OK

No objection from me

Sure. I think they could just stay where they are until the refactor but OK.

EFI_MEMORY_ATTRIBUTE_PROTOCOL is the interface to change mapping

I'll land it with the message changed

I've been AFK for a bit and am just working through the backlog from while I was away, pushing accepted reviews. I'm fine with you making the change.

Review was for reference only; update has been done.

@jrtc27 are you fine with this approach?

Name change is fine with me

See also D14031. I didn't like the idea of offering all tools with both efi* and uefi* names, but standardizing on efi* and leaving backwards compat links for tools that started as uefi* seems fine.

LGTM

Oh, via for_each_pci_dev of course

Are there actually any callers with non-NULL odev? (I.e., is this not an issue in practice today)

This is an alternative view of D58083, and that one is most useful for review.

Let me know your diff after freebsd-configure.sh and freebsd-namespace.sh, as there were a few changes there as well.

servconf_add_hostkey fixup reported by @jlduran

Do any file systems besides tmpfs support exchange?

LGTM

Switching to bool and cgem_get_mac rework LGTM and could be committed independently if you like. I haven't looked at the rest yet.

pdopenpid() should be not allowed lightly for capsicum

The pdopenpid's pid arg is accessing a global namespace and thus not permitted in Capsicum (other than potentially 0 for self), but as you note there are several ways to obtain the pd.

I don't have enough background to give it a thorough review but I'm happy with it.

Detect symlink in _toolchain_tools_to_symlink loop and add symlink target also.

Yes, not unconditionally. I'd want to add it in _toolchain_tools_to_symlink though.

Alternately we could just add ld.lld to _toolchain_tools_to_symlink