Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 11 Feb 2003 09:34:06 -0500 (EST)
From:      Robert Watson <rwatson@FreeBSD.org>
To:        Juli Mallett <jmallett@FreeBSD.org>
Cc:        Peter Wemm <peter@wemm.org>, src-committers@FreeBSD.org, cvs-src@FreeBSD.org, cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/sbin/newfs Makefile mkfs.c newfs.c newfs.h ref.test
Message-ID:  <Pine.NEB.3.96L.1030211093249.20581A-100000@fledge.watson.org>
In-Reply-To: <20030210220423.A36674@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help

On Mon, 10 Feb 2003, Juli Mallett wrote:

> > Umm, I object!  This is a massive performance loss on disks without track
> > write buffering (eg: many raid systems)..  While it isn't so bad for ufs2
> > newfs, it is a Big Deal for ufs1 for interchangeability.
> 
> OK.  I was kind of hoping to put a general write cache (and since we do
> reads, a general buffer cache, in a sense) into libufs, but I suppose I
> could just put back the wc stuff, and avoid calling into libufs for
> anything other than bread/bwrite. 
> 
> Which would you prefer?  Do you know of anything other than newfs where
> it matters? 

If you're going to use a write cache with fsck, you need to be Really
Really Careful.  The last thing we want is a power-down during fsck to
completely hose the system because it violates write-order constraints.
Hopefully fsck isn't in the business of producing such things, but caution
should be applied.

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert@fledge.watson.org      Network Associates Laboratories



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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.96L.1030211093249.20581A-100000>