On Thu Jul 9, 2020 at 5:36 PM CEST, Florian Fainelli wrote: > Certainly, the current interface clearly has deficiencies and since > mdio_device instances were introduced, we should have an interface to > debug those from user-space ala i2c-dev or spidev. > > Can you post the kernel code for review? Would you entertain having mdio
Certainly. I just linked to the repo to show how the userspace part would look in combination with the netlink interface, and to see if I was in ice-cube-in-hell territory or slightly better :) > as an user-space command being part of ethtool for instance (just to > ease the distribution)? If by "ethtool" you mean the project, then yes. But I think it should be a separate binary as ethtool is very interface centric. We might not want to restrict ourselves to a single tool either. A binary that can do basic register read/write seems like a good fit for shipping alongside ethtool. But a tool to read/write the LinkCrypt registers of a Marvell PHY, as an example, might be too specific and better managed in a separate repo. > -- > Florian
