From owner-freebsd-arch Mon Jun 10 11:29:43 2002 Delivered-To: freebsd-arch@freebsd.org Received: from critter.freebsd.dk (critter.freebsd.dk [212.242.86.163]) by hub.freebsd.org (Postfix) with ESMTP id 4555337B401 for ; Mon, 10 Jun 2002 11:29:38 -0700 (PDT) Received: from critter.freebsd.dk (localhost [127.0.0.1]) by critter.freebsd.dk (8.12.3/8.12.2) with ESMTP id g5AISAV7098153 for ; Mon, 10 Jun 2002 20:28:10 +0200 (CEST) (envelope-from phk@critter.freebsd.dk) To: arch@freebsd.org Subject: Statement of architectural direction: disklabel64 / GPT. From: Poul-Henning Kamp Date: Mon, 10 Jun 2002 20:28:10 +0200 Message-ID: <98152.1023733690@critter.freebsd.dk> Sender: owner-freebsd-arch@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG The issue of 32->64 bit migration of struct disklabel has come up now that daddr_t is 64 bit wide. I have pondered the issue and researched what material I could find and come to the conclusion that we should not make a 64bit version of struct disklabel, but instead use the industry standard GPT format. A 64bit BSD style disklabel would in best case be a {Free,Net,Open}BSD thing, worst case just a FreeBSD thing. Either way I will argue that it would be a private format. Unless it can provide functionality otherwise not possible there is no point in adding a new private format. The GPT handles 16k partitions, 64 bit addressing, has decently checksummed and redundant meta-data and space for per partition meta-data. I can't think of anything else we might need, so I have decided not to make a 64bit disklabel format and instead use the GPT format. This may result in us using the GPT format in ways not anticipated by the standard (ie: embedded in a native partition on some odd-ball platform) but that should not give any more or less trouble than embedding a disklabel64 there. We will need to support GPT for at least ia64 anyway, and I predict that it will sneak into ia32 RSN as well, so this is actually less work for us than doing a disklabel64. Any objections ? -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 phk@FreeBSD.ORG | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-arch" in the body of the message