Date: Fri, 09 Nov 2001 10:03:11 -0700 From: Warner Losh <imp@harmony.village.org> To: Poul-Henning Kamp <phk@critter.freebsd.dk> Cc: Sansonetti Laurent <lorenzo@linuxbe.org>, freebsd-hackers@freebsd.org Subject: Re: Measuring interrupt latency Message-ID: <200111091703.fA9H3B753981@harmony.village.org> In-Reply-To: Your message of "Fri, 09 Nov 2001 17:42:58 %2B0100." <574.1005324178@critter.freebsd.dk> References: <574.1005324178@critter.freebsd.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
In message <574.1005324178@critter.freebsd.dk> Poul-Henning Kamp writes: : >We found at Timing Solutions when we were trying to measure interrupt : >latency that the system time (getnanotime()) gave us measurements with : >a larger variance than our expensive scopes that does statistical : >gathering. : : Uhm, you should have used nanotime(), not getnanotime(). : : getnanotime() returns a timestamp in nanoseconds of the last : stored timestamp which may be up to 1/hz seconds old. Hmmmm. Actually looking at the code, we did use nanotime()... But we do have a few drivers that are using getnanotime() incorrectly... The larger variance was something like a factor of 2 larger than the one that we'd measured on the scopes. I don't have the data online, or I'd be more precise. Warner To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200111091703.fA9H3B753981>