From owner-cvs-all  Thu Jan 11 12:50:16 2001
Delivered-To: cvs-all@freebsd.org
Received: from lists01.iafrica.com (lists01.iafrica.com [196.7.0.141])
	by hub.freebsd.org (Postfix) with ESMTP
	id CCA4537B400; Thu, 11 Jan 2001 12:49:46 -0800 (PST)
Received: from nwl.fw.uunet.co.za ([196.31.2.162])
	by lists01.iafrica.com with esmtp (Exim 3.12 #2)
	id 14GofI-0005Ul-00; Thu, 11 Jan 2001 22:49:44 +0200
Received: (from nobody@localhost) by nwl.fw.uunet.co.za (8.8.8/8.6.9) id WAA15064; Thu, 11 Jan 2001 22:49:41 +0200 (SAST)
Received: by nwl.fw.uunet.co.za via recvmail id 15050; Thu Jan 11 22:49:03 2001
Received: from sheldonh (helo=axl.fw.uunet.co.za)
	by axl.fw.uunet.co.za with local-esmtp (Exim 3.20 #1)
	id 14Goec-0002g9-00; Thu, 11 Jan 2001 22:49:02 +0200
From: Sheldon Hearn <sheldonh@uunet.co.za>
To: Jordan Hubbard <jkh@winston.osd.bsdi.com>
Cc: obrien@freebsd.org, Doug Barton <dougb@freebsd.org>,
	cvs-committers@freebsd.org, cvs-all@freebsd.org
Subject: Re: cvs commit: src/etc crontab rc src/etc/defaults rc.conf src/etc/mtree BSD.root.dist src/libexec Makefile src/libexec/save-entropy Makefile save-entropy.sh 
In-reply-to: Your message of "Thu, 11 Jan 2001 12:36:23 PST."
             <19283.979245383@winston.osd.bsdi.com> 
Date: Thu, 11 Jan 2001 22:49:02 +0200
Message-ID: <10300.979246142@axl.fw.uunet.co.za>
Sender: owner-cvs-all@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG



On Thu, 11 Jan 2001 12:36:23 PST, Jordan Hubbard wrote:

> I also appreciate the arguments for having /var be "too late" in the
> boot process, at least as things currently stand, but simply assuming
> that root is writable as your work-around isn't a very safe solution
> either.

*sigh*

I made a mistake in not explaining the situation _fully_ right at the
beginning.  I assumed that people would actually _look_ at rc as it was
before and after this commit before commenting. :-(

So...

Before Doug's commit:

	Because of limitations in mount_mfs(8), the entropy seed files
	could not be stored in /var/db/ without causing a hang on
	post-crash startup.  The hang would persist until a sufficient
	amount of keyboard activity was registered by the entropy
	device.

	Therefore, the entropy seed file was stored in / by rc.shutdown.
	On startup, rc would try to read this file.  On failure, it
	would fall back to trying /var/db/, but this behaviour would
	still enjoy the hang.

	Also, a crash would mean that rc.shutdown would not store the
	entropy seed file at all!

	This system did _NOT_ cater completely for read-only root
	partitions.

After Doug's commit:

	Now, cron(8) stores entropy seed files in /.entropy .  We can't
	use /var/db/ yet because of (at least) the same mount_mfs(8)
	problems we had before problems.

	Since multiple files are used, there is no race condition and we
	can rest assured that, even after a crash, there will always be
	entropy seed files available on startup.

	This system still does not cater completely for read-only root
	partitions.

In the future:

	When those mount_*(8) utilities that require entropy (currently,
	mount_mfs(8) and mount_nfs(8) are the only suspects) have been
	modified to avoid using the new entropy device, all we'll need
	to do is swing the rc.conf variable entropy_dir from /.entropy/
	to /var/db/ .

	This system _WILL_ cater completely for read-only root
	partitions.

I trust that this clears up the hype. :-)

Seriously, folks, what Doug has implemented is no worse than what we had
before and is a step toward something much better than what we had
before.

Ciao,
Sheldon.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message