Date: Wed, 11 May 2011 13:10:48 -0400 From: Doug Rabson <dfr@rabson.org> To: Andriy Gapon <avg@freebsd.org> Cc: FreeBSD Arch <freebsd-arch@freebsd.org> Subject: Re: arch-independent disks in sys/boot, libstand Message-ID: <BANLkTint4ZtpbRLMXUAbhDLkc1EYstAVxA@mail.gmail.com> In-Reply-To: <4DCA684B.4020400@FreeBSD.org> References: <4DC95006.1020404@FreeBSD.org> <BANLkTik6709-66dXy2RB3Qy6FgRBmJ5hGA@mail.gmail.com> <4DCA684B.4020400@FreeBSD.org>
index | next in thread | previous in thread | raw e-mail
On 11 May 2011 06:43, Andriy Gapon <avg@freebsd.org> wrote: > on 11/05/2011 12:16 Doug Rabson said the following: > > There is an example of this in sys/boot/zfs. The approach taken is to > scan > > physical drives as they appear and piece them together into ZFS pools > based on the > > drive metadata. The pools appear as devices in the device table and the > actual ZFS > > filesystem reader is built on top of that. > > Yes, thank you, I've seen that code. > My question was not specific enough - I am interested in the things like > i386_fmtdev vs other archs, i386_parsedev vs other archs, struct > i386_devdesc vs > other archs, etc. > I.e. I would like to make ZFS representable (in some shape) via struct > devdesc. > But it seems that every supported arch has its own extension of struct > devdesc > that it uses and its own set of functions to work with that struct. > Or, using another example, I'd like to be able to specify something like > "zfs:/pool/fs:/boot/kernel" as a kernel name and have that work for all > archs. > I'd hate to duplicate code that would handle that acress all xxx_parsedev > implementations for all architectures. > The fmtdev/parsedev stuff is a huge mess. In many if not all cases, the device and partition code is identical across the different arches and should be shared.home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?BANLkTint4ZtpbRLMXUAbhDLkc1EYstAVxA>