czw., 28 lip 2022 o 11:16 Vladimir Oltean [off-list ref] napisał(a):

On Thu, Jul 28, 2022 at 08:52:04AM +0200, Marcin Wojtas wrote:

quoted

Yes, indeed. After recent update, I think we can assume the current
implementation of fwnode_find_parent_dev_match should work fine with
all existing cases.

What you should really be fixing is the commit message of patch 4,
that's what threw me off:

| As a preparation to switch the DSA subsystem from using
| of_find_net_device_by_node() to its more generic fwnode_
| equivalent, the port's device structure should be updated
| with its fwnode pointer, similarly to of_node - see analogous
| commit c4053ef32208 ("net: mvpp2: initialize port of_node pointer").
|
| This patch is required to prevent a regression before updating
| the DSA API on boards that connect the mvpp2 port to switch,
| such as Clearfog GT-8K or CN913x CEx7 Evaluation Board.

There's no regression to speak of. DSA didn't work with ACPI before, and
fwnode_find_net_device_by_node() still works with the plain dev->of_node
assignment.

There was a regression even for OF in v1, but after switching to
device_match_fwnode() it works indeed. Anyway patch v4 is imo useful,
I'll only reword the commit message.

Thanks,
Marcin