enami tsugutomo wrote:

> - And I finanly found that if I defer actual device close operation
>   until detach time (see the diff attached below.  the diff includes
>   some garbage tho), the scanner works fine (with both uhci and ehci,
>   not tested with ohci controller yet).  But I'm not sure if this
>   should be integrated in main tree or not.  Especially if there is
>   another fix...

I suspect that the device is broken (by design).  To transfer data
the alternating bit protocol is used over the wire, so the host and
the device have to be in sync.  To sync them, the USB stack performs
a CLEAR_ENDPOINT_HALT which is supposed to set the data
toggle to 0.  If the device doesn't do that, something like your problem
would occur.

I'll make a fix to the driver very soon.

    -- Lennart