Re: 2.6.1: process start times by procps

[Petri Kaukasoina <kaukasoi@elektroni.ee.tut.fi>]

On Thu, Jan 29, 2004 at 02:51:48PM -0800, Andrew Morton wrote:
> Petri Kaukasoina <kaukasoi@elektroni.ee.tut.fi> wrote:
> >
> > On Thu, Jan 29, 2004 at 11:48:49AM -0800, john stultz wrote:
> > > 
> > > Does George Anzinger's patch work as well?
> > 
> > I must have missed that... Any references?
> 
> This one.
> 
> diff -puN fs/proc/proc_misc.c~proc-stat-btime-fix-2 fs/proc/proc_misc.c

Yes, it works fine: btime stays at the correct value. Just the
ps-output-in-future-problem left.

Linux-2.6.1+patch without ntpd:
 passed time according to 'date':               11430.68524 s
 timer interrupts:                              11432383
 interrupts per second:                         1000.149
 error compared to 1000 interrupts per second:  +149 ppm
 ps times go towards future about 2.0 s:  about 170 ppm
 btime stays constant!
 clock too fast compared to ntp time by:        +0.260461 s
 so 'date' is too fast compared to UT by:       +22.8 ppm

Linux-2.6.1+patch with ntpd:
 passed time according to 'date':               14792.60765 s
 timer interrupts:                              14795154
 interrupts per second:                         1000.172
 error compared to 1000 interrupts per second:  +172 ppm
 ps times go towards future about 2.7 s:  about 180 ppm
 btime stays constant!
 (ntp.drift: -21.943)

-Petri