EJ Hsu wrote: > Oliver Neukum wrote: > > > > I just checked the code, and the reference count will be incremented > > > in scsi_scan_host(), precisely speaking, in scsi_autopm_get_host(). > > > So, I still think we need to manually add reference count of interface here. > > > Please correct me if there is anything wrong. > > > > AFAICT the generic power model will not suspend a parent while a child > > is active. The includes SCSI children of USB parents. So I think we > > are safe. Could you resubmit your patch without the PM counter > > manipulation and we can fix any issue that may exist in theory later? > > > > The child_count of usb device should be incremented in the scsi_scan_host(). > I can simulate this situation by forcibly enable the autosuspend of uas driver > and add a delay before scsi_scan_host() is called in asynchronous way. > > But for now, as uas driver does not support autosuspend, it is indeed safe to > remove the PM counter manipulation. If you have concern about it, it's OK to > me to re-submit it. > By the way, if autosuspend will be enabled in uas driver in the future, the PM counter manipulation should be added before scsi_add_host(), just as MSC driver did. Otherwise, the child_count of usb device might be accidentally decremented. (the rpm "idle" message will be propagated from scsi host to usb interface and usb device) Thanks, EJ --nvpublic
