Re: [patch-v2.6.39 03/12] usb: musb: gadget: do not poke with gadget's list_head

[Pavol Kurina <pavol.kurina@emsys.de>]

Am 28.02.2011 09:43, schrieb Felipe Balbi:
> On Fri, Feb 25, 2011 at 07:41:47PM +0000, Pavol Kurina wrote:
>> Felipe Balbi<balbi<at>  ti.com>  writes:
>>> struct usb_request's list_head is supposed to be
>>> used only by gadget drivers, but musb is abusing
>>> that. Give struct musb_request its own list_head
>>> and prevent musb from poking into other driver's
>>> business.
>> Hi,
>>
>> I think, the patch misses to fix the usage of
>> "request->list" in musb_gadget_dequeue in
>> musb_gadget.c.
>>
>> I found out by having troubles with f_mass_storage.
>> It needs musb_gadget_dequeue to work properly...
>>
>> I backported the patch to android-2.6.35 kernel on a
>> omap4 system and also fixed the musb_gadget_dequeue
>> there so f_mass_storage seem to work at least there
>> now...
>>
>> Can you check this patch regarding some missing bits
>> please?
> Looks like we need some extra tests to be added to g_zero, I tested with
> g_zero and it didn't find that problem. Anyway, here's a patch which
> probably fixes what you need to be fixed. Would you test it for me ?
>
> diff --git a/drivers/usb/musb/musb_gadget.c b/drivers/usb/musb/musb_gadget.c
> index da8c93b..bf88a61 100644
> --- a/drivers/usb/musb/musb_gadget.c
> +++ b/drivers/usb/musb/musb_gadget.c
> @@ -1274,7 +1274,8 @@ cleanup:
>   static int musb_gadget_dequeue(struct usb_ep *ep, struct usb_request *request)
>   {
>          struct musb_ep          *musb_ep = to_musb_ep(ep);
> -       struct usb_request      *r;
> +       struct musb_request     req = to_musb_request(request);
> +       struct musb_request     *r;
>          unsigned long           flags;
>          int                     status = 0;
>          struct musb             *musb = musb_ep->musb;
> @@ -1285,10 +1286,10 @@ static int musb_gadget_dequeue(struct usb_ep *ep, struct usb_request *request)
>          spin_lock_irqsave(&musb->lock, flags);
>
>          list_for_each_entry(r,&musb_ep->req_list, list) {
> -               if (r == request)
> +               if (r == req)
>                          break;
>          }
> -       if (r != request) {
> +       if (r != req) {
>                  DBG(3, "request %p not queued to %s\n", request, ep->name);
>                  status = -EINVAL;
>                  goto done;
>

I've tested this on my omap4 system with the android-2.6.35 kernel. It 
seems like f_mass_storage and the f_adb
work fine.

I've added one more change to your patch.

diff --git a/drivers/usb/musb/musb_gadget.c b/drivers/usb/musb/musb_gadget.c
index a1e1d24..7877ccc 100755
--- a/drivers/usb/musb/musb_gadget.c
+++ b/drivers/usb/musb/musb_gadget.c
@@ -1250,7 +1250,8 @@ cleanup:
  static int musb_gadget_dequeue(struct usb_ep *ep, struct usb_request 
*request)
  {
         struct musb_ep          *musb_ep = to_musb_ep(ep);
-       struct usb_request      *r;
+       struct musb_request     *req = to_musb_request(request);
+       struct musb_request     *r;
         unsigned long           flags;
         int                     status = 0;
         struct musb             *musb = musb_ep->musb;
@@ -1261,17 +1262,17 @@ static int musb_gadget_dequeue(struct usb_ep 
*ep, struct usb_request *request)
         spin_lock_irqsave(&musb->lock, flags);

         list_for_each_entry(r, &musb_ep->req_list, list) {
-               if (r == request)
+               if (r == req)
                         break;
         }
-       if (r != request) {
+       if (r != req) {
                 DBG(3, "request %p not queued to %s\n", request, ep->name);
                 status = -EINVAL;
                 goto done;
         }

         /* if the hardware doesn't have the request, easy ... */
-       if (musb_ep->req_list.next != &request->list || musb_ep->busy)
+       if (musb_ep->req_list.next != &req->list || musb_ep->busy)
                 musb_g_giveback(musb_ep, request, -ECONNRESET);


Best regards,
Pavol.