git: e77cf2a4ab32 - main - Restore /boot/efi to mtree.

Nathan Whitehorn nwhitehorn at freebsd.org
Sat Mar 6 12:44:15 UTC 2021



On 3/6/21 4:43 AM, Mark Millard wrote:
> On 2021-Mar-6, at 01:01, Mark Millard <marklmi at yahoo.com> wrote:
>
>> On 2021-Mar-5, at 22:05, Mark Millard <marklmi at yahoo.com> wrote:
>>> Nathan Whitehorn nwhitehorn at FreeBSD.org wrote on
>>> Sat Mar 6 02:01:30 UTC 2021 :
>>>
>>>>    Restore /boot/efi to mtree.
>>>>
>>>>    Instead of whether /boot/efi exists, which it now always does, including
>>>>    on systems that don't and can't use EFI, use whether /boot/efi is
>>>>    present in fstab to signal to the installer that it is a valid ESP and
>>>>    should be configured. This has essentially the same semantics, but allows
>>>>    /boot/efi to be created unconditionally.
>>>>
>>> Sounds like the documentation about /etc/fstab content
>>> should indicate the special/reserved /boot/efi usage
>>> context, be that comments in initial default files or
>>> whatever.
>>>
>>> I wonder if anyone puts / at the end in an fstab: /boot/efi/
>>>
>> I tried using a trailing / in /etc/fstab and it is
>> one place were the notational variation is not
>> equivalent: I had to remove it.
>>
> FYI:
>
> Reviewing/adjusting my /etc/fstab files I notice that
> I have examples with things like:
>
> /dev/label/Rock64boot   /boot/efi       msdosfs rw,noatime,noauto       0 0
>
> #/dev/msdosfs/RPI4EFIFS /boot/efi       msdosfs rw,noatime,noauto       0 0
>
> Some might have a space after the #, shifting the
> /boot/efi to be at $3 ? Some /etc/fstab files have both
> types of /boot/efi lines (commented vs. uncommented),
> associated with root-file-system-media that I move
> between machines sometimes and toggle what is commented
> (changing what media ends up referenced).
>

This change *only* applies to boot partitions created by the installer, 
which have a well-known (to the installer) and controlled format.

As you note, though, there are a variety of ways someone could write 
this in their fstab, which makes this approach tricky for updating, and 
is why I originally preferred the simpler method of seeing whether the 
mountpoint existed.
-Nathan


More information about the dev-commits-src-main mailing list