[PATCH v2] usb: gadget: fsl_udc_core: Fix pullup status

[PATCH v2] usb: gadget: fsl_udc_core: Fix pullup status

[Dmitry Osipenko]

udc->softconnect should be set regardless of the VBUS state, otherwise
the USB peripheral device, connected during suspend, won't be detected
since can_pullup() would return false and the UDC won't be enabled.

Fixes: 252455c40316 (usb: gadget: fsl driver pullup fix)
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
---
Changelog:
V2: "(is_on != 0)" changed to "!!is_on" as per Sergei Shtylyov comment,
    cleaned up commit message.

 drivers/usb/gadget/udc/fsl_udc_core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/usb/gadget/udc/fsl_udc_core.c b/drivers/usb/gadget/udc/fsl_udc_core.c
index aab5221..4309b4f 100644
--- a/drivers/usb/gadget/udc/fsl_udc_core.c
+++ b/drivers/usb/gadget/udc/fsl_udc_core.c
@@ -1220,10 +1220,11 @@ static int fsl_pullup(struct usb_gadget *gadget, int is_on)
 
 	udc = container_of(gadget, struct fsl_udc, gadget);
 
+	udc->softconnect = !!is_on;
+
 	if (!udc->vbus_active)
 		return -EOPNOTSUPP;
 
-	udc->softconnect = (is_on != 0);
 	if (can_pullup(udc))
 		fsl_writel((fsl_readl(&dr_regs->usbcmd) | USB_CMD_RUN_STOP),
 				&dr_regs->usbcmd);
-- 
2.7.4

Re: [PATCH v2] usb: gadget: fsl_udc_core: Fix pullup status

[Felipe Balbi]

[-- Attachment #1: Type: text/plain, Size: 1364 bytes --]

Dmitry Osipenko <digetx@gmail.com> writes:
> udc->softconnect should be set regardless of the VBUS state, otherwise
> the USB peripheral device, connected during suspend, won't be detected
> since can_pullup() would return false and the UDC won't be enabled.
>
> Fixes: 252455c40316 (usb: gadget: fsl driver pullup fix)
> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
> ---
> Changelog:
> V2: "(is_on != 0)" changed to "!!is_on" as per Sergei Shtylyov comment,
>     cleaned up commit message.
>
>  drivers/usb/gadget/udc/fsl_udc_core.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/usb/gadget/udc/fsl_udc_core.c b/drivers/usb/gadget/udc/fsl_udc_core.c
> index aab5221..4309b4f 100644
> --- a/drivers/usb/gadget/udc/fsl_udc_core.c
> +++ b/drivers/usb/gadget/udc/fsl_udc_core.c
> @@ -1220,10 +1220,11 @@ static int fsl_pullup(struct usb_gadget *gadget, int is_on)
>  
>  	udc = container_of(gadget, struct fsl_udc, gadget);
>  
> +	udc->softconnect = !!is_on;
> +
>  	if (!udc->vbus_active)
>  		return -EOPNOTSUPP;
>  
> -	udc->softconnect = (is_on != 0);

if we're suspended and VBUS was cut off, why would we keep softconnect
set to true ? That would also cause a discrepancy between SW state and
HW state.

I don't have this HW to test, but this patch seems wrong to me.

-- 
balbi

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 818 bytes --]

Re: [PATCH v2] usb: gadget: fsl_udc_core: Fix pullup status

[Dmitry Osipenko]

