On Sun, 12 Jul 2020 at 13:26, Russell King - ARM Linux admin <li...@armlinux.org.uk> wrote: > If you have machine A with address 192.168.2.1/24 on lan1 and machine B > with address 192.168.2.2/24 on lan2, then they should be able to ping > each other - the packet flow will be through the DSA switch without > involving the CPU.
I can confirm that bridging the lan ports and connecting two physical machines does allow me to ping between the two machines, with and without the commit reverted. > If it is a port issue, that should help pinpoint it - if it's a problem > with the CPU port configuration, then ethtool can't read those registers > (and the only way to get them is to apply some debugfs patch that was > refused from being merged into mainline.) It looks like this is the next step then. Do you have a link to the patch? I'll give it a go. What outputs would be useful for me to provide? > > I'm now less confident about my git bisect... > That seems really weird! And ended up being unrelated to the current issue; I had other things connected to the switch while I was testing it and I think STP was getting involved. Either way, I've isolated it now and have confirmed that 5.8-rc4 with that one commit reverted is working perfectly. Sorry for the confusion. Martin
