On Tue, 2018-10-30 at 11:47 +0100, Marco Felsch wrote: > > > > > I disagree. That functionality is essential to the driver as submitted. > > You're right thats important for my use-case, but I got you, thats not a > hwmon related problem. If I understood the device-model correctly there > are absolut no problems to hot-unplug devices, this is always the case > if we unbind a driver from user-space. So why we can't handle it > directly in the kernel. Are there any concerns? If not can you give me > some hints to get the logic at thr right place. It is already common to have a userspace process that is signaled via a hwmon alarm attribute. Is there a reason one couldn't have a userspace process poll() on the alarm and unbind or remote devices via sysfs when it goes off? Seems like this might also give the opportunity to stop using the device in a more controlled way, instead of having it just disappear.
