Exposing just ETH2 through switch on a Verdin iMX8M Plus


I am currently designing a carrier board for the Verdin iMX8M Plus. We would like to have two Ethernet ports exposed on our product, and they’re likely going to be 100BASE-TX ports. We’d like them to be hardware switched to avoid overhead when passing traffic through (although the overhead seems to be small).

I was wondering if we’d run into any issues only using the RGMII interfaces with ETH_2, and leaving ETH_1 unconnected. Issues with the preinstalled TEZI or the precompiled U-Boot for instance, that would make it less convenient for manufacture/flashing purposes.

One side question - the design guide recommends using the same Gigabit PHY for the best software support - would this simply mean we can reuse the same DTS - or is there more to it?

Although you can leave ETH_1 unconnected and use only the RGMII interfaces with ETH_2, I’m confused about how this configuration aligns with your requirement to have two Ethernet ports exposed on your product.

Regarding the use of pre-compiled U-Boot and TEZI with such a configuration:

  • U-Boot: TFTP boot will not be available.
  • TEZI: You will not be able to load and flash an image from a network server. However, flashing images from an SD card or USB stick will still be possible, which is the most commonly used method for production programming.

We’d have a managed Ethernet switch IC on ETH_2, which would have any number of PHYs.