Re: [PATCH net] phy: mscc: Fix when PTP clock is register and unregister

[Horatiu Vultur]

The 08/22/2025 15:58, Jakub Kicinski wrote:
> 
> On Fri, 22 Aug 2025 08:27:26 +0200 Horatiu Vultur wrote:
> > > > +void vsc8584_ptp_deinit(struct phy_device *phydev)
> > > > +{
> > > > +     switch (phydev->phy_id & phydev->drv->phy_id_mask) {
> > > > +     case PHY_ID_VSC8572:
> > > > +     case PHY_ID_VSC8574:
> > > > +     case PHY_ID_VSC8575:
> > > > +     case PHY_ID_VSC8582:
> > > > +     case PHY_ID_VSC8584:
> > > > +             return __vsc8584_deinit_ptp(phydev);
> > >
> > > void function has no return value. as well as it shouldn't return
> > > anything. I'm not quite sure why do you need __vsc8584_deinit_ptp()
> > > at all, I think everything can be coded inside vsc8584_ptp_deinit()
> >
> > I understand, I can update not to return anything.
> 
> It does look a little unnecessary.
> 
> > Regarding __vsc8584_deinit_ptp, I have created that function just to be
> > similar with the __vsc8584_init_ptp.
> 
> Alternatively you could only deinit if the clock pointer is valid,
> regardless of chip id. Unclear which one is the cleanest form, TBH.

I will update in the next version to check if the
vsc8531->ptp->ptp_clock is valid and in that case call
ptp_clock_unregister and skb_queue_purge.


-- 
/Horatiu