linux-raid.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: John Robinson <john.robinson@anonymous.org.uk>
To: "Lars Täuber" <taeuber@bbaw.de>
Cc: linux-raid@vger.kernel.org
Subject: Re: grub2/grub-pc install not possible on mdadm
Date: Mon, 05 Mar 2012 11:31:54 +0000	[thread overview]
Message-ID: <4F54A42A.4020903@anonymous.org.uk> (raw)
In-Reply-To: <20120305111402.e63fee5b49046ed146cb9dcc@bbaw.de>

On 05/03/2012 10:14, Lars Täuber wrote:
> Hi John,
>
> Am Mon, 05 Mar 2012 09:38:34 +0000
> John Robinson<john.robinson@anonymous.org.uk>  schrieb:
>> On 05/03/2012 07:50, Stefan Priebe - Profihost AG wrote:
>>> Am 02.03.2012 16:32, schrieb Lars Täuber:
>>>> Hi Stefan,
>>>>
>>>> the metadata versions 1.0 and 1.1 and 1.2 are the same but for the
>>>> position of the metadata on disk. Simply use metadata version 1.0 and
>>>> everything works as with 0.9.
>>>>
>>>> Grub does to read partition tables inside raid devices.
>>
>> Sure it does, but that's not the point.
>
> that's new to me. This is true for grub2 only?
> You mean if Stefan would have used an old fashioned partition table inside a
> Linux SW-RAID with metadata 1.2, grub would have booted like normal?

Yes, both grub2 and grub legacy (0.97) read partition tables. Only grub2 
can really read inside RAID arrays - any metadata, any RAID level - 
while grub legacy can only read the first drive of RAID-1 with 0.90 or 
1.0 metadata, and that only because they look like bare drives/partitions.

>>>> Because a raid1 with metadata 1.0 or 0.9 looks like a plain disk to grub
>>>> it can read a partition table like there was no raid array.
>>
>> For the fourth time, no.
>>
>> Because a RAID1 with metadata 0.9 or 1.0 looks like a plain disc to the
>> BIOS, it can read a partition table like there was no RAID array, and
>> therefore the BIOS can load GRUB.
>
> This might be true for UEFI firmware but for the plain old BIOS too?
> I thought the BIOS isn't interested in partitions at all.

That's true, but it is interested in the MBR, sector zero of the drive 
it's booting from, which is also where the partition table lives, so for 
plain old BIOS you have to put the stage 1 boot loader there, which can 
only happen with metadata 0.90 and 1.0 if you use whole-drive arrays.

With plain old BIOS, grub legacy and grub2 also depend on the next 31K 
of the disc, up to sector 62, being unused, as traditionally the first 
partition starts at sector 63. (More recent fdisk make it 1023.5K, 
sector 2047 and sector 2048 respectively). This means you have to have a 
partition table, and though the plain old BIOS doesn't use it, grub 
does, to find the /boot or / partition to load a kernel from. With grub 
legacy, it has to be an MBR partition table, not one buried inside an 
array with metadata 1.1 or 1.2.

I suppose you could possibly contrive to have no MBR partition table for 
grub2, but you'd still need that space on the drive for the stage 1.5 
code, which isn't available with a whole-drive metadata 1.2 array 
(because its metadata starts 4K into the drive, leaving 3.5K after the 
stage 1 on the MBR), so you do effectively need an MBR partition table 
to reserve the space.

But you're right, plain old BIOS don't use the MBR partition table, just 
the MBR. I was fashioning my sentence after yours, so sorry for any 
confusion. My point - that it's the BIOS that has to be able to read the 
array like it was a bare drive, rather than grub - remains.

Cheers,

John.
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

  reply	other threads:[~2012-03-05 11:31 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-02-28  9:10 grub2/grub-pc install not possible on mdadm Stefan Priebe - Profihost AG
2012-03-01 10:10 ` Jes Sorensen
2012-03-01 10:25   ` Stefan Priebe - Profihost AG
2012-03-01 10:38     ` Jes Sorensen
2012-03-02  8:44       ` Stefan Priebe - Profihost AG
2012-03-02  9:06         ` Lars Täuber
2012-03-02  9:11           ` Stefan Priebe - Profihost AG
2012-03-02  9:24             ` John Robinson
2012-03-02  9:33               ` Stefan Priebe - Profihost AG
2012-03-02 11:14                 ` John Robinson
2012-03-02 13:15                   ` Stefan Priebe - Profihost AG
2012-03-02 13:45                     ` John Robinson
2012-03-02 15:32                     ` Lars Täuber
2012-03-05  7:50                       ` Stefan Priebe - Profihost AG
2012-03-05  9:38                         ` John Robinson
2012-03-05 10:14                           ` Lars Täuber
2012-03-05 11:31                             ` John Robinson [this message]
2012-03-05 11:38                               ` Lars Täuber
2012-03-05 12:34                                 ` John Robinson
2012-03-02  9:14           ` Lars Täuber
2012-03-02  9:15           ` Lars Täuber
2012-03-02 10:56         ` Jes Sorensen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4F54A42A.4020903@anonymous.org.uk \
    --to=john.robinson@anonymous.org.uk \
    --cc=linux-raid@vger.kernel.org \
    --cc=taeuber@bbaw.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).