On Fri, 1 Dec 2023 11:41:14 +0100 Johannes Berg wrote: > So I had put this aside for a while, but really got annoyed by all > the test failures now ... thinking about this again I basically now > arrived at a variant of solution #3 previously outlined, and I've > kind of convinced myself that userspace should always get an event > with a new carrier_up_count as it does today. Would it work if we exposed "linkwatch is pending" / "link is transitioning" bit to user space? Even crazier, would it help if we had rtnl_getlink() run linkwatch for the target link if linkwatch is pending?