From owner-freebsd-hackers  Wed Feb 17 15:18:49 1999
Delivered-To: freebsd-hackers@freebsd.org
Received: from uni4nn.gn.iaf.nl (osmium.gn.iaf.nl [193.67.144.12])
	by hub.freebsd.org (Postfix) with ESMTP
	id 60AE21131F; Wed, 17 Feb 1999 15:18:45 -0800 (PST)
	(envelope-from wilko@yedi.iaf.nl)
Received: (from uucp@localhost)
	  by uni4nn.gn.iaf.nl (8.9.2/8.9.2) with UUCP id AAA25948;
	  Thu, 18 Feb 1999 00:04:16 +0100 (MET)
Received: (from wilko@localhost) by yedi.iaf.nl (8.8.8/8.6.12) id TAA00897; Wed, 17 Feb 1999 19:36:35 +0100 (CET)
From: Wilko Bulte <wilko@yedi.iaf.nl>
Message-Id: <199902171836.TAA00897@yedi.iaf.nl>
Subject: Re: savecore before swapon?
In-Reply-To: <19990217142628.F69668@bitbox.follo.net> from Eivind Eklund at "Feb 17, 99 02:26:28 pm"
To: eivind@freebsd.org (Eivind Eklund)
Date: Wed, 17 Feb 1999 19:36:35 +0100 (CET)
Cc: brandon@roguetrader.com, freebsd-hackers@freebsd.org
X-Organisation: Private FreeBSD site - Arnhem, The Netherlands
X-pgp-info: PGP public key at 'finger wilko@freefall.freebsd.org'
X-Mailer: ELM [version 2.4ME+ PL38 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

As Eivind Eklund wrote...
> On Tue, Feb 16, 1999 at 08:26:00AM -0700, Brandon Gillespie wrote:
> > I havn't checked the source, but from my understanding shouldn't
> > savecore be run before swapon is run, incase the swap device is the
> > dump device?  Or to look at it another way, when swapon is run on a
> > swap device, does it look first to see if there is a dump in it, and
> > if so what does it do?  Right now we run swapon, then considerably
> > later we run savecore.  Assuming swapon just trashes whatever was in
> > that device, running savecore is pretty much irrelevant, as most
> > people I know (not necessarily in FreeBSD) use their swap device as a
> > dump point.
> 
> This is done intentionally, as fsck may in some cases require swap
> to be able to fsck large filesystems.
> 
> It would make sense to have a sysctl
> 'only_swap_as_absolutely_last_resort_not_for_performance' AKA
> 'do_linux_swapping_scheme' which was set before the swapon, and unset
> after the crashdump.  I don't know how to implement this.  (It might
> be that one of the present sysctls does something like this - I've not
> checked).

Or start allocating swap 'backwards' from the high block# on the disk
downwards. Assuming swap > physical mem and dumps starting at swap offset 0
this should be safe at all times (unless fsck really allocates all swap
of course ;-)

But this might be a lot of work, I'm not familiar with the VM.

Wilko
_     ______________________________________________________________________
 |   / o / /  _  				Arnhem, The Netherlands
 |/|/ / / /( (_) Bulte 				WWW  : http://www.tcja.nl
______________________________________________ Powered by FreeBSD __________


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