Date: Fri, 01 Jun 2018 19:37:02 -0500
From: Robert Nestor <[email protected]>
Message-ID: <[email protected]>
| Im not convinced that booting in UEFI mode actually works in 8.0_RC1.
I think it does, but ...
| (On my system UEFI takes precedence and so it will load the NetBSD kernel
which
| then fails to run.
That indicates tat the booting worked, but then the kernel failed to deal with
the hardware properly. That's true on my laptop as well, though it is now a
little better in NetBSD current.
UEFI firmware seems to not configure the hardware the same as BIOS firmware
does, and NetBSD is largely assuming the latter, as it has previously never had
to deal (on x86 type systems) with anything else.
| As far as I can tell use of GPT wedges is separate from the style of boot
in NetBSD,
It is.
| So, unless you build a system without the BIOS boot files and/or totally
disable Legacy/BIOS
| boot capability in your system you cant be sure if you booted via UEFI
just because the
| system came up or is using wedges for partitions.
Not from just those, but the boot program prints a different message, so you
can see which one was used, and there's a sysctl (at least in current) which
indicates which boot method was used.
machdep.bootmethod = BIOS
(it says UEFI when that is what was used.)
| Not sure if the non-working UEFI boot is generally known or if a PR has
been filed though.
UEFI booting works, the kernel has some issues (more in 8 than in current, and
the "fix" is somewhat controversial, and not likely to be pulled up soon)
There is a PR which is related (port-amd64/53286)
In reply [email protected] said:
| However, I wouldn't be surprised if GPT was required to be able to boot via
| UEFI.
It is. UEFI booting loads the boot program from a specially designated (FAT
formatted) GPT partition. What the boot program does after that is up to it.
kre
