Re: i386 Installer questions: MBR slices, ext2fs root, bootblock location

[Nutchanon Wetchasit <xwindows%xwindows.in.th@localhost>]

On Thu, 5 Nov 2026, Greg Troxel wrote:

>>  2. Is it possible to install NetBSD in a manner which everything
>>     (root, home, dev, usr, tmp, var, etc) were confined in a single
>>     root filesystem, with *NO* swap space? If so, would it require completing
>>     some parts of the installation outside of a TUI installer
>>     in order to make it so?
[...]
> Yes that's fine.  I think you can do it with the installer.   Not even odd.

Thanks for confirming.

> not sure it is possible and it is definitely not a good idea.
[...]
> My advice is don't fight nested labels within a block that's hosted.
> It won't help you to win that at all.

I just saw Mr. Elz's answer [1], and I have some hope there; so...
challenge accepted!

> The VM guest doesn't need to see the host partitioning; it can just have
> space from your host LVM, and do GPT in that.

I'm explicitly using a configuration that'd mimic what the host system uses;
so I can get full understanding about all the preparation/configuration steps
("idiosyncrasies", if you must) which would entail if/when
I would migrate (or re-install) to bare-metal.

^ As I already said, my ideal goal is to eventually `dd` the system
  into a new partition on the host disk, or one of my flash drives/SD cards;
  all of which use MS-DOS disklabel.

> If you use GPT, you can avoid all this.

Using GUID disklabel (and everything associated to UEFI) is actually one
of the things I personally oppose doing so [2] for both technical
and *political* reasons-- which I'll not get into here to avoid turning
this email into a rant.

I acknowledge that using GUID disklabel _technically_ does not bar me
from starting my system under Freedom Boot [16-bit BIOS/UEFI-CSM];
however, it does bar other "legacy" OSes that I currently have
(as said in footnote "[2]" of the original question) from sharing
the same disk under bare-metal configuration, or start under
the same BIOS/firmware setting.

So, no thanks. I'm willing to go through the trouble in order to
avoid using such thing under NetBSD, or any other OS for that matter.

> Maybe, but NetBSD's ext2fs code is not 100% ok and it is missing
> features so you have to create any shared ext2fs from NetBSD.

Thanks for the heads up.

> unless you have to use userconfig at boot time.  The kernel has
> drivers, and things are probed at boot.
[...]
> Yes, if an install would have succeeded on the new one.

Acknowledged. And I take that the "userconfig" you mentioned is being
a machine-specific custom kernel build/modules-set that could optionally
be created at the installation time?

Regards,
Nutchanon Wetchasit

-----

[1] "Re: i386 Installer questions: MBR slices, ext2fs root, bootblock location"
    [2026-11-05 02:18:01 +0000]
    <https://mail-index.netbsd.org/netbsd-users/2026/11/05/msg035199.html>
    <news:15017.1793845081%jacaranda.noi.kre.to@localhost>

[2] The only single situation I would tolerate GPT in my setup,
    is on a disk larger than 2 TiB; which none of mine currently are.