RE: [EXT] Re: [PATCH v51 5/7] firmware: imx: adds miscdev

From: Pankaj Gupta

Date: Thu Nov 19 2026 - 12:47:40 EST


> >>>>>
> >>>>> It is in linux-next about v37, but build bot find many problems, I
> >>>>> dropped it and request to fix it. And sashiko is online recently
> >>>>> and also find many risk. I request to clean up it, and local AI
> >>>>> environment have difference with sashiko, so bump to v51.
> >>>>>
> >>>>> I put to linux-next again to make sure that pass all build test.
> >>>>>
> >>>>> Other two problems, need more time to check whole review history.
> >>>>> reply you later.
> >>>>
> >>>> So this is still in linux-next, but was not sent in the pull
> >>>> request for 7.4. It's rather confusing to me that maintainers add
> >>>> stuff to linux-next for build-testing and then leave it there,
> >>>> neither removing it, nor continuing the mainlining process.
> >>>>
> >>>> Frank, what is the plan?
> >>>>
> >>>> I'm currently rebasing my work for the NVMEM driver onto
> >>>> linux-next, but as there have been so many iterations and this
> >>>> whole thing took so long, I want to make sure that this is not
> >>>> another dead end where I'm wasting my time on.
> >>>
> >>> I need to ping here, again. What is the plan at NXP to move on with
> >>> this? This is blocking multiple features that depend on ELE firmware
> >>> support for i.MX9 and similar platforms from being upstreamed.
> >> To be more constructive, here is a proposal: let's get the
> >> kernel-internal part of this series merged first (patch 2-4, DT
> >> patches 6-7, doc patch 1 reduced to kernel side). I currently don't
> >> see any objections or issues with those.
> >>
> >> This would unblock drivers using the in-kernel API, such as the NVMEM
> >> driver and this is also the part I already tested when working on [1].
> >>
> >> The somewhat more controversial uAPI in patch 5 can then be further
> >> discussed until there is a consensus on it.
> >>
> >> Arnd, Frank, would that work for you?
> >
> > It needs at least one user for the in-kernel API.
>
> I can provide the NVMEM driver as user. I just need to rebase my series on the
> new reduced firmware driver patchset once that has been posted.
>
> Or if you want to have this all in one series, maybe someone at NXP
> (Pankaj?) can provide the firmware driver as proposed above in a public git tree
> somewhere or send it as RFC. Then I will provide the NVMEM patches on top of
> this for you to pick up and send in a combined series.
>
> Just let me know what works best for you.
>
> >>
> >> Pankaj, can you prepare such a reduced series?

Hi Frieder,

Thanks for the proposal, and sorry that you had to ping more than once.

Yes, I will prepare the reduced series as you describe:

- patches 2-4 (in-kernel interface only),
- DT patches 6-7,
- patch 1 cut down to the kernel side,
- patch 5 (the uAPI / character device) dropped from this series. It
will be posted separately so the uAPI discussion can continue
without blocking the in-kernel users.

It will include an in-kernel user from our side, OCOTP/NVMEM support
for i.MX8ULP, and I will make it available on a public branch so you
have a fixed base to rebase your i.MX93 NVMEM series [1] onto.

I will reply here with the links once it is out.

Thanks,
Pankaj

[1] https://patchwork.kernel.org/project/linux-arm-kernel/cover/20260727-upstreaming-next-20260609-imx-ocotp-ele-v4-0-1a515de33093@xxxxxxxxxx/

NXP Confidential