Date: Tue, 04 Jul 2000 17:40:15 +1000 From: Stephen McKay <mckay@thehub.com.au> To: Andreas Persson <pap@garen.net> Cc: mckay@thehub.com.au, freebsd-stable@FreeBSD.ORG Subject: Re: microuptime() going backwards Message-ID: <200007040740.RAA16962@dungeon.home>
next in thread | raw e-mail | index | archive | help
Andreas Persson wrote: >On Tue, Jun 27, 2000 at 08:21:18PM -0700, Mike Smith wrote: >>I explained this to you at Usenix, actually. It has nothing to do with >>APM, it has to do with the selection of clocks available with/without APM >>compiled into the kernel - there is probably either a bug in the TSC >>hardware on this CPU, or (more likely) a bug in the timecounter code >>(since people see this on !APM systems already). > >Setting the sysctl kern.timecounter.method to 1 seems to have solved this >on one of my 4.0-RELEASE boxes. No microuptime() messages for almost 3 >weeks now. This didn't help me. I have an Athlon 700 on an EPOX 7KXA. I had apm compiled in but disabled. The microuptime messages made work impossible. I tried kern.timecounter.method=1 but the problem continued as long as kern.timecounter.hardware=i8254. When I removed apm totally, FreeBSD started using the CPU's TSC register (ie kern.timecounter.hardware=TSC) and my problems went away. Broken (chipset emulated) 8254 hardware? Or broken time code? I don't know yet. Stephen. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200007040740.RAA16962>