On 03/28/2017 06:31 PM, Andrew Lunn wrote: > On Tue, Mar 28, 2017 at 05:33:18PM -0700, Florian Fainelli wrote: >> On 03/28/2017 02:45 PM, Andrew Lunn wrote: >>> Register the switch and its ports with devlink. >>> >>> Signed-off-by: Andrew Lunn <and...@lunn.ch> >> >> Reviewed-by: Florian Fainelli <f.faine...@gmail.com> >> Tested-by: Florian Fainelli <f.faine...@gmail.com> >> >> root@net-vm:~# devlink -p port >> mdio_bus/fixed-0:1f/0: type eth netdev lan1 >> mdio_bus/fixed-0:1f/1: type eth netdev lan2 >> mdio_bus/fixed-0:1f/2: type eth netdev lan3 >> mdio_bus/fixed-0:1f/3: type eth netdev lan4 >> mdio_bus/fixed-0:1f/5: type notset >> root@net-vm:~# devlink -p dev >> mdio_bus/fixed-0:1f > > Thanks Florian > > I have another patch somewhere which adds new types for CPU and DSA > ports. Once these 2 patches have landed, and the reports of broken > builds stop, i will dig it out. > > But my main priority at the moment is to play with dpipe and make sure > it can represent tables we have in DSA switches, before the Binary API > is frozen.
Yes, that is a very good point actually. Tangential question: we do not currently report to user space what kind of switch tagging protocol is used (DSA, eDSA etc.) I was going to add a sysfs attribute for that under the switch device's directory, but maybe devlink would be more appropriate for that? -- Florian
