Return-Path: owner-linux-activists@joker.cs.hut.fi
Return-Path: <owner-linux-activists@joker.cs.hut.fi>
Received: from kruuna.helsinki.fi by hydra.Helsinki.FI (4.1/SMI-4.1/39)
	id AA13940; Sun, 2 May 93 00:16:03 +0300
Received: from joker.cs.hut.fi by kruuna.helsinki.fi with SMTP id AA09095
  (5.65c8/IDA-1.4.4 for <wirzeniu@cc.helsinki.fi>); Sun, 2 May 1993 00:16:02 +0300
Received: from joker.cs.hut.fi by niksula.hut.fi id <61658-4>; Sun, 2 May 1993 00:15:24 +0300
From: "Linux Activists" <linux-activists@joker.cs.hut.fi>
To: "Linux-Activists" <linux-activists@joker.cs.hut.fi>
Reply-To: "Linux-Activists" <linux-activists@joker.cs.hut.fi>
X-Note1: Remember to put 'X-Mn-Key: KERNEL' to your mail body or header
Subject: Linux-Activists - KERNEL Channel digest. 93-4-1-18:28
X-Mn-Key: KERNEL
Sender: owner-linux-activists@joker.cs.hut.fi
Message-Id: <93May2.001524eet_dst.61658-4@niksula.hut.fi>
Date: Sun, 2 May 1993 00:15:23 +0300
X-Charset: LATIN1
X-Char-Esc: 29
Status: RO
X-Status: 


Topics:
	 BUG: cat /proc/dev/unix


----------------------------------------------------------------------

From: Stephen Tweedie <sct@dcs.ed.ac.uk>
Subject: BUG: cat /proc/dev/unix
Date: Sat, 1 May 1993 21:10:46 +0300


Brian Edmonds <edmonds@cs.ubc.ca> writes:

> The following kernel burp was (repeatably) observed while catting 
> /proc/dev/unix.  ...

> Apr 30 09:49:37 edmonds vmlinux: Unable to handle kernel paging request at address c0000000
> ...

> PS: why does my hard drive seem to spin every time I get these kernel 
>     messages? 

It's the behaviour of the syslogd(8) logging daemon.

The idea is that a kernel log message might conceivable indicate
imminent kernel failure.  As a precaution, syslogd tries to ensure that
the log file is updated on physical media whenever a new message is
recorded, so that after a crash it is possible to recover the error.

The original syslogd program used fsync(2) to ensure that just the log
file was updated correctly; however, since the default Linux kernel does
not support fsync(), most syslogd distributions for Linux use sync()
instead - and hence they result in a lot more hard disk activity than
necessary.

The latest libc.so contains an fsync(2) call, although it currently just
maps to sync().  syslogd and related programs should probably now be
converted to use fsync; they will continue to work in the mean time, and
when (if!) my fsync patches reach the official kernel, a simple library
update will allow the proper fsync facility to be used.

Cheers,
 Stephen Tweedie.
---
Stephen Tweedie <sct@uk.ac.ed.dcs>   (Internet: <sct@dcs.ed.ac.uk>)
Department of Computer Science, Edinburgh University, Scotland.



------------------------------

End of KERNEL Digest
********************
-------
