* Felipe Balbi | 2011-08-29 14:35:29 [+0300]: >Hi, Hi, >On Mon, Aug 29, 2011 at 01:28:50PM +0200, Sebastian Andrzej Siewior wrote: >> * Felipe Balbi | 2011-08-29 11:51:34 [+0300]: >> So I'm for fixing the gadget. > >We could go that path, sure. But then this means that gadget driver will >be doing something different than requested on specification, just >because this one controller is quirky. IMHO this is just an implementation detail and it is not seen on wire / by the host. So the gadget *knows* it becomes only 5 bytes but it queues maxpacketsize anyway. On completion it should becomes only 5 bytes right? Anything else would be wrong. So it will become 5 bytes only. Should 6 bytes arrive then it can complain to the host. >A better approach might be to introduce a "quirks" field on struct >usb_gadget, as have gadget drivers check for that. Something like: > > if (gadget->quirks & GADGET_NEEDS_ALIGNED_OUT) > req->length = gadget->ep0->maxpacket; > else > req->length = len; This is bad because you have to ensure the buffer is large enough. For ep0 the composite framework allocated bMaxPacketSize0 anyway right? >but then, gadget driver needs to cope with whatever size we might need >to send. In case of most gadget drivers, it's very simple because they >will only receive data which is less than or equal to wMaxPacketSize, >but on gadget drivers such as DFU, we might need to be careful here. Sebastian -- 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
