[ntp:questions] PPS kit for 2.6 linux kernels?
Michael L. Semon
mlsemon33 at verizon.net
Fri Apr 13 12:42:11 UTC 2007
Hal Murray wrote:
>> No. Even the hardware of ten to fifteen years ago has sufficient
>> capacity to run ntpd. I run ntpd on Sun Ultra 10 machines with 440 MHz
>> CPU's (late 1990's vintage).
> Are you using PPS? Does your kernel have hardpps?
> My question was not whether ntpd would run on old machines, but
> why the hardpps code was needed in the kernel.
> I was guessing that old machines had performance issues.
> That seems like the best explaination even if it doesn't
> fit today's data.
Even not-so-old machines have performance issues. Someone here once
wrote that if you have good hardware, it's best to let ntpd take care of
the PPS because ntpd has superior algorithms to the NTP-style algorithms
in the kernel. However, my PC has either a clock or voltage regulation
that's absolute garbage, and hardpps would be nice. My current viable
options are these:
1) Use my current Linux 22.214.171.124 + LinuxPPS and sit and watch as ntpd
darts +/- 120 us over the course of a few hours, even when idle.
2) Reboot to FreeBSD and let the kernel, hardpps, and the NTP kernel
code make more radical short-term adjustments, and ntpd will dart +/- 10
us over a few hours.
3) Recompile glibc with support with older kernels without breaking my
system, then fetch Linux 126.96.36.199, fetch Ulrich's last PPSkit alpha, and
hope it all works.
4) Run ntpd on an ancient Pentium 75, where the clock on it is good but
the hardware is noisy, have the added network delay, have my choice of
FreeBSD or Linux 2.4 + PPSkit, and have everything working as designed.
There's a certain amount of frustration on my part. There are actually
two issues here, it seems, and it goes to Ulrich's point: a) hardpps
support, and b) what to do with it, in the way of more NTP algorithms in
the kernel. But when I Google for Linux PPS support and hit an archive
from a kernel mailing list or something, it's usually of someone wanting
to do something with PPS, followed by two or three answers on why it
can't be done. Is there just a resistance about it all?
It's a testament to look in the PPSkit archive, to see all of those file
dates on all of the various verions, and seeing it all stop a few months
ago. If you think of what Linux has gone through in that time, one
can't help but admire the author's dedication. I'm sure that there
would be more PPSkit versions if the conditions were right.
More information about the questions