On Tue, 2006-06-20 at 18:39 +1000, Benjamin Herrenschmidt wrote: > > The LMU node on my tree is below uni-n@f8000000/i2c@f8001000/i2c-bus@0 > > which is the i2c bus named "uni-n 0". (combined from uni-n@....@0 as far > > as I can tell). Maybe we can convince benh to add a property to the i2c > > busses that tell which path they have in the device tree. Ben? > > I suppose we could ... We may still want to use this on older kernels though. If I knew how to add arbitrary sysfs attributes I'd submit a patch right away, but I don't, and I don't have time to research it right now. Anyway, I checked the code, and it looks like my analysis wasn't quite correct, it seems to use snprintf(..., "%s %d", parentname, channel) but channel isn't assigned based on the i2c-bus' reg property but just an arbitrary number, so we'd have to probe both uni-n busses. Which is fine, still *much* better than probing all of them. johannes
Attachment:
signature.asc
Description: This is a digitally signed message part