On Wed, 22 Apr 2020, Alan Stern wrote: > > Jiri, you should know: Are HID drivers supposed to work okay when the > > ->close callback is issued after (or concurrently with) the ->stop > > callback? > > No response. Sorry, I've been a bit swamped recently. Thanks a lot for taking care of this. > I'll assume that strange callback orderings should be supported. Let's > see if the patch below fixes the race in usbhid. Unfortunately I don't believe the supportability of this is fully defined. I have tried to quickly go over the few major drivers and didn't find anything relying various orderings, but I might have easily missed some case. So unless we have a programatic way to check it, the patch you created for mutual exclusion is a good bandaid I believe. Thanks again Alan, I'll push it to Linus for 5.7. -- Jiri Kosina SUSE Labs