29.03.2016 13:31, Felipe Balbi пишет:
> Dmitry Osipenko <digetx@gmail.com> writes:
>> udc->softconnect should be set regardless of the VBUS state, otherwise
>> the USB peripheral device, connected during suspend, won't be detected
>> since can_pullup() would return false and the UDC won't be enabled.
>>
>> Fixes: 252455c40316 (usb: gadget: fsl driver pullup fix)
>> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
>> ---
>> Changelog:
>> V2: "(is_on != 0)" changed to "!!is_on" as per Sergei Shtylyov comment,
>>      cleaned up commit message.
>>
>>   drivers/usb/gadget/udc/fsl_udc_core.c | 3 ++-
>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/usb/gadget/udc/fsl_udc_core.c b/drivers/usb/gadget/udc/fsl_udc_core.c
>> index aab5221..4309b4f 100644
>> --- a/drivers/usb/gadget/udc/fsl_udc_core.c
>> +++ b/drivers/usb/gadget/udc/fsl_udc_core.c
>> @@ -1220,10 +1220,11 @@ static int fsl_pullup(struct usb_gadget *gadget, int is_on)
>>
>>   	udc = container_of(gadget, struct fsl_udc, gadget);
>>
>> +	udc->softconnect = !!is_on;
>> +
>>   	if (!udc->vbus_active)
>>   		return -EOPNOTSUPP;
>>
>> -	udc->softconnect = (is_on != 0);
>
> if we're suspended and VBUS was cut off, why would we keep softconnect
> set to true ? That would also cause a discrepancy between SW state and
> HW state.
>
> I don't have this HW to test, but this patch seems wrong to me.
>

Yeah, you are right. I'm using a fork of this driver with some minor differences 
to make it work with other hardware and missed that upstream driver enables 
controller on resume unconditionally. Sorry for the noise and please ignore this 
patch.

-- 
Dmitry

Re: [PATCH v2] usb: gadget: fsl_udc_core: Fix pullup status

[Felipe Balbi]

[-- Attachment #1: Type: text/plain, Size: 2122 bytes --]

Dmitry Osipenko <digetx@gmail.com> writes:
> 29.03.2016 13:31, Felipe Balbi пишет:
>> Dmitry Osipenko <digetx@gmail.com> writes:
>>> udc->softconnect should be set regardless of the VBUS state, otherwise
>>> the USB peripheral device, connected during suspend, won't be detected
>>> since can_pullup() would return false and the UDC won't be enabled.
>>>
>>> Fixes: 252455c40316 (usb: gadget: fsl driver pullup fix)
>>> Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
>>> ---
>>> Changelog:
>>> V2: "(is_on != 0)" changed to "!!is_on" as per Sergei Shtylyov comment,
>>>      cleaned up commit message.
>>>
>>>   drivers/usb/gadget/udc/fsl_udc_core.c | 3 ++-
>>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/usb/gadget/udc/fsl_udc_core.c b/drivers/usb/gadget/udc/fsl_udc_core.c
>>> index aab5221..4309b4f 100644
>>> --- a/drivers/usb/gadget/udc/fsl_udc_core.c
>>> +++ b/drivers/usb/gadget/udc/fsl_udc_core.c
>>> @@ -1220,10 +1220,11 @@ static int fsl_pullup(struct usb_gadget *gadget, int is_on)
>>>
>>>   	udc = container_of(gadget, struct fsl_udc, gadget);
>>>
>>> +	udc->softconnect = !!is_on;
>>> +
>>>   	if (!udc->vbus_active)
>>>   		return -EOPNOTSUPP;
>>>
>>> -	udc->softconnect = (is_on != 0);
>>
>> if we're suspended and VBUS was cut off, why would we keep softconnect
>> set to true ? That would also cause a discrepancy between SW state and
>> HW state.
>>
>> I don't have this HW to test, but this patch seems wrong to me.
>>
>
> Yeah, you are right. I'm using a fork of this driver with some minor differences 
> to make it work with other hardware and missed that upstream driver enables 
> controller on resume unconditionally. Sorry for the noise and please ignore this 
> patch.

you shouldn't send patches unless you have tested them. Simply
cherry-picking from another tree and sending it is NOT good enough. You
MUST test what you're sending with latest tag from Linus (right now,
that's v4.6-rc1). If you can't update your kernel, then don't try to
patch something you can't validate.

Thanks

-- 
balbi

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 818 bytes --]