On Thu, Jun 30, 2022 at 11:17:59AM +0300, Heikki Krogerus wrote: > On Wed, Jun 29, 2022 at 11:32:19PM +0000, Prashant Malani wrote: > > Introduce a retimer device class and associated functions that register > > and use retimer "switch" devices. These operate in a manner similar to > > the "mode-switch" and help configure retimers that exist between the > > Type-C connector and host controller(s). > > > > Type C ports can be linked to retimers using firmware node device > > references (again, in a manner similar to "mode-switch"). > > > > Signed-off-by: Prashant Malani <pmalani@xxxxxxxxxxxx> > > Cool! This looks really good to me. > > I'll add Mika here, just to keep him in the loop. Thunderbolt/USB4 can > control the same physical retimers over the SBU line. Right now there > is no conflict, but I think we want to later be able to use these > devices to upgrade the retimer firmware, and that is something that > the Thunderbolt/USB4 already does. So let's keep an eye on this. > > I wonder, would it make sense to later make the thunderbolt_retimer > devices also part of the device class that's introduced here? I think > that way it would be easier to later figure out which > thunderbolt_retimer and which retimer_switch represent the same > physical retimer. And perhaps it would also be more clear for the user > space to have a single device class for the retimers? I agree this makes sense.
