[PATCH] musb: sunxi: Ignore VBus errors in host-only mode

[Hans de Goede]

Hi,

On 05-08-15 00:57, Felipe Balbi wrote:
> Hi,
>
> On Wed, Aug 05, 2015 at 12:05:02AM +0200, Hans de Goede wrote:
>> Hi,
>>
>> On 08/04/2015 11:35 PM, Felipe Balbi wrote:
>>> On Tue, Aug 04, 2015 at 11:25:53PM +0200, Hans de Goede wrote:
>>>> For some unclear reason sometimes we get VBus errors in host-only mode,
>>>> even though we do not have any vbus-detection then. Ignore these.
>>>>
>>>> Signed-off-by: Hans de Goede <hdegoede at redhat.com>
>>>> ---
>>>>   drivers/usb/musb/sunxi.c | 4 ++++
>>>>   1 file changed, 4 insertions(+)
>>>>
>>>> diff --git a/drivers/usb/musb/sunxi.c b/drivers/usb/musb/sunxi.c
>>>> index f9f6304..34ce5df 100644
>>>> --- a/drivers/usb/musb/sunxi.c
>>>> +++ b/drivers/usb/musb/sunxi.c
>>>> @@ -194,6 +194,10 @@ static irqreturn_t sunxi_musb_interrupt(int irq, void *__hci)
>>>>   		musb_writeb(musb->mregs, MUSB_FADDR, 0);
>>>>   	}
>>>>
>>>> +	/*  Ignore Vbus errors when in host only mode */
>>>> +	if (musb->port_mode == MUSB_PORT_MODE_HOST)
>>>> +		musb->int_usb &= ~MUSB_INTR_VBUSERROR;
>>>
>>> check with a scope if VBUS is really dropping. Host does VBUS detection
>>> indeed, at a minimum, for overcurrent protection. You might have
>>> something causing VBUS to drop and that something needs to be found,
>>> rather than masked.
>>
>> The boards in question do not have any vbus detection, the usb-phy on
>> allwinner boards do not have a vbus sense pin, instead a gpio is used
>> in designs which use otg. Designs which use host-only mode typically
>> do not have any form of vbus detection at all. In this case we always
>> report vbus as being valid to the musb core, so I've no idea why
>> the musb core is still generating vbus errors.
>
> PHY must tell MUSB about VBUS level using either ULPI or UTMI+. I'd
> check to see if the PHY is at fault, at least.

Right, and the phy code has:

                 if (data->id_det_gpio) {
                         /* OTG mode, force ISCR and cable state updates */
                         data->id_det = -1;
                         data->vbus_det = -1;
                         queue_delayed_work(system_wq, &data->detect, 0);
                 } else {
                         /* Host only mode */
                         sun4i_usb_phy0_set_id_detect(_phy, 0);
                         sun4i_usb_phy0_set_vbus_detect(_phy, 1);
                 }

Where we enter the host only path (no id-pin in host only mode) and
then sun4i_usb_phy0_set_vbus_detect updates the ISCR register of
the phy to make it report vbus valid. I've done an mmio-dump of
the iscr register while running Linux, and this code does the right
thing.

I agree with you that this is weird, but atm I see no other way
to fix this then the submitted patch.

Regards,

Hans


>