Bootstd for sunxi

[Simon Glass]

Hi Tom,

On Wed, 24 Sept 2025 at 19:06, Tom Rini <trini at konsulko.com> wrote:
>
> On Wed, Sep 24, 2025 at 06:34:53PM -0600, Simon Glass wrote:
> > Hi Tom,
> >
> > On Wed, 24 Sept 2025 at 10:34, Tom Rini <trini at konsulko.com> wrote:
> > >
> > > On Wed, Sep 24, 2025 at 10:28:45AM -0600, Simon Glass wrote:
> > >
> > > > Hi,
> > > >
> > > > After the discussion on the call this week and Andre's suggestion of
> > > > putting FEL as the first global boothmeth, I took a look this morning.
> > > >
> > > > I created a new v7 series without the 'efi_loader: bootstd: Drop
> > > > bootmgr for sunxi' patch.
> > > >
> > > > It seems to run OK on a Linksprite_pcDuino3 but I'm not sure that is
> > > > enough. The FEL boothmeth is first, which is good. But Heinrich did a
> > > > series to reorder bootmgr which hasn't landed and specifically I had
> > > > concerns about this patch[1].
> > > >
> > > > Given that Heinrich can't make the calls, I'll talk it through with
> > > > him separately and reply on this thread. For now I don't see much
> > > > benefit in sending the new v7 series, but I can if desired.
> > >
> > > The biggest blocking part of the sunxi series became that it showcased
> > > the more generic problem with integration of EFI and standard boot [2]
> > > so yes, getting that resolved first is the next big blocker for trying
> > > to migrate more platforms. It was unclear to me then, and still unclear
> > > to me now what the value is in exposing more platforms to what I believe
> > > ends up being a less than optimal experience is compared with what they
> > > experience today.
> >
> > I'm not sure about the less-than-optimal, but I suspect this is the
> > bootmgr kicking in and checking the network. Heinrich has something
> > for that and I'll talk to him in a day or two.
>
> Well, that's not how [2] was summarized, but it's hard to know without
> anyone being able to try it.

My understanding is that Andre wanted to disable bootmgr for sunxi, or
have it look only at devices mentioned by its boot variables, but
Heinrich did not.

>
> > > But since no one can actually test things out and see what is or is not
> > > different with the switch, yes, you should post v7.
> >
> > OK, done! Something wrong with CI, but not sure what.
>
> It seems something is going on with docker.com.

OK

>
> > > [2]: https://patchwork.ozlabs.org/project/uboot/patch/20241113150938.1534931-4-sjg@chromium.org/#3418689


- Simon