Re: [PATCH 0/5] Shikra: Add DT support for ice, rng and qce

[Bartosz Golaszewski]

On Mon, 22 Jun 2026 20:19:09 +0200, Eric Biggers <ebiggers@xxxxxxxxxx> said:
> On Mon, Jun 22, 2026 at 04:25:12AM -0400, Bartosz Golaszewski wrote:
>> On Fri, 19 Jun 2026 18:45:06 +0200, Eric Biggers <ebiggers@xxxxxxxxxx> said:
>> > On Fri, Jun 19, 2026 at 02:13:28PM +0530, Kuldeep Singh wrote:
>> >> On 21-05-2026 18:47, Kuldeep Singh wrote:
>> >> > This patchseries attempt to enable sdhc-ice, rng and qce on shikra
>> >> > platform similar to other platforms.
>> >> >
>> >> > Previously, the 3 dt-bindigs/DT changes were sent as individual series
>> >> > and with feedback received, clubbed them together as all belong to same
>> >> > crypto subsystem.
>> >> >
>> >> > Here's link to old patchsets.
>> >> > QCE: https://lore.kernel.org/lkml/20260515-shikra_qcrypto-v1-0-80f07b345c29@xxxxxxxxxxxxxxxx/
>> >>
>> >> Hi Eric,
>> >>
>> >> As selftests issues for QCE are now fixed[1], so shikra series should be
>> >> good to proceed? as your concerns[2] are now addressed.
>> >> I am waiting for merge window to end and will send next rev post that.
>> >>
>> >> [1]
>> >> https://lore.kernel.org/linux-arm-msm/20260617-qce-fix-self-tests-v3-0-ecc2b4dedcfd@xxxxxxxxxxxxxxxx/
>> >> [2] https://lore.kernel.org/lkml/20260522024912.GC5937@quark/
>> >
>> > If you think that then it sounds like you need to read what I actually
>> > said.  The fixes are appreciated but don't change the big picture.
>> >
>> > - Eric
>> >
>>
>> Eric,
>>
>> I mentioned it in another thread[1]. This series is not adding any new features
>> to the QCE driver, it describes the hardware. The SoC *does have* this IP and
>> no matter the state of the support in the kernel, there's nothing wrong in
>> extending the existing bindings and adding new dts nodes.
>>
>> Thanks,
>> Bartosz
>
> It enables the driver on a new platform.  So it very much has a real
> effect.  It's not just adding a hardware description without a user.
>

The driver can be disabled by not building it. The hardware description in dts
must reflect the reality. This is what dt-bindings and devicetree sources do:
they *describe* the hardware. It's up to the systems integrator to build the
relevant driver or not.

Let's wait for DT maintainers to respond to v2 once it's out.

Bartosz