Date: Sun, 6 Oct 2002 02:50:04 -0700 (PDT) From: Valentin Nechayev <netch@iv.nn.kiev.ua> To: freebsd-bugs@FreeBSD.org Subject: Re: i386/43491: microuptime () went backwards Message-ID: <200210060950.g969o46u009186@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
The following reply was made to PR i386/43491; it has been noted by GNATS. From: Valentin Nechayev <netch@iv.nn.kiev.ua> To: "Bjoern A.Zeeb" <bzeeb@zabbadoz.net> Cc: bug-followup@FreeBSD.ORG Subject: Re: i386/43491: microuptime () went backwards Date: Sun, 6 Oct 2002 12:42:45 +0300 Sat, Oct 05, 2002 at 15:10:12, bzeeb (Bjoern A.Zeeb) wrote about "Re: i386/43491: microuptime () went backwards": BAZ> pcib2: <VIA 82C598MVP (Apollo MVP3) PCI-PCI (AGP) bridge> at device 1.0 on pci0 [...] BAZ> Oct 5 23:28:23 e0-0 /kernel: microuptime() went backwards (100768.401070 -> 100768.378034) BAZ> Oct 5 23:28:23 e0-0 /kernel: microuptime() went backwards (100768.401070 -> 100768.379457) [...] BAZ> though read that you cannot really fix it there is one thing you might BAZ> want to do: BAZ> remove the time changes from each and every printf so syslog can BAZ> aggregate the messages and simply print BAZ> "last message repeated 490345 times". BAZ> if you need it to be dumped (though you cannot correct anything with BAZ> this information) add a flag for those people who want to see the BAZ> addition information. BAZ> This would at least reduce disc IO and save some bytes ;-) and perhaps BAZ> help to stop the system from misbehaving. You can do it for your particular system. For most systems, it is not such useful. Pity for you but I suggest you to change hardware. VIA Apollo MVP3 has too strange relations with time counting. I saw a system with it setting kern.timecounter.method=1 stopped timer totally; without it, "microuptime went backwards" appeared each minute and I had to run ntpdate every minute. Of course one can argue that total PC time counting is horribly brain-damaged, but it is common place and we should live with it. (Only PIIX4 seems to be correct. But 4.5 secs is too small period.) Change it to Intel chipset based motherboard and you will be satisfied. I don't see a reason to keep this PR open. You may want to open PR to fix total FreeBSD time counting subsystem or simply ask phk@ to describe it in details, but this shall be another PR. /netch To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200210060950.g969o46u009186>