On Mon, Oct 03, 2022 at 04:51:32PM +0300, Ilias Apalodimas wrote:
> Hi Jassi,
>
> On Mon, 3 Oct 2022 at 16:40, Jassi Brar <jassisinghb...@gmail.com> wrote:
> >
> > On Mon, Oct 3, 2022 at 6:04 AM AKASHI Takahiro
> > <takahiro.aka...@linaro.org> wrote:
> >
> > > > diff --git a/board/socionext/developerbox/developerbox.c
> > > > b/board/socionext/developerbox/developerbox.c
> > > > index f5a5fe0121..a0db26eaf3 100644
> > > > --- a/board/socionext/developerbox/developerbox.c
> > > > +++ b/board/socionext/developerbox/developerbox.c
> > > > @@ -20,6 +20,13 @@
> > > >
> > > > #if CONFIG_IS_ENABLED(EFI_HAVE_CAPSULE_SUPPORT)
> > > > struct efi_fw_image fw_images[] = {
> > > > +#if defined(CONFIG_FWU_MULTI_BANK_UPDATE)
> > > > + {
> > > > + .image_type_id = DEVELOPERBOX_FIP_IMAGE_GUID,
> > > > + .fw_name = u"DEVELOPERBOX-FIP",
> > > > + .image_index = 1,
> > > > + },
> > > > +#else
> > >
> > > From curiosity, why do you want to use different capsule formats
> > > for multi-bank update and normal case?
> > >
> > normal/legacy layout has one image for each component - uboot, tfa and
> > optee, whereas the new layout contains everything in one fip image.
Yes, that is exactly what I understand here.
> > So I thought it would be better to make the image_index consistent by
> > making the fip's as 1.
>
> FWIW this does make a lot of sense. Since the SCP firmware is not
> included in the capsule and that SCP firmware is needed to transition
> from old -> new layout, I think we are better off having those in
> different GUIDs. On top of that those GUIDs can be used in LVFS if we
> ever decide to upload firmwares there.
>
> Not having discrete GUIDs means there's a chance to brick the board on
> old -> new update, unless the SCP is explicitly updated.
SCP? I don't care.
My question is why you use a single capsule (FIP) in A/B update while you use
three separate capsule files in normal case.
-Takahiro Akashi
> Cheers
> /Ilias
> >
> > cheers.
