Mobile Pentium MMX @266MHz).

Somewhere in 1.8.11 is code that causes a precise, repeatable 3.99 second 
whole system freeze very 1020 seconds no matter how busy or idle, or task 
mix.  "whole system" means the entire laptop, including RTC, is frozen. 
After the freeze, everything continues as if nothing happened except that 
the system time has jumped ahead by 3.990 seconds.  This does not happen on 
my Asus and Abit based desktops which all have the same environment.

There have never been any problems like dropped packets or data loss, only 
  the RTC gaining 4.seconds.  I ran ntpdate to my LAN ntp server in a 1 
second loop to illustrate:

> 28 Jun 11:56:52 ntpdate[2780]: adjust time server offset 0.002125 sec
> 28 Jun 11:56:53 ntpdate[2782]: step time server offset 3.992613 sec
> ...
> 28 Jun 12:13:52 ntpdate[4599]: step time server offset 4.003295 sec
> 28 Jun 12:13:53 ntpdate[4601]: step time server offset 7.993790 sec
> ...
> 28 Jun 12:30:52 ntpdate[6409]: step time server offset 8.004875 sec
> 28 Jun 12:30:53 ntpdate[6411]: step time server offset 11.995373 sec

When I switch to either gnome or kde, no time jumps.  100% repeatable.

I am happy to track and fix.  I've looked at lib/{sleep,tick}.c and can see 
no obvious 1020s or 4s patterns.  Any suggestions as to where else to look?

As-users mailing list