Re: [PATCH net-next v8 6/7] of: property: do not let "pses" block a consumer's probe

[Carlo Szelinsky]

On Mon, 05 Oct 2026 17:33:31 +0000, netdev-bot+sashiko@xxxxxxxxxx wrote:
> [Severity: Low]
> Is this description accurate at this point in the series?

No, not at this commit: it describes the phy patch that comes later.
v9 folds this change into that phy patch, so the description and the
code now arrive together.

> [Severity: Medium]
> fwnode_mdio still does the PSE lookup at this point. Does that mean every
> deferral of the MDIO bus owner now runs a full PHY driver probe and
> remove, including the hardware setup done in probe?

Yes, with only this patch applied. Because v9 folds it into the phy
patch, which removes the PSE lookup from fwnode_mdio, no commit has this
probe and remove cycle any more.

> [Severity: Medium]
> Can this also make deferred probing loop forever while the PSE controller
> is absent?

Yes, I could reproduce the hang with only this patch applied. The same
fix applies: in v9 this change is part of the phy patch, and with that
patch the PHY no longer waits for the PSE controller at all, so there is
nothing to loop on. A boot test with a PSE controller that never binds
now boots at every commit.

Carlo