hmmm..... I don't know.... :) devstat_compute_etime(struct bintime *cur_time, struct bintime *prev_time) is responsable for calculating the difference between cur_time and prev_time, but knows nothing about the correctness of the result. someone could use the function for calculating negative differences or something (you can never know.... ;-) so i think, it would be better in vmstat.c (usr.bin/systat), as there is already a check, if busytime = devstat_compute_etime(...) is bigger than 100%, so <0 should be checked right there too, i think look at the patche I posted to -hackers.... Friedemann On Wed, 9 Apr 2003, Poul-Henning Kamp wrote: > In message <20030409023754.P92580_at_alpha.siliconlandmark.com>, Andre Guibert de > Bruet writes: > > > >I just spotted this today... > > > >Disks ad4 ad6 acd0 ofod intrn > >KB/t 0.00 0.00 0.00 %slo-z 114272 buf > >tps 0 0 0 tfree 13 dirtybuf > >MB/s 0.00 0.00 0.00 70224 desiredvnodes > >% busy -0. -0. 0 1046 numvnodes > > > Yeah, I saw that too. It's a rounding error. If you dive into it you > will see that the number is incredibly close to zero. > > We should probably catch that in libdevstat. > > -- > Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 > phk_at_FreeBSD.ORG | TCP/IP since RFC 956 > FreeBSD committer | BSD since 4.3-tahoe > Never attribute to malice what can adequately be explained by incompetence. >Received on Wed Apr 09 2003 - 06:42:12 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:37:03 UTC