[U-Boot] [PATCH 1/2] usb: dwc2: correctly handle binding for g-tx-fifo-size

[Marek Vasut]

On 6/18/19 9:32 AM, Patrick DELAUNAY wrote:
> Hi Marek,
> 
>> From: Marek Vasut <marex at denx.de>
>> Sent: lundi 17 juin 2019 17:54
>>
>> On 6/14/19 1:08 PM, Patrick Delaunay wrote:
>>> Manage g-tx-fifo-size as a array as specify in the binding.
>>>
>>> Signed-off-by: Patrick Delaunay <patrick.delaunay at st.com>
>>> ---
>>>
>>>  arch/arm/dts/stm32mp157c-ev1-u-boot.dtsi |  4 ----
>>>  drivers/usb/gadget/dwc2_udc_otg.c        | 17 ++++++++++++++++-
>>>  2 files changed, 16 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/arch/arm/dts/stm32mp157c-ev1-u-boot.dtsi
>>> b/arch/arm/dts/stm32mp157c-ev1-u-boot.dtsi
>>> index 5b19e44..994092a 100644
>>> --- a/arch/arm/dts/stm32mp157c-ev1-u-boot.dtsi
>>> +++ b/arch/arm/dts/stm32mp157c-ev1-u-boot.dtsi
>>> @@ -56,10 +56,6 @@
>>>  	};
>>>  };
>>>
>>> -&usbotg_hs {
>>> -	g-tx-fifo-size = <576>;
>>
>> Should this really be in this patch ?
> 
> As I change the binding parsing, the stm32mp1 will don't work without this patch.
> I make a commun patch only to allow bisec, but I can split the serie with 2 patches.
> 
>>
>>> -};
>>> -
>>>  &v3v3 {
>>>  	regulator-always-on;
>>>  };
>>> diff --git a/drivers/usb/gadget/dwc2_udc_otg.c
>>> b/drivers/usb/gadget/dwc2_udc_otg.c
>>> index 494ab53..7e6b5fc 100644
>>> --- a/drivers/usb/gadget/dwc2_udc_otg.c
>>> +++ b/drivers/usb/gadget/dwc2_udc_otg.c
>>> @@ -1039,6 +1039,8 @@ static int dwc2_udc_otg_ofdata_to_platdata(struct
>> udevice *dev)
>>>  	int node = dev_of_offset(dev);
>>>  	ulong drvdata;
>>>  	void (*set_params)(struct dwc2_plat_otg_data *data);
>>> +	u32 tx_fifo_sz_array[DWC2_MAX_HW_ENDPOINTS];
>>
>> Can't you just read directly into platdata->tx_fifo_sz_array[] below, and thus drop
>> this temporary variable ?
> 
> It was the case in in my first internal version.
> 
> if (platdata->tx_fifo_sz_nb) {
> 		ret = dev_read_u32_array(dev, "g-tx-fifo-size",
> 					 &platdata->tx_fifo_sz_array,
> 					 platdata->tx_fifo_sz_nb);
> 		if (ret)
> 			return ret;
> 	}
> 
> And I add it to avoid the warning / potential issue:
> 
> /local/home/frq07632/views/u-boot/u-boot/drivers/usb/gadget/dwc2_udc_otg.c:1062:7: 
> 	warning: passing argument 3 of ‘dev_read_u32_array’ from incompatible pointer type [-Wincompatible-pointer-types]
>        &platdata->tx_fifo_sz_array,
>        ^~~~~~~~~~~~~~~~~~~~~~~~~~~
> In file included from /local/home/frq07632/views/u-boot/u-boot/include/dm.h:12,
>              from /local/home/frq07632/views/u-boot/u-boot/drivers/usb/gadget/dwc2_udc_otg.c:22:
> 	/local/home/frq07632/views/u-boot/u-boot/include/dm/read.h:710:15: note: expected ‘u32 *’ {aka ‘unsigned int *’} but argument is of type ‘unsigned int (*)[16]’
>           u32 *out_values, size_t sz)
>           ~~~~~^~~~~~~~~~

Looks like GCC is complaining because you're passing an array of
pointers to dev_read_u32_array() , instead of plain pointer .

Try

 		ret = dev_read_u32_array(dev, "g-tx-fifo-size",
- 					 &platdata->tx_fifo_sz_array,
+ 					 platdata->tx_fifo_sz_array,
 					 platdata->tx_fifo_sz_nb);

>> btw is this fix for current release or next ?
> 
> I hope it for the v2019.07 (as it is only impact the stm32mp1 arch/board). 
> But it is not blocking.

OK

-- 
Best regards,
Marek Vasut