[Macchiato] Latest EFI hangs writing efivars

[Marcin Wojtas]

Hi Ard,

Hm, my dtb update fixing (or rather working around EFI var access hang) was
done by adding clk entry to one of i2c nodes, sharing same clock. After
that I did not observe hang with efibootmgr and 'apt install
grub-efi-arm64'.

I think there's a big chance, that the installer does not initialize i2c
driver... I'll try if clk_ignore_unused helps for the installation hang
(it's the last stage of it...) if yes, I think more proper fix should be
applied in the variable driver then.

Best regards,
Marcin

2018-04-19 15:28 GMT+02:00 Ard Biesheuvel <ard.biesheuvel at linaro.org>:

> On 13 March 2018 at 10:19, Marcin Wojtas <mw at semihalf.com> wrote:
> > Hi Ard,
> >
> > 2018-03-13 9:49 GMT+01:00 Ard Biesheuvel <ard.biesheuvel at linaro.org>:
> >> On 12 March 2018 at 16:51, Marcin Wojtas <mw at semihalf.com> wrote:
> >>> Hi Riku,
> >>>
> >>> Thanks for your debug. It was a DT issue - I reproduced it and was
> >>> able to finalize installation of grub. Long story short, the SPI clock
> >>> remained not enabled (normally the mtd layer does that). It doesn't
> >>> happend with ACPI, where all clocks are enabled by default.
> >>>
> >>> I worked-around the issue and pushed to the github
> >>> (https://github.com/MarvellEmbeddedProcessors/
> edk2-open-platform/commit/b316449f24fdac9fc14c87d23f4680dfd789c7b8)
> >>> - please try newest marvell-armada-wip and let know.
> >>>
> >>
> >> Is the SPI controller that UEFI uses also exposed to the OS?
> >
> > No. I removed the reference from the DT.
>
> I am seeing the same issue with today's buster mini.iso installer: the
> install proceeds until the point where grub gets installed, and then
> it hangs. After a reboot, there's no grub in the ESP.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.einval.com/pipermail/macchiato/attachments/20180419/1c389e76/attachment.html>