On Mon, Jan 12, 2026 at 12:43:50PM +0100, Niklas Cassel wrote:

Hello Koichiro,

On Sat, Jan 10, 2026 at 11:29:06PM +0900, Koichiro Den wrote:

(snip)

quoted

quoted

Your new feature (epc_features->subrange_mapping) in epc_features appears
to depend on epc_features->dynamic_inbound_mapping, so it is a shame that
we don't have a epc_features->dynamic_inbound_mapping bit, so that this new
feature could have depended on that bit.

	if (epf_bar->use_submap &&
	    !(epc_features->dynamic_inbound_mapping &&
	      epc_features->subrange_mapping))
		return -EINVAL;


I think adding some documentation is a good step.

Perhaps we should also introduce a epc_features->dynamic_inbound_mapping bit?
Since you are making DWC glue drivers return a mutable EPC features, we could
set this bit in the DWC driver after that commit. What do you think?

As you pointed out, support for dynamic_inbound_mapping is needed
independently of my series. Given that, it would make sense to handle it
either before this series, or to fold it into the next iteration (=v6) of
the series if that is preferred.

Please fold it into the next iteration (=v6).

It should be a one liner patch in the DWC driver, at least if you put it
after your "PCI: dwc: Allow glue drivers to return mutable EPC features"
patch.

Thank you for all your efforts on improving the endpoint framework.

I just submitted v6:
https://lore.kernel.org/all/20260113023715.3463724-1-den@valinux.co.jp/ (local)

Thank you very much for the review,
Koichiro

Kind regards,
Niklas