Re: [PATCH v2 4/5] usb: dwc3: Allow end transfer commands to be sent during soft disconnect

[Thinh Nguyen <Thinh.Nguyen@synopsys.com>]

On 7/21/2022, Wesley Cheng wrote:
> Hi Thinh,
>
> On 7/21/2022 5:04 PM, Wesley Cheng wrote:
>> Hi Thinh,
>>
>> On 7/21/2022 5:00 PM, Thinh Nguyen wrote:
>>> On 7/21/2022, Wesley Cheng wrote:
>>>> Hi Thinh,
>>>>
>>>> On 7/21/2022 3:20 PM, Thinh Nguyen wrote:
>>>>> On 7/21/2022, Wesley Cheng wrote:
>>>>>> Hi Thinh,
>>>>>>
>>>>>> On 7/21/2022 3:08 PM, Thinh Nguyen wrote:
>>>>>>> Hi Wesley,
>>>>>>>
>>>>>>> On 7/12/2022, Wesley Cheng wrote:
>>>>>>>> If soft disconnect is in progress, allow the endxfer command to be
>>>>>>>> sent,
>>>>>>>> without this, there is an issue where the stop active transfer 
>>>>>>>> call
>>>>>>>> (during pullup disable) wouldn't actually issue the endxfer 
>>>>>>>> command,
>>>>>>>> while clearing the DEP flag.
>>>>>>>>
>>>>>>>> In addition, if the DWC3_EP_DELAY_STOP flag was set before soft
>>>>>>>> disconnect
>>>>>>>> started (i.e. from the dequeue path), ensure that when the EP0
>>>>>>>> transaction
>>>>>>>> completes during soft disconnect, to issue the endxfer with the 
>>>>>>>> force
>>>>>>>> parameter set, as it does not expect a command complete event.
>>>>>>>>
>>>>>>>> Fixes: e4cf6580ac740 ("usb: dwc3: gadget: Wait for ep0 xfers to
>>>>>>>> complete during dequeue")
>>>>>>>> Suggested-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
>>>>>>>> Signed-off-by: Wesley Cheng <quic_wcheng@quicinc.com>
>>>>>>>> ---
>>>>>>>> Link:
>>>>>>>> https://urldefense.com/v3/__https://lore.kernel.org/linux-usb/1a1a5485-790e-79ce-f5a6-1e96d9b49a47@synopsys.com/__;!!A4F2R9G_pg!cXW2TlALYWnXNPg-NoMFUrQ8K1egEZiQizZ5CA1DOM1Gcw4tfOULy-_2eMGvL8pQPte5dScFON-0wxH2eXu8ijEQUbs$ 
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>      drivers/usb/dwc3/ep0.c    | 3 +--
>>>>>>>>      drivers/usb/dwc3/gadget.c | 5 ++++-
>>>>>>>>      2 files changed, 5 insertions(+), 3 deletions(-)
>>>>>>>>
>>>>>>>> diff --git a/drivers/usb/dwc3/ep0.c b/drivers/usb/dwc3/ep0.c
>>>>>>>> index 506ef717fdc0..5851b0e9db0a 100644
>>>>>>>> --- a/drivers/usb/dwc3/ep0.c
>>>>>>>> +++ b/drivers/usb/dwc3/ep0.c
>>>>>>>> @@ -290,8 +290,7 @@ void dwc3_ep0_out_start(struct dwc3 *dwc)
>>>>>>>>              if (!(dwc3_ep->flags & DWC3_EP_DELAY_STOP))
>>>>>>>>                  continue;
>>>>>>>>      -        dwc3_ep->flags &= ~DWC3_EP_DELAY_STOP;
>>>>>>>
>>>>>>> If we don't clear this flag here,
>>>>>>>
>>>>>>
>>>>>> This is why I added the dwc->connected argument to control the
>>>>>> "interrupt" argument.
>>>>>>
>>>>>>>> - dwc3_stop_active_transfer(dwc3_ep, true, true);
>>>>>>>> +        dwc3_stop_active_transfer(dwc3_ep, true, dwc->connected);
>>>>>>>>          }
>>>>>>>>      }
>>>>>>>>      diff --git a/drivers/usb/dwc3/gadget.c
>>>>>>>> b/drivers/usb/dwc3/gadget.c
>>>>>>>> index ee85b773e3fe..41b7007358de 100644
>>>>>>>> --- a/drivers/usb/dwc3/gadget.c
>>>>>>>> +++ b/drivers/usb/dwc3/gadget.c
