Re: HEADS UP: panic behaviour changed

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Sun, 1 Feb 2009 00:52:17 +0000 (UTC)
    From:        Eduardo Horvath <eeh%NetBSD.org@localhost>
    Message-ID:  <Pine.NEB.4.64.0902010028070.17004%mail.netbsd.org@localhost>

  | If you check the original email, X was the justification for making this 
  | change.

I know, as I said in another message, the reasons given for changing things
were not really what should have been given, and fixing the cvs log would be
a good idea (I suppose), but that has no impact on the validity of the
change itself.

Note that really two things were changed simultaneously here, one of those
I totally agree with, and I haven't seen anyone argue against - that is the
way that the default value was being set into the kernel.   The other change
was what the default should be - that one I personally don't think matters,
for reasons I have given before.

If you decide the default value matters, and is important, and is currently
wrong, then please change it the right way - add the option to the kernel
config file to set the default to 1, don't just undo the change ad made.

  | It's a bogus justification, but I don't think we can ignore it.

The former is now understood I believe.   That means that by continuing
to reference it you're just shooting down a straw man, over and over again,
which does your argument no good.

  | The only time the default setting of this value is important is between 
  | the time the power is turned on and the time the rc script is run that can 
  | change the sysctl value.  As soon as that happens it is no longer defalt 
  | behavior but whatever the sysadmin managing the system desires.

Yes.   That is what I thought I said too.

The rest of your arguments are reasonable, if not perhaps totally
convincing to me for having the default value set back to 1.   Once
again, I'm not sure it matters very much, as I don't think what happens
when the kernel panics is something we should spend huge amounts of
effort on - far better to make it so the kernel doesn't panic...

lacombar%gmail.com@localhost said:
 | This is a real PITA if you're doing early code debugging and don't reach
 | multi-user. It must be possible to tune this behaviour at compile time. 

It definitely is - the only question is what the distributed default should
be (and just perhaps, how implementing it should be done - as the way it was
done before this change was just plain absurd.)

kre