Re: [PATCH] BIOS boot vs EFI system partition mountpoint

[Taylor R Campbell <riastradh%NetBSD.org@localhost>]

> Date: Sun, 21 Aug 2022 01:11:39 +0700
> From: Robert Elz <kre%munnari.OZ.AU@localhost>
> 
>     Date:        Sat, 20 Aug 2022 16:14:46 +0000
>     From:        Taylor R Campbell <riastradh%NetBSD.org@localhost>
>     Message-ID:  <20220820161441.BDEFD609EB%jupiter.mumble.net@localhost>
> 
>   | Nothing in this change will automatically move /boot to /biosboot.
> 
> When one builds a system (if that change gets included) what
> gets complied and installed in ${RELEASEDIR}/ and then in the
> base.tgz set?   /boot or /biosboot?   What do the sets lists
> expect to exist?   Does the mtree file contain /boot as a directory
> that needs to be created, or not?

The x86 /boot file does not appear in any of these, and never has.

The only way x86 /boot has ever been created is by sysinst -- and only
when installing on systems with BIOS boot -- or by manually copying it
there.  All I changed is (a) where bootxx_* looks, (b) where sysinst
puts it, and (c) some documentation.

As far as I know, there is nothing else automated in-tree that touches
any of this -- not even postinstall or installboot.

The MI dtb sets, and MD evbarm and evbmips (and riscv) sets, and the
corresponding mtree files, _do_ have /boot as a directory, and they
use that for both EFI and FDT boot.  Changing this is a much bigger
pain that I'm not inclined to deal with -- even if there were a better
name (`/efi' is obviously wrong because it's not just for EFI).

Some MD base sets have /boot as a file (mipsco and vax) but I doubt
EFI or FDT will ever be relevant for them.  I'm also not sure what
part of the build puts them there or whether it's relevant at all,
because, e.g., sysinst/arch/mipsco/md.c has logic to copy
/usr/mdec/boot to /boot anyway.