On Wed, Apr 08, 2020 at 11:24:20PM +0200, Max Staudt wrote: > Thank you, that looks much nicer indeed. I have also tested it on my Amiga just to be sure - it works. Thank you. > However, no signature yet because the call stack is giving me a headache: > > icy_probe() > -> i2c_new_scanned_device() > -> i2c_new_client_device() > -> device_add_properties() > > And here we are in drivers/base/property.c looking at device_add_properties(): > > * WARNING: The callers should not use this function if it is known that there > * is no real firmware node associated with @dev! In that case the callers > * should create a software node and assign it to @dev directly. > > > Why is this warning there? It flies right in the face of what we're trying to achieve here. > > It was introduced in 2018 with commit caf35cd52242 . > > > So either the warning is superfluous, or i2c_new_client_device() should be creating a software fwnode, I guess? No and no. First one because the mechanism is added to have quirks, it must not replace the actual possibility to provide this via firmware (DT / ACPI). Second one, because software node API should have (has?) the same warning. +Cc Heikki. Heikki, am I correct? -- With Best Regards, Andy Shevchenko
