Re: [PATCH v2 3/3] arm64: dts: rockchip: Add Luckfox Omni3576 Board support

[Andrew Lunn <andrew@lunn.ch>]

> Hi Andrew,
> Thank you for your valuable feedback and for pointing me toward
> investigating the PHY configuration further. After digging deeper into the
> MAE0621A-Q3C PHY (PHY ID 0x7b744412), I now understand why it performs
> reliably: the generic PHY driver relies on the GMAC to set RGMII delays
> (tx_delay=0x20, rx_delay=0x10), enabling a stable 1Gbps link for gmac0 in
> rgmii-rxid mode, while rgmii-id failed in my tests due to the driver’s lack
> of internal delay configuration. Given the critical role of networking for
> development, I’d like to retain the GMAC nodes in v3 using this setup, but
> I’d greatly appreciate your approval on whether the generic PHY driver is
> suitable in rgmii-rxid mode for now.

Device Tree describes the hardware. The hardware does not appear to
have the 2ns delay added by extra long clock lines, so your DT
description of the hardware using 'rgmii-rxid' is wrong.

If we where to accept this broken description now, it really will
break when the MAE0621A driver is merged and it correctly sets the
RGMII delays.

So sorry, no we cannot accept it.

For you own development work, you can merge the driver i pointed to
into your tree. It needs work, but it is not terrible. So it should
work for you.

> I’m eager to explore a Maxio-specific
> driver for mainline compatibility and would be grateful for any guidance on
> existing upstreaming efforts for this PHY.

This is the first time i've heard of this device, so i have no idea if
anybody else is working on it. I would suggest you reach out to the
author of that patch.

	Andrew