Re: "Loader needs to be updated" (azure guest context)

From: Edward Sanford Sutton, III <mirror176_at_hotmail.com>
Date: Fri, 03 Jan 2025 17:15:11 UTC
On 12/30/24 10:59, Warner Losh wrote:
> On Mon, Dec 30, 2024, 8:45 AM void <void@f-m.fm> wrote:
> 
>> Hi,
>>
>> context is
>>
>> freebsd amd64 vm on Azure (initially installed via azure marketplace
>> several
>> years ago (12-releng)), and has been upgraded as updates became available
>> with freebsd-update over the years all the way up to 13.4-p1 now.
>> System updates/upgrades have *always* been managed with freebsd-update.
>>
>> Latest upgrade from 13.3-p6 to 13.4-p1 shows "loader needs to be updated"
>> in the beastie menu in the console now. This is new.
>>
>> # zfs --version
>> zfs-2.1.14-FreeBSD_gd99134be8
>> zfs-kmod-2.1.15-FreeBSD_gd99134be8
>>
>> (perhaps side issue - different version numbers, same -FreeBSD_gd99134be8
>> ?!!)
>>
>> This system is *not* root-on-zfs. There is zfs, but it's data on
>> a non-boot virtual disk.
>>
>> "zpool status" invites me to upgrade the pool. I've not done this (have
>> never done it with this vm, either), and don't want to unless I'm
>> absolutely certain upgrading the pool won't break everything.
>>
>> I note from a (similar, but different context) thread last September
>>
>> https://lists.freebsd.org/archives/freebsd-current/2024-September/006378.html
>> that FreeBSD uses "the guest's boot loader and the host's /boot/lua files"
>> but I'm clueless how this would apply in an amd64 context with Azure
>> as the host.
>>
>> What do i have to do? Also, is the warning safe to ignore in this context?
>>
> 
> In this context, it's a known false positive. It's too risky to fix in a pX
> for 13.4, so will be in 13.5 since the fix is already in stable/13. It's
> just cosmetic, there's no bug it exposes. And even if you hadn't actually
> updated the loader, it will fail safe for loaders installed from FreeBSD 11
> and newer (though not relevant to your use case). It should be in the
> release notes for 13.4 as a known issue, but the process for post release
> revision is murky at best.

Why is there no errata entry for it?

> The host vs guest issue you highlighted is a different thing and applies
> only to bhyve.
> 
> Warner
> 
>>
>