Re: [Bug 8094] ipaq oops on connecting "Vodafone VPA-II"
Alan Stern <[email protected]>
| Newsgroups | gmane.linux.usb.devel |
|---|---|
| Message-ID | <[email protected]> |
On Tue, 18 Sep 2007, Oliver Neukum wrote: > > The reporter has supplied the lsusb output; the CDC Data interface does > > have both an IN and an OUT endpoint. But the other interface isn't CDC > > Control, and I don't know how that will affect the driver. > > It'll crash the way it does. The ipaq driver needs both a bulk in and a bulk > out endpoint. It assumes URBs for them are available, as the usb serial > layer does guarantee, if the endpoints for them are there. > The ipaq driver specifies that it wants only interfaces with certain > attributes: > > .num_interrupt_in = NUM_DONT_CARE, > .num_bulk_in = 1, > .num_bulk_out = 1, > > But I cannot find out where usb_serial_probe() uses that information. > Is there something fishy in the generic usb serial support? It looks like a bug. Greg, isn't search_serial_device() or get_iface_id() supposed to check the number of actual endpoints against the driver's requirements? > I seriously question whether this device should be bound to the ipaq > driver at all. Interface 0 shouldn't, but maybe interface 1 should. However the oops seems to occur during the bind to interface 0, thanks to the bug in the serial core. Alan Stern ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ [email protected] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel