Re: [PATCH V4 1/3] iommu/arm-smmu-v3: Add device-tree support for CMDQV driver

[Nicolin Chen]

On Wed, Dec 10, 2025 at 12:39:49PM +0000, Robin Murphy wrote:
> On 2025-12-10 5:19 am, Ashish Mhetre wrote:
> [...]
> > Hi Robin, Nic,
> > We removed ACPI dependency in Kconfig but driver still depends
> > on ACPI for these functions. I will be protecting ACPIspecific
> > tegra241-cmdqv code under CONFIG_ACPI similar to what is done
> > in arm-smmu-v3 driver. Is this the correct thing to do or do you
> > have any other suggestions?
> 
> Yes, when I commented that "depends on ACPI || OF" was functionally the same
> as just removing "depends on ACPI", I didn't mean to suggest that wasn't
> necessarily a genuine dependency still.
> 
> I guess if you can wrap the ACPI-specific functions in a single #ifdef block
> that's reasonable, however I do now wonder whether things couldn't be
> factored out a bit more - if it's a standard DSDT/SSDT namespace device,
> shouldn't there also be a corresponding platform device created for it,
> which we could look up instead of delving directly into the _CRS of the ACPI
> node itself? (not sure off-hand if there's a straightforward inverse of
> ACPI_COMPANION()...)

Ah, I did a quick tracer at acpi_create_platform_device(). And I
do see platform devices being created. So, we could have made it
ACPI-independent from the beginning, as you expected :)

Ashish, let's attach the following patch in your series: