Re: [PATCH net-next 1/3] net: phy: Let drivers check Wake-on-LAN status

[Florian Fainelli]

On 5/8/23 12:17, Florian Fainelli wrote:
> On 5/8/23 12:02, Andrew Lunn wrote:
> > On Mon, May 08, 2023 at 11:43:07AM -0700, Florian Fainelli wrote:
> > > A few PHY drivers are currently attempting to not suspend the PHY when
> > > Wake-on-LAN is enabled, however that code is not currently executing at
> > > all due to an early check in phy_suspend().
> > >
> > > This prevents PHY drivers from making an appropriate decisions and put
> > > the hardware into a low power state if desired.
> > >
> > > In order to allow the PHY framework to always call into the PHY driver's
> > > ->suspend routine whether Wake-on-LAN is enabled or not, provide a
> > > phydev::wol_enabled boolean that tracks whether the PHY or the attached
> > > MAC has Wake-on-LAN enabled.
> > >
> > > If phydev::wol_enabled then the PHY shall not prevent its own
> > > Wake-on-LAN detection logic from working and shall not prevent the
> > > Ethernet MAC from receiving packets for matching.
> >
> > Hi Florian
> >
> > Did you look at using late_suspend for this? Then there would not be
> > any need to change all these drivers which are happy as they are.
>
> I did not know its existence until you mentioned it, this would require 
> plumbing all the way from the MDIO bus driver down to the PHY driver 
> level, which could be done, but for a single driver? The way Linux 
> suspends devices currently, and the fact that the interrupt is driven 
> level low, it does not give much room if at all for missing the 
> interrupt AFAICT.

I suppose that a middle ground could be to introduce a specific callback 
or flag that says: please call my suspend routine and I will do what's 
necessary.

> phy_suspend() is called both from the system suspend path, but also 
> whenever the PHY is unused, and this is a nice property because we do 
> not really need to differentiate these paths usually, that includes 
> Wake-on-LAN.
>
> Besides there are drivers like drivers/net/phy/at803x.c that wish to 
> isolate the PHY if WoL is enabled and which are currently not doing it 
> because this never gets called.

--
Florian