On Tue, 8 Mar 2005, Alan Stern wrote: > How would you tell the core or a bus driver not to add/delete devices or > bind/unbind drivers? The logical answer is by locking a semaphore or the > equivalent. This would be tantamount to reintroducing the bus rwsem! No, the rwsem is a bad idea. That locks the entire bus and prevents anything from happening to any device. That's not good. It's best to have a per-device lock for that type of exclusion. And, you really do want to exclude a device power transition from happening while a driver is binding/releasing from a device. Ultimately, that would be a driver-based operation, and the lock would exist in a driver-allocated, per-device structure. But, we could easily add a semaphore to struct device that we take around ->probe(), ->release() and power state transition calls. [ Actually, I recently added a spinlock to struct device for a similar purpose, so it'll be trivial to change it to a semaphore, which we need for this purpose, since ->probe() or ->suspend() could easily sleep on memory allocations. ] > Can anyone point out a place where the kernel has to iterate over the > entire driver-model tree, other than the suspend/resume routines? Device shutdown, which is a related situation. They're the only things that need to operate on the entire device space. Pat
