Re: [PATCH] hci-usb bugfix

[Marcel Holtmann]

Hi Oliver,

> > which results in the same as if we set NULL for the private pointer when
> > we claim the second interface. If this really happens then we have more
> > problems in the driver itself, because this case won't be handled in
> > either way. However I don't think that this will happen, because for
> 
> You can trigger it in software through usbfs.

I've never done this before. Can you show me how?

> > Bluetooth devices interface 0 and 1 can be seen as a unit. The only
> > reason that this was split over two interfaces, was that you don't have
> > to stop the bulk transfers when you change the altsetting on the second
> > interface.
> 
> Yes, but you should really stop using the second interface _before_
> returning returning from disconnect() for _that_ interface. You will
> operate correctly if the primary interface is disconnected first,
> but you cannot depend on that. If the secondary interface is
> disconnected first, you have a window where you illegally use an
> interface you no longer own.

You are absolutely right and this needs to be fixed, but this problem is
a different one than that my patch fixes. However this problem is now on
my todo list. Thanks for making me aware of.

Regards

Marcel


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/