On Wednesday, September 13, 2006 at 8:06:40 +0200, Petter Reinholdtsen wrote:
> I do realise that hwclock have its purpose if the hardware clock is > using local time, and not UTC, but am not equally convinced that it is > useful if it is using UTC. What about drift correction? The RTC chip of my machine here actually has 2m09s of advance. If I reboot now, the kernel will just copy this wrong raw time from the RTC to its own system clock. While with my carefully calibrated /etc/adjtime correction factor, hwclock --hctosys will provide the correct time, to the second: | # date | Fri Sep 15 09:44:51 CEST 2006 | | # cat /proc/rtc | grep time | rtc_time : 09:47:00 | | # adjtimex --compare=1 | --- current --- -- suggested -- | cmos time system-cmos 2nd diff tick freq tick freq | 1158306291 0.149082 0.149082 10000 -5974812 Note "cmos time" and "system-cmos" offset: That's not raw time, but time after drift correction. In other words, the drift correction reduced my RTC error from +2 minutes to -150 milliseconds. Wow! No sorry: Your wish to disable hwclock by default for everyone seems quite a bad idea. You could disable it for yourself, but even this is probably a bad idea I would not advice you. Alain. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]