Re: [PATCH v7 1/2] module: add SCMI device table alias support

[Sudeep Holla]

On Sun, Sep 20, 2026 at 12:18:47PM +0200, Uwe Kleine-König wrote:
> On Sun, Sep 20, 2026 at 08:36:10AM +0100, Sudeep Holla wrote:

[...]

> > 
> > I am confused. I am asking to align with what's done in
> > Commit ad428f5811bd ("mod_devicetable.h: Split into per subsystem headers")
> > and you are disagreeing with that ?
> > 
> > I am not comfortable taking this patch unless there are other similar
> > examples that are agreed upon to break the pattern followed in the above
> > mentioned commit.
> > 
> > Uwe, please suggest.
> 
> I didn't start removing stuff from mod_devicetable.h yet, as changes
> there are expensive (regarding build dependencies). So the plan is to
> stop drivers consuming mod_devicetable.h and then drop it. (Not sure
> this is sensible as new users come in faster than I can fight them, but
> I didn't give up yet.)
> 
> So while adding new device_id structures to mod_devicetable.h is
> consistent with the existing practise, it still makes the situation
> worse, as every consumer of mod_devicetable.h then also depends on the
> new header for the new structure. That's why I think it's a good idea to
> deviate here, even if you're the first one doing so, to not worsen the
> already bad situation.
> 

It would be good if you provide your reviewed-by for mod_devicetable.h
and associated parts specifically here.

Also there is another patch under discussion [1], it would be good to have
similar suggestion there as well, so that there is less divergence in the
newly added code from now on.

> If it helps you, I can propose a patch changing modpost to not use that
> catch-(nearly)-all header and use the needed linux/device-id/*.h
> instead. I put this on my list for tomorrow.
> 

That's helpful too, but for the things that are being planned to be in
v7.4, they may not see that unless you push that as a fix.

-- 
Regards,
Sudeep


[1] https://lore.kernel.org/all/20260921053807.354802-2-aneesh.kumar@xxxxxxxxxx