Jump to content

Welcome to MSFN Forum
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. This message will be removed once you have signed in.
Login to Account Create an Account



Photo

Possible to Boot from 4 TB System Volume with BIOS?


  • Please log in to reply
11 replies to this topic

#1
NoelC

NoelC

    Software Engineer

  • Member
  • PipPipPipPipPipPip
  • 1,435 posts
  • Joined 08-April 13
  • OS:Windows 8.1 x64
  • Country: Country Flag

I'm not completely solid in my understanding of the restrictions of installing / booting Windows with a huge system volume.

 

I have a Dell Precision T5500 Workstation based on BIOS, without the possibility of UEFI as far as I know.

 

Presently I have a 2 TB system volume (C:) with MBR partitioning.  Works great.

 

But what if I wanted to make a 4 TB system volume?

 

The advantages would be that with more drives in the array making up the volume performance would be better, and also there would be a huge amount of free space.

 

What would the steps be to partition 4 TB of space using GPT?  Would I be able to install Windows on it?  Would it boot?

 

Thanks in advance for any wisdom and experience you are willing to share.  There's surprisingly little how-to on this out there.

 

-Noel




How to remove advertisement from MSFN

#2
MagicAndre1981

MagicAndre1981

    after Windows 7 GA still Vista lover :)

  • Patrons
  • 6,161 posts
  • Joined 28-August 05
  • OS:Vista Ultimate x86
  • Country: Country Flag

Donator

booting a GPT drive requires UEFI. Create 2 MBR partitions where the boot partition is smaller compared to 2.2 TB.


Posted Image

#3
jaclaz

jaclaz

    The Finder

  • Developer
  • 14,822 posts
  • Joined 23-July 04
  • OS:none specified
  • Country: Country Flag

Well, not really-really.

 

The limit in size is still there of course, but it is perfectly possible with a few tricks to boot a GPT disk from BIOS, and what happens later depends on the size of disk sector and on the actual OS.

 

A MBR partition entry still has a field for size that is a double word so anything more than FFFFFFFF or 4294967295 can't be wrtten to it, 

so that anything above 4294967295*512=2199023255040 will still become a suffusion of yellow.

 

So, you need anyway a small "system" (MS terminology reversed, "wrong") or "boot" (logical terminology, "right") volume, and then the OS needs to be able to use GPT disks natively.

 

This is normally achieved using a hybrid MBR/GPT disk, which has a number of issues:

http://www.rodsbooks...isk/hybrid.html

 

The above site http://www.rodsbooks.com/gdisk/is currently the most complete (and seemingly only) source of information on GPT disks in the "Windows world".

 

However:

http://reboot.pro/to...in-bios-to-gpt/

http://reboot.pro/to...-gpt-partition/

 

It is possible to Boot from BIOS on a non-hybrid GPT disk :yes:.

 

Final (at the moment) solution here:

http://reboot.pro/to...o-gpt/?p=186656

 

jaclaz



#4
NoelC

NoelC

    Software Engineer

  • Member
  • PipPipPipPipPipPip
  • 1,435 posts
  • Joined 08-April 13
  • OS:Windows 8.1 x64
  • Country: Country Flag

Ah yes, I do remember running across your writeup before - some of the only info out there on the subject.  Though the backflips required to create such a setup initially seem daunting, I'm gathering that once it works, it keeps working unless you try to re-partition the drive.

 

I'm not sure how the needs of the RAID controller might fit into it all as well.  That might be a deal-breaker.  I don't know how many assumptions they make between the BIOS setup and the run-time driver about what's where.

 

Perhaps if I find myself retiring my existing array and building a whole new one (i.e., where I can just plug the old hardware back in and restore functionality in a few moments) I'll experiment.

 

-Noel



#5
jaclaz

jaclaz

    The Finder

  • Developer
  • 14,822 posts
  • Joined 23-July 04
  • OS:none specified
  • Country: Country Flag

Though the backflips required to create such a setup initially seem daunting, I'm gathering that once it works, it keeps working unless you try to re-partition the drive.