>>>>>>>> @@ -1693,6 +1693,7 @@ static int 
>>>>>>>> __dwc3_stop_active_transfer(struct
>>>>>>>> dwc3_ep *dep, bool force, bool int
>>>>>>>>              dep->flags &= ~DWC3_EP_TRANSFER_STARTED;
>>>>>>>>          else if (!ret)
>>>>>>>>              dep->flags |= DWC3_EP_END_TRANSFER_PENDING;
>>>>>>>> +    dep->flags &= ~DWC3_EP_DELAY_STOP;
>>>>>>>>             return ret;
>>>>>>>>      }
>>>>>>>> @@ -3686,8 +3687,10 @@ void dwc3_stop_active_transfer(struct 
>>>>>>>> dwc3_ep
>>>>>>>> *dep, bool force,
>>>>>>>>          if (dep->number <= 1 && dwc->ep0state != EP0_DATA_PHASE)
>>>>>>>>              return;
>>>>>>>>      +    if (interrupt && (dep->flags & DWC3_EP_DELAY_STOP))
>>>>>>>> +        return;
>>>>>>>> +
>>>>>>>
>>>>>>> then it may not go through here. I think I made this same mistake
>>>>>>> in my
>>>>>>> previous suggestion.
>>>>>>>
>>>>>>
>>>>>> Since dwc->connected is set to FALSE before we call stop active
>>>>>> transfers, if we ever run into a situation where delayed stop is set
>>>>>> already, then it should go through.
>>>>>>
>>>>>
>>>>> Then the check for request dequeue that we did previously will not 
>>>>> work
>>>>> anymore.
>>>>>
>>>>
>>>> Could you help clarify?  The pullup refactor kind of shifted some of
>>>> the previous checks we had in the dequeue path, and combined with with
>>>> the stop active transfer checks.
>>>>
>>>> I think there was an issue w/ the patch I submitted though. the above
>>>> snippet should be replacing what is there:
>>>>
>>>> void dwc3_stop_active_transfer(struct dwc3_ep *dep, bool force,
>>>>      bool interrupt)
>>>> {
>>>> ...
>>>>      if (!(dep->flags & DWC3_EP_TRANSFER_STARTED) ||
>>>>          /* Following should be removed --- (dep->flags &
>>>> DWC3_EP_DELAY_STOP) || */
>>>>          (dep->flags & DWC3_EP_END_TRANSFER_PENDING))
>>>>          return;
>>>>
>>>
>>> Request dequeue can occur while the device is connected. The
>>> DWC3_EP_DELAY_STOP intention is to delay sending the End Transfer
>>> command until the SETUP stage is prepared. If we don't clear the flag
>>> before the flag check, then the End Transfer command will not go 
>>> through.
>>>
>>
>> Thanks, got it.  Understand what you mean now.  Let me think about 
>> how to go about it.  I probably don't see any issues as of now, 
>> because my test does the soft disconnect, which will do the stop 
>> active transfers forcefully.
>>
> What do you think about just removing the
>
> (dep->flags & DWC3_EP_DELAY_STOP)
>
> check?  In the end, as long as the conditions are satisfied (ie we 
> aren't in the middle of a SETUP transaction) then we don't care too 
> much about who called dwc3_stop_active_transfer().  We would still be 
> able to avoid issuing the endxfer here:
>
>     if (dwc->ep0state != EP0_SETUP_PHASE && !dwc->delayed_status) {
>         dep->flags |= DWC3_EP_DELAY_STOP;
>         return 0;
>     }

That's good. Then check for interrupt-on-completion there instead.

Thanks,
Thinh