Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

RTC's and system time

8 views
Skip to first unread message

Warren Brown

unread,
Sep 20, 1994, 4:30:34 PM9/20/94
to
In <4...@pepkfb.pep-kaufbeuren.de> r...@pepkfb.pep-kaufbeuren.de (Ruediger Nahm-Elstner) writes:

>Hello Folks,

>i have some small problem with synchronisation of an RTC and the system time.
>Imagine you have a very accurate RTC, but the system tick has some
>inacceptable deviation (10 minutes/day). Or you have some time synchronisation
>running between some hosts via TCP/IP.
>How is it possible to correct the system time ?

>If i run some task in the background that will periodically call "setime -s" -
>what will happen to alarms if we skip one second ?

This would be ok. If you set the time "past" a process/alarm wakeup time,
it will be awoken.

>What will happen to alarms that would occur just when we increment D_Ticks by
>2 instead of 1 (if the clock driver makes a correction) ?

Attempting to manipulate system globals will not achieve what you desire.

>RN

Hope this helps.

Warren

Ruediger Nahm-Elstner

unread,
Sep 20, 1994, 1:15:00 PM9/20/94
to
Hello Folks,

i have some small problem with synchronisation of an RTC and the system time.
Imagine you have a very accurate RTC, but the system tick has some
inacceptable deviation (10 minutes/day). Or you have some time synchronisation
running between some hosts via TCP/IP.
How is it possible to correct the system time ?

If i run some task in the background that will periodically call "setime -s" -
what will happen to alarms if we skip one second ?

What will happen to alarms that would occur just when we increment D_Ticks by
2 instead of 1 (if the clock driver makes a correction) ?


RN

+-----------------------------------------------------------------+
| Ruediger Nahm-Elstner Technical Support Group |
| PEP Modular Computers Phone: 0049/8341/4302/0 |
| Apfeltranger Strasse 16 Fax: 0049/8341/4302/39 |
| D-87600 Kaufbeuren Telex: 541233 |
| Germany |
| E-MAIL: r...@pepkfb.uucp, r...@pepkfb.pep-kaufbeuren.de |
+-----------------------------------------------------------------+

Ruediger Nahm-Elstner

unread,
Sep 21, 1994, 2:33:31 AM9/21/94
to
war...@microware.com (Warren Brown) writes:

>In <4...@pepkfb.pep-kaufbeuren.de> r...@pepkfb.pep-kaufbeuren.de (Ruediger Nahm-Elstner) writes:

>>Hello Folks,

>>i have some small problem with synchronisation of an RTC and the system time.
>>Imagine you have a very accurate RTC, but the system tick has some
>>inacceptable deviation (10 minutes/day). Or you have some time synchronisation
>>running between some hosts via TCP/IP.
>>How is it possible to correct the system time ?

>>If i run some task in the background that will periodically call "setime -s" -
>>what will happen to alarms if we skip one second ?

>This would be ok. If you set the time "past" a process/alarm wakeup time,
>it will be awoken.

What will happen to cyclic alarms ?
Will they drift away with the correction ?

J. Jansen

unread,
Sep 23, 1994, 3:22:31 AM9/23/94
to
r...@pepkfb.pep-kaufbeuren.de (Ruediger Nahm-Elstner) writes:

>Hello Folks,


>RN

EKF uses radio controlled clock boards in their VMEbus systems and has
a small driver that periodically synchronizes the system clock to the
DCF77 accurate time. I think this driver/descriptor could be modified
in order to adapt to your needs. If interested, write to the author

Uwe Petersen - u...@ekf.werries.de

--------------------- EKF Elektronik Messtechnik GmbH ---------------------
Joachim Jansen e-mail: j...@ekf.werries.de
Philipp Reis Str. 4 phone: ++49 (0)2381-6890-0
D-59065 Hamm (Germany) fax: ++49 (0)2381-6890-90
------------------- Advanced VMEbus Hardware + Software -------------------


--
---------------------------------------------------------------------------
j...@ekf.werries.de
---------------------------------------------------------------------------

0 new messages