I've audited the code, added some debug and it looks algorithmically
correct. I think this is a problem with the granularity of time
accounting being different for real time versus the cpu times.
You can see the same issue independent of shping ... see NOTE cases
below.
For more serious commands that run longer than a few milliseconds, I
suspect this would be less of a problem.
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143743326
real 0m0.003s
user 0m0.000s
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143744588
real 0m0.003s
user 0m0.004s NOTE
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143751635
real 0m0.006s
user 0m0.000s
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143752872
real 0m0.003s
user 0m0.000s
sys 0m0.004s NOTE
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143758166
real 0m0.003s
user 0m0.000s
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143759937
real 0m0.004s
user 0m0.000s
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143760486
real 0m0.004s
user 0m0.000s
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143761346
real 0m0.004s
user 0m0.000s
sys 0m0.000s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143762237
real 0m0.004s
user 0m0.000s
sys 0m0.004s
kenj@bozo:~/src/pcp/src/pmdas/shping$ time pminfo -f kernel.all.pswitch
kernel.all.pswitch
value 143765283
real 0m0.003s
user 0m0.000s
sys 0m0.004s NOTE
On Wed, 2009-01-21 at 17:52 +1100, Nathan Scott wrote:
> Hi,
>
> Couple of as-yet-unresolved things coming out of LCA so far
> (that others have reported to me):
> - lex/yacc not being installed should cause configure failure.
> - Makepkgs doesn't work if /bin/sh is /bin/dash (something to
> do with sourcing the VERSION file).
>
> - one I found - shping config:
> null exit 0
> pmcd pminfo -f kernel.all.pswitch
>
> ... running this, I notice that the pmcd response time is often
> _less_ than the sum of user/kernel time ... what the?
>
> Several kmchart things too ... chopped off menu bar margin'ing
> and/or font issues again. I'll look into this some more if I
> get some time this week.
>
> cheers.
>
> --
> Nathan
>
> _______________________________________________
> pcp mailing list
> pcp@xxxxxxxxxxx
> http://oss.sgi.com/mailman/listinfo/pcp
|