Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 24 May 2007 09:57:14 +0000
From:      Darren Reed <darrenr@freebsd.org>
To:        Bruce Evans <brde@optusnet.com.au>
Cc:        Craig Boston <craig@xfoil.gank.org>, Pawel Jakub Dawidek <pjd@freebsd.org>, freebsd-fs@freebsd.org, freebsd-current@freebsd.org, Kris Kennaway <kris@obsecurity.org>, Vince <jhary@unsane.co.uk>
Subject:   Re: ZFS committed to the FreeBSD base.
Message-ID:  <20070524095714.GA52149@hub.freebsd.org>
In-Reply-To: <20070523212559.V10628@besplex.bde.org>
References:  <20070410011125.GB38535@xor.obsecurity.org> <20070410013034.GC8189@nowhere> <20070410014233.GD8189@nowhere> <4651BD6F.5050301@unsane.co.uk> <20070522083112.GA5136@hub.freebsd.org> <4652B15D.5060505@unsane.co.uk> <20070523085532.GA27542@hub.freebsd.org> <20070523093231.GA29797@xor.obsecurity.org> <20070523100631.GA30143@xor.obsecurity.org> <20070523212559.V10628@besplex.bde.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, May 23, 2007 at 09:45:16PM +1000, Bruce Evans wrote:
> On Wed, 23 May 2007, Kris Kennaway wrote:
> 
> >On Wed, May 23, 2007 at 05:32:31AM -0400, Kris Kennaway wrote:
> >
> >>I would actually be interested to know how Solaris gets away with
> >>this.  It sounds like there must be less of a distinction between
> >>memory allocated to the kernel and to userland, and the ability for
> >>memory to flow between these two with some form of backpressure when
> >>userland wants memory that is currently gobbled by up solaris ZFS.
> >>
> >>This kind of system probably makes good sense (although maybe there
> >>are trade-offs), but anyway it's not how FreeBSD does it.
> >
> >After some further thought I guess the difference is just that on a
> >64-bit kernel you don't have KVA issues and can indeed map all of
> >physical RAM into the kernel for caching.
> 
> This should probably happen for 64-bit kernels in FreeBSD too.  FreeBSD
> sizes the buffer map part of KVA in the same way on all arches, to squeeze
> it into the limited available space on i386's, and has large complexity
> and some loss of performance in the buffer cache in order to work with
> the limited KVA.  (Very old versions had less complexity and a large loss
> of performance.)

Do we support variable page sizes in the kernel for amd64?

Darren



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20070524095714.GA52149>