Re: dropping kerneld...

Luca Berra (bluca@rettore.rett.polimi.it)
Tue, 30 Apr 1996 16:56:58 +0200 (METDST)

Kevin Buettner said:
>=20
> Michael Meskes writes:
> > Kevin Buettner writes:
>=20
> Linus decided to remove it when someone (Simon Shapiro?) reported tha=
t
> his kernel went into a "coma". Here is the relevant post:
>=20
>=20
> : Date: Thu, 25 Apr 1996 13:48:34 +0300 (EET DST)
> : From: Linus Torvalds <torvalds@cs.helsinki.fi>
> : Reply-To: Linus Torvalds <torvalds@cs.helsinki.fi>
> : Subject: Re: 1.3.94 Ooops For Sale...
> : To: Simon Shapiro <Shimon@i-connect.net>, bj0rn@blox.se
> : cc: linux-kernel@vger.rutgers.edu
> : In-Reply-To: <XFMail.960424091114.Shimon@i-Connect.Net>
> : Message-ID: <ML-2.2.830429314.8565.torvalds@linux.cs.helsinki.fi>
> : MIME-Version: 1.0
> : Content-Type: TEXT/PLAIN; CHARSET=3DUS-ASCII
> : Content-Transfer-Encoding: 8BIT
> : Sender: owner-linux-kernel@vger.rutgers.edu
> : Precedence: bulk
> : Status: RO
> :=20
> : > Few more tidbits;
> : >=20
> : > master.debian.org went into a coma less than 12 hours after booti=
ng 1.3.94.
> : > What happened? No clue. the only symptoms (differentiating coma=
from
> : > death):=20
> : >=20
> : > 1. ``Ouch, kerneld wanted to sleep in interrupt'' every second o=
r two
> : > 2. Login freezes after taking the login name on the console.
> : > 3. Existing telnet session totally frozen.
> : > 4. ``Cannot load interpreter'' in response to login - sometimes.
> : >=20
> : > This was preceeded by the same unholy stream of prophanity as des=
cribed in
> : > my previous posting under this title (``route to %p was born dead=
'').
> :=20
> : Ok, "kerneld" goes disabled in the next kernel, and won't be resusc=
itated until
> : after 2.0 is out unless somebody really starts to look into this th=
ing. I'm not
> : using kerneld personally, and for a few reasons I don't think I _wi=
ll_ be using
> : it in the near future, so I won't be fixing this.=20
> :=20
> : Bj=F6rn?
> :=20
> : Linus
>=20
> Kevin
>=20
Sorry, i hadn't time to read linux-kernel for a while, but i had the
same problem. AND I FOUND THE REASON.
It happened when I was so stupid to enable ARPD in the kernel
and had NOT started the arpd daemon.
I believe kerneld gets some messages which were really meant for arpd a=
nd
it freezes, i had no time to look further into this, since i am overloa=
ded.

I believe a way to fix this kind of problems could be:
- All user space programs using the kerneld interface register with
kerneld at startup.
- All messages from the kernel are tagged with a 'magic number'
specifiyng who is it for.
- kerneld gets the message, and if it finds someone able to handle
it passes it on, else it tells the kernel that it failed to
process it, and operation continues.

comments? Bj0rn? Linus?

I repeat it is not a kerneld problem,
so Linus please leave in kerneld support.

--=20
#include <disclaimer.h>
+----------------------------------------------------------------------=
------+
| Luca Berra -- S.I.W.A. Politecnico di Milano =
|
| Phone: +39-2-23992413 - e-mail: bluca@www.polimi.it =
|
| PGP public key available via finger =
|
+----------------------------------------------------------------------=
------+
"She was infatuated with their male prostitutes, whose members were =
like
those of donkeys and whose seed came in floods like that of stallio=
ns."
Ezekiel 23:20