Process start times moving in reverse on 2.6.8.1

Process start times moving in reverse on 2.6.8.1

[Phil Oester]

ISTR discussion on the mailing list about this problem,
and recently upgraded from 2.6.3 to 2.6.8.1 to hopefully
solve it, but alas the problem still exists.

Example:

# date ; ps -ef | grep ps | grep -v grep
Mon Oct  4 14:53:39 EDT 2004
root     29412 29351  0 14:51 pts/0    00:00:00 ps -ef

Notice the two minute difference between now and what the
process start time is.  Uptime on this box is 48 days, so
it is a gradual drift.

Any ideas on this?  Or has it been fixed since 2.6.8.1?

Phil

Re: Process start times moving in reverse on 2.6.8.1

[Andrew Morton]

Phil Oester <kernel@linuxace.com> wrote:
>
> ISTR discussion on the mailing list about this problem,
> and recently upgraded from 2.6.3 to 2.6.8.1 to hopefully
> solve it, but alas the problem still exists.
> 
> Example:
> 
> # date ; ps -ef | grep ps | grep -v grep
> Mon Oct  4 14:53:39 EDT 2004
> root     29412 29351  0 14:51 pts/0    00:00:00 ps -ef
> 
> Notice the two minute difference between now and what the
> process start time is.  Uptime on this box is 48 days, so
> it is a gradual drift.
> 
> Any ideas on this?  Or has it been fixed since 2.6.8.1?

It's allegedly fixed by
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.9-rc3/2.6.9-rc3-mm2/broken-out/fix-process-start-times.patch
but I've seen no confirmation of that.

Re: Process start times moving in reverse on 2.6.8.1

[Phil Oester]

On Mon, Oct 04, 2004 at 04:35:11PM -0700, Andrew Morton wrote:
> Phil Oester <kernel@linuxace.com> wrote:
> > Notice the two minute difference between now and what the
> > process start time is.  Uptime on this box is 48 days, so
> > it is a gradual drift.
> > 
> > Any ideas on this?  Or has it been fixed since 2.6.8.1?
> 
> It's allegedly fixed by
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.9-rc3/2.6.9-rc3-mm2/broken-out/fix-process-start-times.patch
> but I've seen no confirmation of that.

After running a patched 2.6.9-rc3-bk7 box for a day vs a freshly
rebooted 2.6.8.1, I can confirm that process start times seem
to coincide with current time, while the 2.6.8.1 box is already
out of sync.

I'd say this is a keeper, but can track it longer if you prefer.

Phil