Re: [arch-releng] GPT support in AIF

On 12/07/2010 10:54 AM, Dieter Plaetinck wrote:

On Tue, 7 Dec 2010 17:19:23 +0100 Heiko Baums wrote:

...

Am Tue, 7 Dec 2010 09:35:29 -0600 schrieb C Anthony Risinger:

...

how about pulling btrfs-progs into core as well? and on that note, where does btrfs stand in terms of AIF? I'd like to see btrfs in [core] and AIF, too.

While we're on it, what's about supporting GPT (GUID Partition Table) in AIF? Since the harddisks are getting bigger nowadays it will soon be necessary as far as I know, because an MBR partition can only have a size of 2 TB.

Heiko I'm not very familiar with this topic. Nor am i familiar with the related topic of block boundary sizes and stuff. Although tpowa tried to explain to me, this is what I wrote down:

* don't use sfdisk or cfdisk, they don't support 1MB bootsectors (which is a new standard for 4k sector drives, better alignment and win7) * parted and fdisk do. * mbr is soon dead * https://ata.wiki.kernel.org/index.php/ATA_4_KiB_sector_issues * parted is also good for GPT and uefi usage * fdisk does not support GPT * but parted has no (ncurses) UI

so yeah, lets figure out if this info is still up to date, and how we should proceed. it looks like we shoud just use parted, if we can find/make an ncurses frontend for it, all problems should be magically solved.

Dieter I believe this is accurate. I have used GPT for large raid arrays on other Linux distros. The prevailing way to deal with GPT is to leave it out of the installer, configure the OS on a non GPT partition, then after install to use parted to make the GPT partitions. I believe to boot to a GPT partition you need EFI or UEFI, booting to GPT from a regular bios is not well supported, and most motherboards presently do not have EFI/UEFI except for some enterprise servers.

A good first step that would put Arch beyond other distros from what I've seen would be support for non-boot GPT partitions in the installer via parted. -JT