* Lokesh Vutla <lokeshvutla@xxxxxx> [190214 17:32]: > Hi Tony, > Please do not snip the on going discussion. > > On 2/14/2019 9:11 PM, Tony Lindgren wrote: > > * Lokesh Vutla <lokeshvutla@xxxxxx> [190214 08:39]: > >> IMHO, device ids are something which can be used in DT. There are many other > >> things like the interrupt ranges etc.. which are discoverable from sysfw and we > >> are implementing it. > > > > We need to describe hardware in the device tree, not firmware. > > > > If you have something discoverable from the firmware, you should > > have the device driver query it from sysfw based on a hardware > > property, not based on some invented enumeration in the firmware. > > Yes we are already querying sysfw for all the irq ranges that can be > discoverable. The topic of discussion here is about the parent interrupt > controller id. I am not sure how you are expecting an id be discoverable > from system firmware especially with a name. Well names are quite standard in dts (but should be used with the phandle + offset). Think for example interrupt-names and reg-names :) > > If there is some device to firmware translation needed, hide that > > into the device driver and keep it out of the device tree. > > If preferred this can be moved to of_match_data attached to each > compatible property. Then for each SoC a new compatible needs to be created. Hiding the ID into the device driver and compatible property makes sense to me if the id is based on SoC + firmware. But I'd rather have a proper hardware based phandle + index type mapping in the dts if possible though. What does this id really consist of? Regards, Tony
