Re: [PATCH] char: misc: make misc_open() and misc_register() killable

[Greg KH]

On Thu, Jul 07, 2022 at 02:06:38PM +0900, Tetsuo Handa wrote:
> On 2022/07/06 20:04, Oliver Neukum wrote:
> > 
> > 
> > On 06.07.22 12:26, Tetsuo Handa wrote:
> >> On 2022/07/06 15:34, Greg KH wrote:
> >>> On Wed, Jul 06, 2022 at 03:21:15PM +0900, Tetsuo Handa wrote:
> >>>> How should we fix this problem?
> >>>
> >>> We can decrease the timeout in usb_stor_msg_common().  I imagine that if
> >>> that timeout is ever hit in this sequence, then all will recover, right?
> > 
> > Not really. The timeout there is supposed to come from the SCSI layer
> > in the general case.
> 
> I couldn't catch. usb_stor_msg_common() belongs to USB subsystem, doesn't it?
> 
> How does SCSI layer matter here?
> Does USB storage device shows up as if SCSI disk (e.g. /dev/sdb) relevant?

Yes.  USB storage is a very tiny wrapper around the whole SCSI protocol,
it is a SCSI disk for all all intensive purposes.

thanks,

greg k-h