As a matter of fact the (smart? :unsure:) main point of that experimental approach is that the "hidden" partition (i.e. if it is not "indexed" in the GPT partitition table) is pretty much "hidden" and "safe" in "normal operation".
It is placed on sectors 63-2047:

  • the MBR is on LBA 0
  • the EFI PART header is on LBA 1
  • the GPT partition table will be starting on sector 2 and extend up to sector 33 for 128 available partition entries - unless you create a zillion partitions/volumes in it, it will *never* reach sector 63
  • the Disk Manager and/or diskpart in the default settings will do nothing to any sector before 2048 on a largish disk
  • the adding or removing of volumes will affect the GPT partition table but not the MBR, leaving untouched the protective entry and the code (that should not be there)
  • changes (if any) in Disk Signature will obviously not affect anything "on disk" though depending on the actual contents of the \boot\BCD in the "hidden" partition this may need to be adapted

So, in practice only if you manage to delete the Magic Bytes 55AA from the MBR (thus creating the need to re-initialize the disk) you will have to "re-start from scratch", but if you re-initialize the disk you need to start from scratch anyway.
 

I'm not sure how the needs of the RAID controller might fit into it all as well.  That might be a deal-breaker.  I don't know how many assumptions they make between the BIOS setup and the run-time driver about what's where. 

If it's a hardware controller, it's operation is AFAIK "completely transparent" to BIOS (and of course as well to MBR code and grub4dos), so it should not change anything.

 

jaclaz



#6
NoelC

NoelC

    Software Engineer

  • Member
  • PipPipPipPipPipPip
  • 1,435 posts
  • Joined 08-April 13
  • OS:Windows 8.1 x64
  • Country: Country Flag

Or maybe, just maybe, Microsoft will work things out so that ReFS becomes available for use as boot / system volumes, and in the process of doing so will rework the boot mechanics so that the boot / system volume can be any (practical) size and the setup of same can be done by any member of the Federation, not just the pointy-eared variety.

 

Has anyone heard of any progress with ReFS?  I've been running data volumes with it for over a year now with no apparent problems.

 

Edit:  But on reflection, I'm thinking they wouldn't give a darn about BIOS any more.  Guess I need to start thinking about a newer workstation.  More cores, faster cores, and of course the wonderful new Unified Extensible Firmware Interface - the answer to all prayers.

 

-Noel


Edited by NoelC, Yesterday, 10:48 AM.


#7
jaclaz

jaclaz

    The Finder

  • Developer
  • 14,822 posts
  • Joined 23-July 04
  • OS:none specified
  • Country: Country Flag

Or maybe, just maybe, Microsoft will work things out so that ReFS becomes available for use as boot / system volumes, and in the process of doing so will rework the boot mechanics so that the boot / system volume can be any (practical) size and the setup of same can be done by any member of the Federation, not just the pointy-eared variety.

But this has nothing to do with the actual filesystem, the "normal" NTFS can do nicely much more than any "practical" size :

http://en.wikipedia....TFS#Scalability

 

The good MS guys introduced artificially (as they often do) an incompatibility between BIOS based hardware and GPT disks, or better didn't bother to update the real mode part of the BOOTMGR, most probably in order to push the EFI/UEFI platform, there is no real reason why the non-EFI bootmgr cannot access/understand/parse/use GPT volumes.

 

jaclaz



#8
dencorso

dencorso

    Iuvat plus qui nihil obstat

  • Supervisor
  • 6,093 posts
  • Joined 07-April 07
  • OS:98SE
  • Country: Country Flag

Donator

@NoelC: With all due respect:

 

ReFS and exFAT are just two more (of the many) ways MS has of imposing closed-source, undocumented, unnecessary standards, where they aren't really needed at all... *and* to support and promote them is tantamount to supporting and promoting evil things like Metro / Universal UI. :puke:

 

Now, creating a free update to the real-mode part of the BOOTMGR to allow the non-EFI version GPT capabilities, OTOH, might be a really worthy project, and maybe even demand not too much time to reach real-word usability. :angel



#9
jaclaz

jaclaz

    The Finder

  • Developer
  • 14,822 posts
  • Joined 23-July 04
  • OS:none specified
  • Country: Country Flag

