On Wed, Apr 13, 2022 at 11:32:42PM -0300, Luiz Angelo Daros de Luca wrote: > > While the code is OK, on second thought I think based on Andrew's points in the > > other subthread that we are better off without this patch. > > I agree, although the rtl8365mb name will make it harder for a > newcomer to find the driver. If it is made clear in the DT bindings, I think it's fine. If it works for Marvell switches, it will work for Realtek switches too. > > Is it too late to get rid of all those compatible strings from > dt-bindings? And rtl8367s from the driver? > We must add all supported devices to the doc as well, similar to mv88e6085. You can always try! I'm OK with those things in principle, but others might object due to ABI reasons. Don't get hung up on the vestigial "realtek,rtl8367s" compatible string though... while it's probably harmless to remove it, it's also relatively harmless to leave it there. Linux is full of such examples. Kind regards, Alvin
