Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 09 Oct 1995 10:47:48 -0700
From:      Paul Traina <pst@cisco.com>
To:        Stu Phillips <stu@cisco.com>
Cc:        hackers@freebsd.org
Subject:   Re: FleeBSD and XNTPD 
Message-ID:  <199510091747.KAA02148@puli.cisco.com>
In-Reply-To: Your message of "Mon, 09 Oct 1995 09:28:38 PDT." <199510091628.JAA21052@feta.cisco.com> 

next in thread | previous in thread | raw e-mail | index | archive | help
That's utterly bizzare that async IO isn't working on the serial lines, I seem
to recall using it way back a million years ago.

What version of FreeBSD are you using?  I assume 2.0.5R, right?


> Date: Mon, 9 Oct 1995 09:28:38 -0700
> From: Stu Phillips <stu@cisco.com>
> To: pst@cisco.com
> Subject: FreeBSD and XNTPD
> Cc: dkatz@cisco.com
> 
> Spent a frustrating weekend trying to get one of the clock drivers for
> XNTPD working - of course, to hook up my GPS using NMEA as a first pass
> to determine just what level of accuracy it might produce.
> 
> Turns out that FleeBSD doesn't seem to support asynchronous IO on serial
> ports.  All the software built just fine but ioctl() or fcntl() calls on
> the serial file handle trying to set the process id to be signalled (ie
> using F_SETMODE or TIOCSPGRP) barfed with 'inappropriate ioctl for device).
> 
> After a lot of kernel code reading I confess to confusion.... the tty driver
> itself definetly supports async IO and the appropriate ioctl operations
> but the sio driver does not.  I thought that the tty driver was always
> layered on top but alas seems not to be the case.
> 
> So, bottom line, failed to get xntp to recognize the clock - its way too
> much work to change the logic to poll since the whole enchilada is built
> around async io. Futz!
> 
> On a more positive note, the G-45 can be interogated for the time of day
> using its own protocol - the answer has 4 uS resolution - date/time plus
> a 256Kbbps sync'd clock.  Of course with the serial link this degrades
> significantly but should be good enough for roughly 1-2 mS precision.
> 
> Sucks for a real GPS but for $300 LAN clock doesn't look too bad.
> 
> Stu
> 




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199510091747.KAA02148>