Re: [PATCH net-next v8 5/7] net: pse-pd: check the PI vpwr supply before registering the controller

[Carlo Szelinsky]

On Mon, 05 Oct 2026 17:33:29 +0000, netdev-bot+sashiko@xxxxxxxxxx wrote:
> [Severity: Low]
> This isn't a bug, but the commit message only describes the future
> phylib notifier case. It leaves out a change that applies today.
...
> Could the commit message mention this?

Agreed. The v9 commit message says so: a PI whose provider appears late
now gets its power domain, which pse_register_pw_ds() used to skip.

> [Severity: Medium]
> Can this check pass while the PI regulator registration still defers its
> supply?

Yes. The check does not copy the regulator core exactly: a provider that
is registered, but whose device is not bound yet, still passes. That PI
then gets no power domain, and the v9 commit message says so. Closing
this fully would need an event from the regulator core when the provider
binds, which does not exist today. The window is small: the provider has
to be registered but not yet bound while the PSE driver probes.

> [Severity: Medium]
> Does this new -EPROBE_DEFER exit leak the pd692x0 power budget claimed on
> the vmain regulator?

Yes. That is an older bug in pd692x0: it already leaks the budget on any
probe failure after setup_pi_matrix(). Its fix goes to net: the budget
is now released through devres.

Carlo