[time-nuts] FreeBSD version

[Josh Smith]

Thanks for the great information guys.

--
Josh Smith
KD8HRX
email/jabber:  juicewvu at gmail.com
phone:  304.237.9369(c)





On Thu, Oct 1, 2009 at 8:29 AM, Magnus Danielson
<magnus at rubidium.dyndns.org> wrote:
> Adam,
>
> Adam Feigin wrote:
>>>
>>> I guess this might be a little ot for this thread but maybe not.  I'm
>>> a newbie to all of this stuff but judging from the archives freebsd
>>> seems to be the OS of choice for these highly accurate NTP servers.
>>> Why is the case?  Is there something specific to it that makes it more
>>> suitable for this application than say some of the other BSD's or
>>> Linux?
>>>
>>
>> Well, FreeBSD had one of the earliest PPSAPI implementations, so its
>> well tested, and it works extremely well.
>>
>> Linux still does not have a PPSAPI implementation in the "standard"
>> kernel. As of 2.6.31, its sort of kinda of halfway there, but not
>> completely, and I've never been able to get it to work. I'm sure one
>> can, if ones time in not very valuable.
>
> Actually, one problem is that glibc is (was?) publishing an older variant of
> the PPSAPI than the kernel actually supports. This is due to an unfortunate
> decission to let kernel-API being made available directly to be used by
> applications be declared by glibc rather than following the kernel. The NTPD
> finds the glibc variant of the header and uses the older interface version
> regardless of the fact that the kernel has advanced since the glibc took its
> snapshot out of the early 2.x series.
>
> The correct way would be either to let glibc discover the actual kernel
> abilites or just provide the kernel-call wrappers and let the kernel provide
> the actual header file.
>
> Usually the glibc approach is correct, but for PPSAPI it is not. Updating
> glibc to the current version heals puts a japanese paper wall in front of
> the problem, but you haven't properly healed the wound.
>
>> FreeBSD works right out of the box (okay, you'll need to add the PPS
>> option to the kernel configuration).
>
> To a higher degree things are integrated there in this respect.
>
> Cheers,
> Magnus
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>