On Fri, 25 Oct 2013, Hans de Goede wrote: > Hi, > > On 10/24/2013 06:32 PM, Alan Stern wrote: > > On Thu, 24 Oct 2013, Hans de Goede wrote: > > > >> uas devices have 2 alternative settings on their usb-storage interface, > >> one for usb-storage and one for uas. Using the uas driver is preferred, so if > >> the uas driver is enabled, and the device has an uas alt setting, don't bind. > > > > We need a mechanism for the user to force a device into usb-storage > > mode, even if the device claims to support UAS. It would not be at all > > surprising to find some devices whose UAS support is broken. > > Agreed, this is one of the reasons why I want to share the uas-detect code > between the uas and usb-storage drivers, so that we can have a single > place for a blacklist. > > So I think the best way to provide such a user override would be a kernel > module option. But that needs to belong to a single module, so I see 2 options; > 1) Turn uas-detect into a separate module > 2) Use the existing unusual_dev stuff for this and make uas depend on usb-storage > to get the unusual_dev (included user added extra quirks) from usb-storage > > 2. Makes the most sense to me, since usb-storage will get loaded anyways when > uas is loaded as they both match on the same usb class. > > If you agree that 2. is probably best then I can do an addon patch adding support > for a NO_UAS quirk which will cause the uas-detect code to bail with an error if > present. How about: 3) Create an unusual_uas.h file containing entries to be shared between usb-storage and uas. Put the NO_UAS quirk entries there. This way uas will know not to bind to the device (and usb-storage will know to bind), and you won't have to make uas depend on usb-storage. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
