what's it doing in "sys" time?

[Hubert Feyrer <hubert%feyrer.de@localhost>]

I have a system (4.99.72/i386 GENERIC in Parallels, 1 CPU) that's doing a 
build, with sources mounted via NFS. I'm seeing pretty high times for the 
CPU spending in "system" (as opposed to user time), see "sy" column:

vmnetbsd% vmstat -w 1
 procs    memory      page                       disks   faults      cpu
 r b w    avm    fre  flt  re  pi   po   fr   sr f0 c0   in   sy  cs us sy id
 0 0 0 158060  12248  233   0   0    0    2    2  0  0  201  233 113  1 13 86
 1 0 0 154880  15460 1232   0   0    0    0    0  0  0 1091  807 1067 4 57 39
 1 0 0 157032  13292  669   0   0    0    0    0  0  0 1281  591 1233 3 63 34
 1 0 0 154684  15672 1144   0   0    0    0    0  0  0 1230  542 1193 2 71 27
 1 0 0 156928  13400  987   0   0    0    0    0  0  0 1149  741 1036 3 69 28
 1 0 0 154956  15388  983   0   0    0    0    0  0  0 1263  542 1262 5 70 25
 0 0 0 156312  14012  997   0   0    0    0    0  0  0 1152  793 1136 3 52 45
 0 0 0 158284  12036  409   0   0    0    0    0  0  0 1414  524 1292 0 59 41
                                                                        ^^

Is the time of ~70% in the kernel expected? My gut feeling tells me it's 
too high. And my question is now: how do I find out what it's doing in the 
system?

top(1) doesn't show the [system] process a lot, and according to ps(1), 
wchan is "schedule" a lot of the time.

What amount of "sys" times do you see during your builds?


 - Hubert