On Thu, 2019-03-21 at 12:20 +-0800, jianchao.wang wrote: +AD4 Do you think we should provide more guarantee when setting sdev state through sysfs ? +AD4 +AD4 In the current implementation, store+AF8-state+AF8-field does nothing but just modify state. +AD4 +AD4 For example, when we set state to 'blocked', it cannot ensure the tasks that has escaped +AD4 the checking of state in scsi+AF8-queue+AF8-rq has quit, when we return from the sysfs. I would like to know which real workloads store +ACI-blocked+ACI in this sysfs attribute. If there are no such workloads, I would prefer no longer to allow +ACI-blocked+ACI being written in this sysfs attribute rather than making the implementation more complex. Thanks, Bart.