Now, creating a free update to the real-mode part of the BOOTMGR to allow the non-EFI version GPT capabilities, OTOH, might be a really worthy project, and maybe even demand not too much time to reach real-word usability.  :angel

And, conversely, now that CHS is not used anymore in actual OS operations, it would be trivial to invent a "new standard" :w00t: where the CHS area of the MBR partition entry is used to hold the further "high bytes" of the LBA addresses, giving it a new "generic" protective partition ID, by stealing  :ph34r: one of the ID's that are unused and that were (senselessly) claimed or "reserved" by Operating Systems that never were or that are totally irrelevant/abandoned: 

http://www.win.tue.n...n_types-1.html 

so that the only limit/requirement for booting with an UNmodified BIOS would be that the Active partition has "conventional LBA addressing" and resides within the first 2.2 Tb of the disk, making m00t of the supposed *need* for GPT disks.

 

For the record the *only* "advantage" GPT disks have in "real world" is (besides the larger addressing extents) that you can have countless (like 128 or more) primary partitions (not really a "limit", since most people seemingly use single extremely large volumes and rarely needs more than 4, and even when they do, logical volumes work fine since what 20 or more years) and the freedom from the 256 partition ID's limit, which, given that the available number is now so mindboggingly huge makes it improbable that there will be a collision (see point #7 here): 

http://reboot.pro/to...o-gpt/?p=186493

 

And - still for the record - more wasted bytes:

http://reboot.pro/to...o-gpt/?p=186341

 

jaclaz



#10
NoelC

NoelC

    Software Engineer

  • Member
  • PipPipPipPipPipPip
  • 1,435 posts
  • Joined 08-April 13
  • OS:Windows 8.1 x64
  • Country: Country Flag

ReFS and exFAT are just two more (of the many) ways MS has of imposing closed-source, undocumented, unnecessary standards, where they aren't really needed at all... *and* to support and promote them is tantamount to supporting and promoting evil things like Metro / Universal UI. :puke:

 

Ah, no.  The file system implementation has substance; Metro/Modern does not.  Probably why we haven't seen advancement of the former.  No one's left in the Republic who commands the light side of the force.

 

Some of the things ReFS brings to the party are actually good.  That said, I've had zero problems with NTFS in all the years since there was NTFS.  I don't see a need for a new file system, honestly, I just brought it up because one claim to fame is virtually unlimited size, and that bears on this problem.

 

-Noel



#11
dencorso

dencorso

    Iuvat plus qui nihil obstat

  • Supervisor
  • 6,093 posts
  • Joined 07-April 07
  • OS:98SE
  • Country: Country Flag

Donator

Sure. I do understand why you you've brought it up.

 

No matter whether ReFS is great, and I actually don't intend dispute that, it's simply not needed. exFAT much less. Why move on to proprietary FSes, when none new is actually needed? Just to shower money onto MS for them to develop their follies? It makes no sense to me, sorry, none at all, actually! :wacko:

 

Now, fact is: there's NTFS, then there's Ext3 and Ext4. And for non-journaling, there's FAT32 and Ext2. :yes:

 

While it sure is good to have alternatives for when those trusty FSes are not enough, I don't think proprietary FSes are the way to go anymore... not now, and much less later on. Just my 2¢, of course!



#12
jaclaz

jaclaz

    The Finder

  • Developer
  • 14,822 posts
  • Joined 23-July 04
  • OS:none specified
  • Country: Country Flag

I don' t know. :unsure:

 

I mean, I do like some of the features of ReFS, but it's not like any of them (BTW nice) are really-really *needed* or *make a difference* in practice when compared with NTFS (the exception would be of course data centers and similar where they most probably use Zfs or the like, not the PC). 

 

exFAT (which the good MS guys managed to bastardize releasing - without documenting it - a few slightly different versions, including TexFAT) makes no sense at all, if not - maybe - on dedicated hardware (like the phones, and the like).

 

In practice the filesystem that never was :w00t: :ph34r: (though it is available in newer systems) given a chance is UDF (that besides the pompous and extremely inaccurate name, since noone uses it if not on optical discs) which would be otherwise particularly suitable for storage or for USB sticks and the like.

 

jaclaz


  • dencorso likes this




1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users