2.6.10 and time drift

2.6.10 and time drift

[Alexander Prokoshev]

Hello,

  after installation of 2.6.10 kernel I've noticed time drift, which
(according to ntpdc's dmpeer command) is about 10-15 seconds per hour.
Downgrade to 2.6.9 solves this problem. I can send any additional
information which may be helpful.

-- 
WBR, Alexander.

Re: 2.6.10 and time drift

[Valdis.Kletnieks]

[-- Attachment #1: Type: text/plain, Size: 1106 bytes --]

On Mon, 27 Dec 2004 14:28:25 +0300, Alexander Prokoshev said:

>   after installation of 2.6.10 kernel I've noticed time drift, which
> (according to ntpdc's dmpeer command) is about 10-15 seconds per hour.
> Downgrade to 2.6.9 solves this problem. I can send any additional
> information which may be helpful.

For starters, the output of 'uname -a', the architecture/hardware you're
running on, and if your dmesg has any hints about which time source it's
using. On my Dell laptop, I see:

Dec 26 19:11:22 turing-police kernel: PID hash table entries: 1024 (order: 10, 16384 bytes)
Dec 26 19:11:22 turing-police kernel: Detected 1595.344 MHz processor.
Dec 26 19:11:22 turing-police kernel: Using pmtmr for high-res timesource
Dec 26 19:11:22 turing-police kernel: Console: colour dummy device 80x25
Dec 26 19:11:22 turing-police kernel: Dentry cache hash table entries: 65536 (order: 6, 262144 bytes)
Dec 26 19:11:22 turing-police kernel: Inode-cache hash table entries: 32768 (order: 5, 131072 bytes)

Most x86 systems should have a similar note of the timesource right
around that point of the dmesg.

[-- Attachment #2: Type: application/pgp-signature, Size: 226 bytes --]

Re: 2.6.10 and time drift

[Gene Heskett]

On Monday 27 December 2004 06:28, Alexander Prokoshev wrote:
>Hello,
>
>  after installation of 2.6.10 kernel I've noticed time drift, which
>(according to ntpdc's dmpeer command) is about 10-15 seconds per
> hour. Downgrade to 2.6.9 solves this problem. I can send any
> additional information which may be helpful.

Until this gets fixed, you might try resetting the kernels tick period 
with the 'tickadj' command. I'm doing fairly well with a setting of 
9926 here, default is 10,000.  I stuck it in my rc.local file.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.30% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com attorneys please note, additions to this message
by Gene Heskett are:
Copyright 2004 by Maurice Eugene Heskett, all rights reserved.

Re: 2.6.10 and time drift

[Miguelanxo Otero Salgueiro]

I just didn't notice the time drift during normal conditions, but it 
gets REALLY bad when I suspend my laptop.
As a reference, I suspended it aproximately at 01:00 and resumed it at 
15:25 but the clock says 22:30. It does not occur in 2.6.9. Dmesg output 
relative to time dmesg|grep time:

$ dmesg|grep time
Using pmtmr for high-res timesource
PCI: Setting latency timer of device 0000:00:1f.5 to 64
PCI: Setting latency timer of device 0000:00:1d.0 to 64
PCI: Setting latency timer of device 0000:00:1d.7 to 64
PCI: Setting latency timer of device 0000:00:1d.0 to 64
PCI: Setting latency timer of device 0000:00:1f.5 to 64