Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!sdd.hp.com!mips!daver!bungi.com!news
From: culberts@hplwbc.hpl.hp.com (Bruce Culbertson)
Newsgroups: comp.sys.nsc.32k
Subject: Re:  /dev/rtc
Message-ID: <9009131644.AA22494@hplwbc.hpl.hp.com>
Date: 13 Sep 90 16:44:43 GMT
Sender: news@daver.bungi.com
Lines: 37
Approved: news@daver.bungi.com

> Karl Swartz writes:
> 
> Johan Myreen <jem@cs.hut.fi> writes:
> 
> > I myself have written something that also might interest you out
> > there: a driver for the real time clock. You talk to it through the
> > special file /dev/rtc
> 
> That would be very nice indeed.  If I can ever manage to get the
> sources I'll enhance the date command to use this.  What will your
> driver do if there's no clock installed?  It would be nice to have
> some automagic way of detecting that so you wouldn't have to diddle
> with a kernel option to use the thing.

What about putting something like this in /etc/rc:

	# set system date
	if (test -c /dev/rtc)
	then
	  # read date from hardware real time clock
	  date_formatter < /dev/rtc | date -q
	else
	  # prompt user for date
	  date -q
	fi

Date_formatter (to be written, but trivial) converts the eight bytes
from the real time clock chip into the mmddyyhhmmss format which
the date command likes.  The date command would be unchanged.  Also,
we would remove the RTC code which I supplied from the kernel.
If you have an RTC, you mknod /etc/rtc; if not, you do not mknod
/dev/rtc.  The kernel would have a new device driver but otherwise
would not know about the RTC.

I think this is clean -- what do you think?

Bruce