On Wed, 30 May 2012, Bjørn Mork wrote: > Bjørn Mork <bjorn@xxxxxxx> writes: > > > I am not at all sure this is the correct way to fix this, > > but at least it solves my problem. > > Been sleeping on this, and I don't think the fix is entirely correct. > > We may have situations where two or more drivers intentionally match the > same device, assuming the serial driver probe will sort out the correct > driver. E.g. one driver doing wildcard pid matching for a single vid, > and another driver matching a single odd device from the same vendor. > The proposed patch will only allow one of those probes to run and will > therefore fail if the first called probe fails. > > But I don't think we want to move the driver matching after the serial > driver probing either. That would cause unnecessary probe calls from > the "wrong" drivers. > > So search_serial_device needs to do the match. Does anyone see a > problem with that? I guess noone here does, unless I express it in code > :-) So I'll go do that now. The new patch looks okay to me. But there's still a problem. Greg, as far as I can see, the generic driver is unique in that it uses _different_ id tables for its USB probing and its serial probing: generic_serial_ids is used for USB probing, generic_device_ids is used for serial probing. So it's still possible that the USB probe may match while the serial probe is rejected. In the 3.3 kernel such cases were rejected by generic_probe(), but that routine doesn't exist any more. Maybe generic_serial_ids should be removed entirely and replaced with generic_device_ids? Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
