On 4/02/21 4:58 pm, Bean Huo wrote: > On Wed, 2021-02-03 at 11:56 +0200, Adrian Hunter wrote: >>> >>> Hallo Adrian >> >> Hi Bean >> >> Thanks for the review >> >>> >>> Would you like sharing the advantage of this debugfs node comparing >>> to >>> sysfs node "attributes/exception_event_control(if it is writable)"? >> >> Primarily this is being done as a debug interface, but the user's >> exception >> events also need to be kept separate from the driver's ones. >> >>> what is the value of this? >> >> To be able to determine if the UFS device is being affected by >> exception events. >> >>> Also, now I can disable/enable UFS event over ufs-bsg. >> >> That will be overwritten by the driver when it updates the e.g. bkops >> control, or sometimes also suspend/resume. > > Hi Adrian > yes, I saw that, they are not tracked by driver. > > I have one question that why "exception_event_mask" cannot represent > the current QUERY_ATTR_IDN_EE_CONTROL value? only after writing it. It represents only the user's exception events (ee_usr_mask), not the driver's ones (ee_drv_mask) as well. ee_usr_mask is updated after successfully ensuring it is set on the device.
