>-----Original Message----- >From: Peter Chen [mailto:hzpeterchen@xxxxxxxxx] >Sent: Wednesday, November 23, 2011 11:02 AM >To: Li Yang-R58472 >Cc: Chen Peter-B29397; balbi@xxxxxx; gregkh@xxxxxxx; linux- >usb@xxxxxxxxxxxxxxx; linuxppc-dev@xxxxxxxxxxxxxxxx >Subject: Re: [PATCH] usb/fsl_udc: fix dequeuing a request in progress > >On Tue, Nov 22, 2011 at 7:48 PM, Peter Chen <hzpeterchen@xxxxxxxxx> wrote: >>>>It seems to can't get the correct qh pointer, you may still need to >>>>use below code to get it >>>> int i = ep_index(ep) * 2 + ep_is_in(ep); >>>> struct ep_queue_head *dQH = &ep->udc->ep_qh[i]; >>> >>> Thanks for trying. It will be much easier if we can dereference QH >from the ep structure. It is really strange that the ep->qh pointer is >not working somehow. >>> >> >> Seems only ep0-out's qh pointer is assigned at ep structure. >> At probe: >> /* setup udc->eps[] for ep0 */ >> struct_ep_setup(udc_controller, 0, "ep0", 0); >> >> >>> We have initialized it in struct_ep_setup(): >>> ep->qh = &udc->ep_qh[index]; >>> >>> Can you do me a favor on investigating why it's failing? >>> >Leo, I have debugged this issue at my board just now, the reason of >failure is we only have one ep struct for ep0, so when talking about ep0, >it always pointers to udc->ep[0]. So even we initialize the current qh >address for ep0in at probe, it still can't get the ep0in's qh address >through ep structure, as ep0 is always udc->ep[0]. Oops. I forgot the fact that we used a single ep structure to handle both IN and OUT ep0 because the gadget layer only knows about one ep0 structure. I guess currently we have no other way out unless the gadget layer do honor ep0 with direction. IMHO, it is a limitation to current gadget APIs that each udc driver need to take too much care of the protocol related stuff on ep0. Thanks, Leo -- 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
