[PATCH net-next 0/4] riscv: sophgo: Add ethernet support for SG2042

[Andrew Lunn]

On Wed, May 07, 2025 at 08:01:29PM +0800, Inochi Amaoto wrote:
> On Wed, May 07, 2025 at 02:10:48AM +0200, Andrew Lunn wrote:
> > On Wed, May 07, 2025 at 06:24:29AM +0800, Inochi Amaoto wrote:
> > > On Tue, May 06, 2025 at 02:03:18PM +0200, Andrew Lunn wrote:
> > > > On Tue, May 06, 2025 at 05:32:50PM +0800, Inochi Amaoto wrote:
> > > > > The ethernet controller of SG2042 is Synopsys DesignWare IP with
> > > > > tx clock. Add device id for it.
> > > > > 
> > > > > This patch can only be tested on a SG2042 x4 evb board, as pioneer
> > > > > does not expose this device.
> > > > 
> > > > Do you have a patch for this EVB board? Ideally there should be a user
> > > > added at the same time as support for a device.
> > > > 
> > > > 	Andrew
> > > 
> > > Yes, I have one for this device. And Han Gao told me that he will send
> > > the board patch for the evb board. So I only send the driver.
> > > And the fragment for the evb board is likes below, I think it is kind
> > > of trivial:
> > > 
> > > &gmac0 {
> > > 	phy-handle = <&phy0>;
> > > 	phy-mode = "rgmii-txid";
> > 
> > And this is why i ask, because this is broken. For more information,
> > please see:
> > 
> > https://patchwork.kernel.org/project/netdevbpf/patch/20250430-v6-15-rc3-net-rgmii-delays-v2-1-099ae651d5e5@lunn.ch/
> > 
> > 	Andrew
> 
> You are right, the right phy-mode is "rgmii-id", the delay is not
> added by the PCB. It seems to be better to ask for the vendor about
> the hardware design before copying params for vendor dts. Anyway,
> thanks for reviewing this.

Please do figure this out. Since you are adding a new compatible, you
have a bit more flexibility. If the MAC driver is doing something
wrong, you can change its behaviour based on this new compatible
without breaking other users of the driver using other compatibles.

	Andrew