Re: UEFI booting survey

On Dec 18, 2017 6:06 PM, "Rodney W. Grimes" <
freebsd-rwg_at_pdx.rh.cn85.dnsmgr.net> wrote:

> On Mon, Dec 18, 2017 at 3:12 PM, Mark Millard <markmi_at_dsl-only.net> wrote:
>
> > Warner Losh imp at bsdimp.com wrote on
> > Mon Dec 18 20:29:45 UTC 2017 :
> >
> > > The specific thing we will stop doing is that in the absence of
> > > instructions to the contrary, we will no longer search for root on a
> > device
> > > other than the one the loader.efi came from.
> >
> > Warner Losh imp at bsdimp.com wrote on
> > Sun Dec 17 19:52:07 UTC 2017 :
> >
> > > In the coming months, we're looking at dropping boot1.efi and instead
> > > installing /boot/loader.efi onto the ESP (most likely as
> > > \efi\freebsd\loader.efi).
> >
> >
> > Combining the two statements would appear to have consequences
> > not obvious from the separate statements in isolation. Rewording
> > the first to substitute where loader.efi comes from based on
> > the second (if I interpret right):
> >
> > MISQUOTE
> > The specific thing we will stop doing is that in the absence of
> > instructions to the contrary, we will no longer search for root
> > on other than the device for the ESP used (which will hold
> > loader.efi).
> > END MISQUOTE
> >
>
> The specific thing we will stop doing is that in the absence of
> instructions to the contrary, we will no longer search for root on other
> than the device for the ESP used (which will hold now loader.efi as
> boot1.efi will shortly be eliminated).

Yes please, that is the correct behavior, our searching can lead to
problems, and as you have pointed out, often more problems than it
ever really fixed.

>
> Or the following pseudo-code with all the weird special cases removed for
> clarity
>
> load loader.efi from ESP
> if BootXXXX uefi variable holds a second path, use that for root/kernel
> otherwise if an override variable holds a kernel/root path, use that
> otherwise scan for a usable ZFS pool, use that if it exists
> otherwise use the same partition loader.efi was booted from for
root/kernel
> if it's usable
> otherwise use the first UFS partition on the ESP that's usable.

use the ACTIVE ufs partition, not the first, I can have more than 1 slice,
only 1 of them can be set active.  Do not use any ufs partitions if they
are not in active slices, it is possible to have 0 partitions set active.


Active is not a GPT concept. UEFI makes it hard to implement since there is
no good API to get and set the flags FreeBSD's gptboot uses to hack this
concept in. Active is done via BootOrder UEFI variable. Loader.efi and
boot.efi completely ignore this today. I have no plans on changing that.

>
> A partition is usable if /boot/loader.rc exists on that path.

A partition is usable if it is in an active slice, and ^above


Active isn't a got thong. So no.

Is there any fallback to skip loader and go direct to
/boot/kernel/kernel, back to /kernel any more?


You are thinking about this wrong. We are loader.efi, not boot2. This is
one of the big advantages of loading directly. We don't have the space
limitations that forced that design, so we should simplify.

Warner

> What is being deleted is one final step: "otherwise use the first UFS
> partition on any drive in a random order that's usable." which used to be
> at the end of the boot1.efi psuedo code. It's my belief that no such
> installations actually use this due to the random factor today (plug in a
> new USB drive and it might take over). If my belief is wrong, it's my
> belief that efibootmgr will solve it, and failing that, the fallback
> mechanism (for platforms that use u-boot + EFI where UEFI variables don't
> work) will allow the two or three people that are doing this today.
>
> Warner
> _______________________________________________
> freebsd-current_at_freebsd.org mailing list
> https://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe_at_freebsd.org"

--
Rod Grimes
rgrimes_at_freebsd.org