On Fri, Aug 30, 2019 at 01:43:36PM -0400, Alan Stern wrote: > > I could instead add a new sysfs node 'force_bind' to the driver core, > > that would work like 'bind' except for skipping the > > driver_match_device() call entirely and forcing a probe(). Do you > > think that would be acceptable? Or is that too big of a hammer to make > > available for all drivers in Linux? Maybe if I do the same thing but > > only for usb drivers, or even only for the usb-storage driver > > specifically, would that be acceptable? > > This is a question for Greg. The problem is that there may be drivers > which can't handle being probed for devices they don't match. > > Still, we ought to have a mechanism for doing manual but not automatic > matches. > > Greg, any thoughts? This should work just fine today. Add a new device id to the "new_id" file and then tell the driver to bind. That's pretty much the same as a "force_bind", right? thanks, greg k-h
