[U-Boot] [PATCH] usb: ehci-mx6: allow board_ehci_hcd_init to fail

[Marek Vasut]

On 05/06/2016 07:11 PM, Stefan Agner wrote:
> On 2016-05-06 09:30, Marek Vasut wrote:
>> On 05/06/2016 06:12 PM, Stefan Agner wrote:
>>> On 2016-05-05 17:39, Marek Vasut wrote:
>>>> On 05/06/2016 01:59 AM, Stefan Agner wrote:
>>>>> There could be runtime determined board specific reason why a EHCI
>>>>> initialization fails (e.g. ENODEV if a Port is not available). In
>>>>> this case, properly return the error code.
>>>>> While at it, that function (board_ehci_hcd_init) has actually two
>>>>> documentation blocks... Use the correct function name for the
>>>>> documentation block of board_usb_phy_mode.
>>>>>
>>>>> Signed-off-by: Stefan Agner <stefan at agner.ch>
>>>>> ---
>>>>>  drivers/usb/host/ehci-mx6.c | 7 +++++--
>>>>>  1 file changed, 5 insertions(+), 2 deletions(-)
>>>>>
>>>>> diff --git a/drivers/usb/host/ehci-mx6.c b/drivers/usb/host/ehci-mx6.c
>>>>> index a981b50..bb48d0d 100644
>>>>> --- a/drivers/usb/host/ehci-mx6.c
>>>>> +++ b/drivers/usb/host/ehci-mx6.c
>>>>> @@ -254,7 +254,7 @@ static void usb_oc_config(int index)
>>>>>  }
>>>>>
>>>>>  /**
>>>>> - * board_ehci_hcd_init - override usb phy mode
>>>>> + * board_usb_phy_mode - override usb phy mode
>>>>>   * @port:	usb host/otg port
>>>>>   *
>>>>>   * Target board specific, override usb_phy_mode.
>>>>> @@ -310,6 +310,7 @@ int ehci_hcd_init(int index, enum usb_init_type init,
>>>>>  #endif
>>>>>  	struct usb_ehci *ehci = (struct usb_ehci *)(USB_BASE_ADDR +
>>>>>  		(controller_spacing * index));
>>>>> +	int ret;
>>>>>
>>>>>  	if (index > 3)
>>>>>  		return -EINVAL;
>>>>> @@ -317,7 +318,9 @@ int ehci_hcd_init(int index, enum usb_init_type init,
>>>>>  	mdelay(1);
>>>>>
>>>>>  	/* Do board specific initialization */
>>>>> -	board_ehci_hcd_init(index);
>>>>> +	ret = board_ehci_hcd_init(index);
>>>>> +	if (ret)
>>>>> +		return ret;
>>>>
>>>> Shouldn't this disable the OH3 clock , which were enabled a few lines above?
>>>
>>> Hm, not sure, the clock does not take a controller index... I guess if
>>> the second instance is returning an error, that would disable the clock
>>> also for the first controller...
>>
>> Siiiiiiigh! All right, that sucks and we absolutely have no time for
>> reworking the whole clock framework on imx for this release. You want
>> this stuff in for this release, right ?
> 
> Wasn't that urgent as the board which will use this is not upstream yet
> anyway... But still, nice to have it in.
> 
> Btw, I really like the new 2 month release cycle.

Heh, tell that to Tom :)

> --
> Stefan
> 


-- 
Best regards,
Marek Vasut