Re: GSoC2007: cnst-sensors.2007-09-13.patch

> Date: Thu, 20 Sep 2007 16:12:13 -0700 (PDT)
> From: Doug Barton <dougb_at_FreeBSD.org>
> Sender: owner-freebsd-current_at_freebsd.org
> 
> On Wed, 19 Sep 2007, Constantine A. Murenin wrote:
> 
> > Thanks for testing!
> 
> Glad to help. In case it's interesting, I was doing the xorg update with 
> portmaster last night and I got several "PROCHOT asserted" messages on my 
> console at different times. I'm assuming that's expected behavior, just 
> curious if it's something bad, as in when that happens it's time to turn 
> off the laptop? (I didn't seem them when the happened, they were there 
> when I got back to check on the compiling.)
> 
> >> Two small comments about the rc.d stuff. First, the empty _flags variable 
> >> in defaults/rc.conf should be commented out. Second, the rc.d 
> >
> > How so?  I don't see any other empty _flags variables in defaults/rc.conf 
> > being commented out.
> 
> Well you missed named_flags. :) But seriously, I didn't realize that 
> things had gotten quite so out of hand with that ... never mind then.
> 
> >> script needs the shutdown KEYWORD.
> >
> > Similarly, I don't see why this is needed -- it was not used by the scripts 
> > on which this script was based on
> 
> Which scripts? I realize that a distressingly large number of scripts that 
> start services don't have this keyword, but they should. I'll work on a 
> patch for that. At the same time, we don't want to add any new scripts 
> that make the same mistake.
> 
> > Reading through rc(8) doesn't seem to suggest that this keyword would 
> > actually be applicable here.
> 
> As far as I can tell, you're starting a daemon, which means that it should 
> be cleanly shut down when the system exits.

I've now read rcorder(8) and rc(8). I am very unclear what the
significance of the shutdown keyword is. I thought it was to cause
rc.shutdown to invoke a "stop" at shutdown time, but only 11 of the
scripts in /etc/rc.d include the shutdown keyword. They do include most
of the daemons which look likely to really need to be shut down
(e.g. random, mixer, nfsd, inetd, auditd), but it is missing from a
number of others.

I think 'shutdown' might belong in others, but, if a 'kill -9' does the
same thing as a 'stop' operation for a given daemon, it might be better
to not have 'shutdown' to avoid having a hung daemon delay the shutdown.

The reality is that it looks like it is usually done right and I don't
see any real reason for 'shutdown' in this case.

Feel free to call me a fool. I don't claim to 
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman_at_es.net			Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4  EADA 927D EBB3 987B 3751