On Sun, Jul 29, 2018 at 04:03:41AM +0000, Bart Van Assche wrote: > On Thu, 2018-07-26 at 06:35 -0700, Tejun Heo wrote: > > Making removal asynchronous this way sometimes causes issues because > > whether the user sees the device released or not is racy. > > Hello Tejun, > > How could this change cause any user-visible changes? My understanding is > that any work queued with task_work_add() is executed before system call > execution leaves kernel context and returns back to user space context. Ah, you're right. This isn't workqueue. Hmm... yeah, needing to allocate sth in removal path is a bit icky but, it should be okay I guess. I *think* the kernfs active break/unbreak is likely gonna be cleaner tho. Thanks. -- tejun
