Re: HEADS UP: panic behaviour changed

[Eduardo Horvath <eeh%NetBSD.org@localhost>]

On Sat, 31 Jan 2009, Jukka Ruohonen wrote:

> On 01.02.2009, Robert Elz wrote:
> > I have no particular opinion on which default is better, nor do I really
> > care, I build my own kernels anyway and set the options the way I like them.
> > I would assume that most people here can (and should) do the same.
> > 
> > So, which is the default as distributed really applies (or should apply)
> > to those regular users who don't build their own kernels, and just keep
> > running GENERIC (personally, I think that should amount to no-one at all,
> > so I don't think this is a very important issue.)
> 
> I must ask: what is exactly again wrong with GENERIC?
> 
> It seems kind of counterproductive to argue that all "those regular users"
> should be tinkering with their kernels -- all "those regular users" who run
> CURRENT [sic], the main development branch that presumably benefits from the
> unified baseline.

Let's consider this logically.

1) Not everyone runs X.  Most servers use a serial console.

2) Even if you run X, some platforms have kernel drivers so they don't 
have problems with being unable to run a useful DDB even while X is 
running.

3) There is a period of time between loading the kernel and when the rc 
scripts run where you can't tweak the ddb_onpanic value.

4) If the machine panics early, say during device configuration due to 
broken hardware, you don't really want it to attempt to reboot, since that 
will result in an infinite reboot loop.

Given all this, the default should be to have the kernel drop to DDB when 
it panics.

Given this, you could have a simple option to set the value of ddb_onpanic 
either in the rc scripts or X initialization scripts.  But I really don't 
think the default behavior should be changed just as a minor convenience 
to some users.

Eduardo