Re: [RFC PATCH v2 3/3] usb: dwc3: Issue core soft reset before enabling run/stop

[Wesley Cheng]

Hi Daehwan,

On 2/16/2022 9:52 PM, Jung Daehwan wrote:
> Hi wesley,
> 
> On Tue, Feb 15, 2022 at 04:08:35PM -0800, Wesley Cheng wrote:
>> It is recommended by the Synopsis databook to issue a DCTL.CSftReset
>> when reconnecting from a device-initiated disconnect routine.  This
>> resolves issues with enumeration during fast composition switching
>> cases, which result in an unknown device on the host.
>>
>> Signed-off-by: Wesley Cheng <quic_wcheng@xxxxxxxxxxx>
>> ---
>>  drivers/usb/dwc3/core.c   |  4 +---
>>  drivers/usb/dwc3/core.h   |  2 ++
>>  drivers/usb/dwc3/gadget.c | 11 +++++++++++
>>  3 files changed, 14 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
>> index 18adddfba3da..02d10e1cb774 100644
>> --- a/drivers/usb/dwc3/core.c
>> +++ b/drivers/usb/dwc3/core.c
>> @@ -115,8 +115,6 @@ void dwc3_set_prtcap(struct dwc3 *dwc, u32 mode)
>>  	dwc->current_dr_role = mode;
>>  }
>>  
>> -static int dwc3_core_soft_reset(struct dwc3 *dwc);
>> -
>>  static void __dwc3_set_mode(struct work_struct *work)
>>  {
>>  	struct dwc3 *dwc = work_to_dwc(work);
>> @@ -261,7 +259,7 @@ u32 dwc3_core_fifo_space(struct dwc3_ep *dep, u8 type)
>>   * dwc3_core_soft_reset - Issues core soft reset and PHY reset
>>   * @dwc: pointer to our context structure
>>   */
>> -static int dwc3_core_soft_reset(struct dwc3 *dwc)
>> +int dwc3_core_soft_reset(struct dwc3 *dwc)
>>  {
>>  	u32		reg;
>>  	int		retries = 1000;
>> diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h
>> index 00348d6d479b..b27ad8dad317 100644
>> --- a/drivers/usb/dwc3/core.h
>> +++ b/drivers/usb/dwc3/core.h
>> @@ -1532,6 +1532,8 @@ bool dwc3_has_imod(struct dwc3 *dwc);
>>  int dwc3_event_buffers_setup(struct dwc3 *dwc);
>>  void dwc3_event_buffers_cleanup(struct dwc3 *dwc);
>>  
>> +int dwc3_core_soft_reset(struct dwc3 *dwc);
>> +
>>  #if IS_ENABLED(CONFIG_USB_DWC3_HOST) || IS_ENABLED(CONFIG_USB_DWC3_DUAL_ROLE)
>>  int dwc3_host_init(struct dwc3 *dwc);
>>  void dwc3_host_exit(struct dwc3 *dwc);
>> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
>> index 0c89baedf220..788889f924f9 100644
>> --- a/drivers/usb/dwc3/gadget.c
>> +++ b/drivers/usb/dwc3/gadget.c
>> @@ -2585,6 +2585,17 @@ static int dwc3_gadget_pullup(struct usb_gadget *g, int is_on)
>>  						dwc->ev_buf->length;
>>  		}
>>  	} else {
>> +		/*
>> +		 * In the Synopsis DesignWare Cores USB3 Databook Rev. 1.90a
>> +		 * Section 4.1.9, it specifies that for a reconnect after a
>> +		 * device-initiated disconnect requires a core soft reset
>> +		 * (DCTL.CSftRst) before enabling the run/stop bit.
>> +		 */
>> +		spin_unlock_irqrestore(&dwc->lock, flags);
>> +		dwc3_core_soft_reset(dwc);
>> +		spin_lock_irqsave(&dwc->lock, flags);
>> +
>> +		dwc3_event_buffers_setup(dwc);
> 
> Could you tell me why you add dwc3_event_buffer_setup?
> 
During my testing if the event buffer re-initialization is not there,
then it resulted in the controller attempting to access an invalid
address space after the soft reset completed. (SMMU fault w/ FAR=0x0)

Likewise, in other scenarios where core soft reset is called, there is
always an event buffer re-init following it.

Thanks
Wesley Cheng