use_generic and usb probing
nick at van-laarhoven.org
Tue Apr 12 22:00:25 UTC 2011
> Well, I think that that's what probe priorities actually for.
> I also think that typically ivars should be set by a bus driver. So maybe it's
> not such a good idea to pass data from probe to attach via ivars in child drivers.
> But I could be mistaken about that.
> Practically speaking, you most likely don't have to worry about that for drivers
> that return BUS_PROBE_SPECIFIC (=0) from their probe methods. And there is only a
> few "generic" drivers that can be handled on a case by case basis.
Newbus priorities where specifically implemented on my request by Doug Rabson to cater for fallback attachments: usb.h (the old code) contained a list of priorities. ugen had the lowest priority and attached if no one else did. An example was a keyboard with a built-in mouse on one interface. It would be attached by either mouse or keyboard but not both. An additional driver could probide both devices. We ended up implementing that differently though: attachment to interfaces instead of devices.
A probe should not pass any information to the attach phase if it can at all avoid it. Or at least that is the assumption. If a driver needs info in both phases, it needs to regenerate it again. The fact that usb_probe is called again is a kludge, specifically for the description.
I guess that documenting this kludge and updating the comment to state this fact would be sufficient to solve the problem.
More information about the freebsd-current