We trying to get our board which is using a Microchip Lan9662 chip connected to
the Jetson ORIN EQOS MAC. The connection between them is a rgmii phy-less design. Using a
6.18.3 Linux kernel we are able to use this design with a few modifications to the nvethernet driver
to accommodate the lack of a phy.
The probblem is when I make the same changes to the R36.5 nvethernet driver the EQOS MAC
never appears to come out of reset. The function call hw_poll_for_swr which polls BIT(0) in
the EQOS_DMA_BMR register and it never goes to 0. It does correctly go to 0 in the 6.18.3 implementation.
I have tried tracing the differences between what is being done in the R36.5 and the 6.18.3 implementations
and they appear to be doing the same initialization steps and the relevant section of the DTB’s appear to
be the same. So my question is what conditions should I be checking for that would keep the EQOS MAC from
coming out of its reset state?
Thank you in advance.