[PATCH 4/5] pxe_utils: Support a backup for localboot

[Simon Glass]

On Wed, 8 Jan 2025 at 10:31, Tom Rini <trini at konsulko.com> wrote:
>
> On Wed, Jan 08, 2025 at 10:04:07AM -0700, Simon Glass wrote:
> > Hi Tom,
> >
> > On Fri, 20 Dec 2024 at 13:48, Tom Rini <trini at konsulko.com> wrote:
> > >
> > > On Fri, Dec 20, 2024 at 10:37:34AM -0700, Simon Glass wrote:
> > > > Hi Tom,
> > > >
> > > > On Fri, 20 Dec 2024 at 10:23, Tom Rini <trini at konsulko.com> wrote:
> > > > >
> > > > > On Fri, Dec 20, 2024 at 10:18:17AM -0700, Simon Glass wrote:
> > > > > > Hi Tom,
> > > > > >
> > > > > > On Fri, 20 Dec 2024 at 07:56, Tom Rini <trini at konsulko.com> wrote:
> > > > > > >
> > > > > > > On Thu, Dec 19, 2024 at 09:01:19PM -0700, Simon Glass wrote:
> > > > > > >
> > > > > > > > The current localboot implementation assumes that a 'localcmd'
> > > > > > > > environment variable is provided, with the instructions to follow. This
> > > > > > > > may not be included, so provide a fallback in that case.
> > > > > > > >
> > > > > > > > Add a test image and test as well.
> > > > > > > >
> > > > > > > > Signed-off-by: Simon Glass <sjg at chromium.org>
> > > > > > >
> > > > > > > This is a pretty niche feature, I had to dig around a bit to see how
> > > > > > > it's specified elsewhere (not really) and how it's used. And I think
> > > > > > > that based on how it's used, making up a bootcmd when localcmd is
> > > > > > > undefined is the wrong approach. It's the hook for "run what I defined
> > > > > > > in the environment", so if not set erroring back out seems appropriate.
> > > > > >
> > > > > > Yes, but unfortunately it seems to be used and we should support it.
> > > > > > The problem with scripts is that we don't know the boot device, etc,
> > > > > > so it needs to be integrated into PXE. I did consider putting
> > > > > > something in bootstd, but we only find out that it is requesting a
> > > > > > localboot when actually running the extlinux bootmeth, so this is what
> > > > > > I came up with.
> > > > > >
> > > > > > It will be interesting to see if any other cases come up.
> > > > >
> > > > > It would be helpful at this point I think if you can point to how the
> > > > > code for handling this case (the LOCALBOOT keyword followed by an
> > > > > integer) in other projects so that we can be compliant with what's
> > > > > expected, even if it's poorly documented.
> > > >
> > > > Yes, it seems very hard to find anything at all.
> > > >
> > > > The implementation in syslinux uses the BIOS to jump to a boot sector:
> > > >
> > > > https://repo.or.cz/syslinux.git/blob/HEAD:/core/bios/localboot.c
> > > >
> > > > I'm really not sure how it is supposed to work with a filesystem.
> > > >
> > > > [1] is a usage of it for rpi
> > >
> > > Yes, I did some searching on extlinux.conf and localboot and got a few
> > > hits on people using localcmd for when they need to run their own
> > > special thing, for various reasons, for example:
> > > https://stackoverflow.com/questions/54258960/applying-fdt-overlay-with-u-boot-using-extlinux-conf
> > >
> > > > But other than that, even Gemini doesn't seems to have much idea.
> > >
> > > I'm only half surprised it didn't make something up for it.
> >
> > I've been quite amazed at its capabilities recently. I can mostly get
> > good answers to quite complex questions and it is very fast now. I'm
> > just waiting for when I can start using it for development.
>
> Oh no, please no. We don't have a policy yet, but no "AI" created code,
> please.

lol me too :-)

>
> > I wonder who maintains the extlinux stuff. Do you have any idea? It
> > would be nice to move it into a git tree and generate the spec from
> > there.
>
> I would say the syslinux project should be the authority unless they say
> otherwise.

Yes I've tried irc and will see if I get any ideas. Will send an email too.

Regards,
Simon