[PATCH] x86_64: add systemd-boot images

[Jonas Gorski]

On Tue, Dec 30, 2025 at 3:05 PM Jonas Lochmann <openwrt at jonaslochmann.de> wrote:
>
> Am Tue, Dec 30, 2025 at 02:01:21PM +0100, schrieb Paul Spooren:
> > Hi,
> >
> > >
> > > I read this twice, and still don't know what this patch is trying to
> > > accomplish. What problem does this solve, or what new feature does
> > > this allow? I am missing a sentence or two explaining what the patch
> > > does and why.
>
> Maybe you read it twice because I posted it twice? I am sorry about
> that, but the first version had a small mistake that blocked updating
> the disk UUID after updates.

No, I definitely meant I read the text twice to see if I missed something.

>
> > From my understanding it’s much simpler/smaller than GRUB, so maybe something inline with the efforts of OpenWrt?

Sure, but we shouldn't need to guess why something is added, that
should be spelt out in the commit message.

>
> Yes, that's the main motivation. Unlike grub, it only supports UEFI
> boot but this is the only boot method supported by recent x86 hardware.
> The legacy boot procedure is relevant only in virtual machines but
> because this implementation uses (like the OpenWrt grub integration)
> the fallback path, no specific values in nvram variables are required
> so that UEFI booting works without effort - like the legacy boot.

The motivation should be part of the commit message, and currently it
isn't. Like is this a potential replacement for something, or does it
something different? Are there any drawbacks? etc.

>
> > > And you can probably trim the history part, e.g. I don't think it
> > > matters much for understanding that it used to be called gummiboot, as
> > > entertaining the name is.
> >
> > I found this rather helpful, since much documentation/form posts out there point to Gummiboot.
>
> Another thing is that gummiboot is packaged in the openwrt "packages"
> repository/feed. I don't know for what it is used (if at all), because
> there are no images that I know of that use it.
>
> Writing this detail (gummiboot is now systemd-boot) into the commit
> message makes it easier to state a reason why gummiboot can be removed
> from the repository once this was merged. Actually, I tried using
> gummiboot first and it resulted in a black screen only. I assume it
> still works at older hardware.

Maybe then this should be also mentioned in the commit message?
Especially since there are 0 mentions of gummiboot in the core OpenWrt
repository. I tried to understand this commit from the perspective of
here, and since gummiboot isn't mentioned (and used) there, the
information seemed irrelevant for core OpenWrt.

Best regards,
Jonas