OpenBSD Journal

MBR Support for EDD Reads (> 8 GB)

Contributed by jose on from the experimental dept.

Snapshots for the i386 architecture since December 16 have included the start of support for booting on partitions above the 8GB marker. You can view the patch on the Arches Consulting OpenBSD pages . From that page:
The following diff makes the OpenBSD master boot record (MBR) able to boot BIOS (fdisk) partitions that start above the 8 GB CHS limit. It will use EDD (LBA) reads if the BIOS indicates that they are available; it will only use CHS reads if the BIOS says they not available. It is possible to override this behaviour, and force CHS reads, by holding down the Shift key as the MBR starts.
Let's see if this works for people.
UPDATE: After discussions with several developers, this story is misleading. This is only partial support for the requirements to boot on partitions over 8GB in. You can get more info in a recently updated FAQ entry from Tom and Nick.

(Comments are closed)

  1. By Anonymous Coward () on

    This is one of my biggest easily-fixed beefs with OpenBSD/i386: the other free OSs have moved beyond this partition limit, but OpenBSD hasn't... and nowadays naive users who aren't familiar with the problem only see that OpenBSD won't boot on a non-first partition and think OpenBSD is junk.

    There are a lot of system-dependent demons in large disk BIOS support. Please, please, *please* test this and report bugs as best you can!

    1. By SH () on

      Unfortunately, I got 'Bad magic' (just after 'reading boot....') when trying to boot via Grub chainloading. I've installed the 31 december snapshot for i386.

      Grub bootloading works fine on another machine (with OpenBSD on the first partition), so I thought the same approach should work on my wifes Windows machine.

      1. By Anonymous Coward () on

        please try the OpenBSD bootloader and not GRUB, if you're experiencing problems due to GRUB, that's a bugreport to the GNU Grub team, not OpenBSD.

        1. By SH () on

          That is also tried. However, on the OpenBSD Misc list it says that the article is premature. Just part of the puzzle is done for booting above 8GB.

  2. By mirabile () on

    Well, for a working LBA capable boot loader
    (which also does El Torito natively and can
    boot even a 10 MB /bsd.rd), you know
    where to look.

    Huh, not? Try for one.

    1. By Sam () on

      If people know where to go, why do you feel the need to post about MirBSD?

  3. By Anonymous Coward () on

    This is probably the best news I've heard this year...

    1. By ciph3r () on

      Nice to hear that december 31 and not january 1 ;)

  4. By Anonymous Coward () on

    I honestly want to know, what's keeping OpenBSD from grabbing code from NetBSD's bootloader, which does support partitions over 8GB? If the problem has already been solved by NetBSD, and NetBSD supports many more platforms compared to OpenBSD, why is OpenBSD not using the NetBSD way of solving things?

    1. By Anonymous Coward () on

      One could ask, if it was so simple, why didn't *YOU*
      do it?
      Given the way openbsd is developed, its either:
      1. Impossible (nah)
      2. Unwanted
      3. Not a priority for the developers

      Given the fact that "global world domination" and "nice
      coexistance with BillOS and others on the same machine"
      seems to not be on the goals.html, the probable
      answer is 3. Those who do grasp bootloading on x86
      most likely run openbsd only on their machines making
      it a non-issue...

    2. By djm () on

      NetBSD's bootloader is a little different to OpenBSD's. But mainly because noone offered a patch until now.

      1. By mirabile () on

        Bläh, NetBSD is different but MirBSD not.
        I offered the code. They rejected because it
        was written in a different assembly language
        dialect than they speak (but gas handles it
        just fine).

        1. By djm () on

          You make it impossible to compare the code, your diffs get rejected. The same rules apply for OpenBSD developers, so why not you?

          1. By mirabile () on

            Of course you can't compare the (low-level) code,
            because it has been rewritten from scratch (sic!).

            How do you expect to compare mashed apples to
            banana milk?

  5. By Anonymous Coward () on

    Dumb question, but does that mean that the Quantity-multipliers in machine mem patch will or are also in -current now?

    That's one thing I'd like to see. I've ran into quite a few Compaq servers and workstations that just don't see the memory properly without adding to boot.conf in hex. FreeBSD and NetBSD's way (if I recall correctly) is a lot nicer in that it's done with base-8.

  6. By Jedi/Sector One () on

    is the unability to boot on extended DOS partitions.

    1. By mirabile () on

      You almost got me. Drop me a mail in two or three
      weeks, when I'll have a smaller TODO, and I might
      implement it and send you diffs.

      1. By Jedi/Sector One () on

        No problem, I will test your diffs.

  7. By STeve Andre' () on

    This is inaccurate. The snapshots do not yet have the ability to boot beyond 8G. They will at some point it sounds like, but not currently.


Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]