RE: [PATCH v6 1/2] dt-bindings: net: Add FSD EQoS device tree bindings

["Swathi K S" <swathi.ks@samsung.com>]

> -----Original Message-----
> From: Andrew Lunn <andrew@lunn.ch>
> Sent: 14 February 2025 18:40
> To: Swathi K S <swathi.ks@samsung.com>
> Cc: krzk+dt@kernel.org; andrew+netdev@lunn.ch; davem@davemloft.net;
> edumazet@google.com; kuba@kernel.org; pabeni@redhat.com;
> robh@kernel.org; conor+dt@kernel.org; richardcochran@gmail.com;
> mcoquelin.stm32@gmail.com; alexandre.torgue@foss.st.com;
> rmk+kernel@armlinux.org.uk; netdev@vger.kernel.org;
> devicetree@vger.kernel.org; linux-stm32@st-md-mailman.stormreply.com;
> linux-arm-kernel@lists.infradead.org; linux-kernel@vger.kernel.org;
'Pankaj
> Dubey' <pankaj.dubey@samsung.com>
> Subject: Re: [PATCH v6 1/2] dt-bindings: net: Add FSD EQoS device tree
> bindings
> 
> On Fri, Feb 14, 2025 at 10:47:39AM +0530, Swathi K S wrote:
> >
> >
> > > -----Original Message-----
> > > From: Andrew Lunn <andrew@lunn.ch>
> > > Sent: 14 February 2025 05:50
> > > To: Swathi K S <swathi.ks@samsung.com>
> > > Cc: krzk+dt@kernel.org; andrew+netdev@lunn.ch;
> davem@davemloft.net;
> > > edumazet@google.com; kuba@kernel.org; pabeni@redhat.com;
> > > robh@kernel.org; conor+dt@kernel.org; richardcochran@gmail.com;
> > > mcoquelin.stm32@gmail.com; alexandre.torgue@foss.st.com;
> > > rmk+kernel@armlinux.org.uk; netdev@vger.kernel.org;
> > > devicetree@vger.kernel.org;
> > > linux-stm32@st-md-mailman.stormreply.com;
> > > linux-arm-kernel@lists.infradead.org; linux-kernel@vger.kernel.org
> > > Subject: Re: [PATCH v6 1/2] dt-bindings: net: Add FSD EQoS device
> > > tree bindings
> > >
> > > > +  phy-mode:
> > > > +    enum:
> > > > +      - rgmii-id
> > >
> > > phy-mode is normally a board property, in the .dts file, since the
> > > board
> > might
> > > decide to have extra long clock lines and so want 'rgmii'.
> > >
> > > The only reason i can think of putting rgmii-id here is if the MAC
> > > only supports 'rgmii-id', it is impossible to make it not add delays.
> > > If that is true, a comment would be good.
> >
> >
> > Hi Andrew,
> > Thanks for reviewing.
> > I think we already discussed this part some time back here [1] [1] :
> > https://patchwork.kernel.org/project/linux-arm-kernel/patch/2023081411
> > 2539.7
> > 0453-2-sriranjani.p@samsung.com/#25879995
> > Please do let me know if there is any other concern on this.
> 
> We partially discussed this in this thread.
> 
> As i said, what value you need here depends on the board design. The PCB
> could provide the 2ns delay, in which case, 'rgmii' would be the correct
value
> to have in the board .dts file. Hence the binding should not restrict the
value
> of phy-mode to just rgmii-id. All 4 rmgii values should be accepted.
> 
> The only reason you would force only rgmii-id is if the MAC/PHY pair
cannot
> do anything else. If that really is true, i would expect a comment in the
> binding, and the MAC driver to return -EINVAL for anything but rgmii-id.

Hi Andrew, 
What you said is right. Generally, PCB provides internal delay.
But in this case, due to customer request, the delay was added into SoC.

The following doc on rgmii says that "Devices which implement internal delay
shall be referred to as RGMII-ID. 
Devices may offer an option to operate with/without internal delay and still
remain compliant with this spec"
https://community.nxp.com/pwmxy87654/attachments/pwmxy87654/imx-processors/2
0655/1/RGMIIv2_0_final_hp.pdf

Also, the driver is in such a way that it handles all four rgmii in the same
way. 

Considering this, could you let us know what will be the right approach to
take in this case?

-Swathi

> 
> 	Andrew