gpio-altera OF binding discrepancy

gpio-altera OF binding discrepancy

[Marek Vasut]

Hi!

I've just checked linux-4.4 and linux-next and discovered the following
issue.

The Documentation/devicetree/bindings/gpio/gpio-altera.txt lists OF
property altr,interrupt-trigger , which is not used anywhere in the
kernel tree.

The drivers/gpio/gpio-altera.c on the other hand checks for a mandatory
OF property altr,interrupt-type and assigns it into variable
altera_gc->interrupt_trigger . Note that if this property is not in the
DT, the driver will fail to probe.

Clearly, there is a typo between altr,interrupt-trigger and
altr,interrupt-type . Since the driver fails to probe if altr,...-type
property is not present, there must be users of this variant in the
wild. I believe the Documentation/devicetree is a part of DT ABI, so the
bindings there should not be randomly changed.

In this case, I would propose to change the Documentation though, since
any users who try to use altr,...-trigger will have probe failures and
fix their DT anyway.

Does this make sense ? If so, I will submit a patch .

Thanks!

-- 
Best regards,
Marek Vasut

Re: gpio-altera OF binding discrepancy

[Tien Hock Loh]

Hi Marex

Yeah, it must've been a typo when I wrote the document, good catch
there. Can you help submit a patch to this?

Thanks

On Sat, 2016-02-27 at 03:10 +0100, marex@denx.de wrote:
> Hi!
> 
> I've just checked linux-4.4 and linux-next and discovered the following
> issue.
> 
> The Documentation/devicetree/bindings/gpio/gpio-altera.txt lists OF
> property altr,interrupt-trigger , which is not used anywhere in the
> kernel tree.
> 
> The drivers/gpio/gpio-altera.c on the other hand checks for a mandatory
> OF property altr,interrupt-type and assigns it into variable
> altera_gc->interrupt_trigger . Note that if this property is not in the
> DT, the driver will fail to probe.
> 
> Clearly, there is a typo between altr,interrupt-trigger and
> altr,interrupt-type . Since the driver fails to probe if altr,...-type
> property is not present, there must be users of this variant in the
> wild. I believe the Documentation/devicetree is a part of DT ABI, so the
> bindings there should not be randomly changed.
> 
> In this case, I would propose to change the Documentation though, since
> any users who try to use altr,...-trigger will have probe failures and
> fix their DT anyway.
> 
> Does this make sense ? If so, I will submit a patch .
> 
> Thanks!
> 

Re: gpio-altera OF binding discrepancy

[Marek Vasut]

On 02/29/2016 04:15 PM, Tien Hock Loh wrote:
> Hi Marex

Hi Tien!

> Yeah, it must've been a typo when I wrote the document, good catch
> there. Can you help submit a patch to this?

Sure, I will do shortly (today or tomorrow).

> Thanks
> 
> On Sat, 2016-02-27 at 03:10 +0100, marex@denx.de wrote:
>> Hi!
>>
>> I've just checked linux-4.4 and linux-next and discovered the following
>> issue.
>>
>> The Documentation/devicetree/bindings/gpio/gpio-altera.txt lists OF
>> property altr,interrupt-trigger , which is not used anywhere in the
>> kernel tree.
>>
>> The drivers/gpio/gpio-altera.c on the other hand checks for a mandatory
>> OF property altr,interrupt-type and assigns it into variable
>> altera_gc->interrupt_trigger . Note that if this property is not in the
>> DT, the driver will fail to probe.
>>
>> Clearly, there is a typo between altr,interrupt-trigger and
>> altr,interrupt-type . Since the driver fails to probe if altr,...-type
>> property is not present, there must be users of this variant in the
>> wild. I believe the Documentation/devicetree is a part of DT ABI, so the
>> bindings there should not be randomly changed.
>>
>> In this case, I would propose to change the Documentation though, since
>> any users who try to use altr,...-trigger will have probe failures and
>> fix their DT anyway.
>>
>> Does this make sense ? If so, I will submit a patch .
>>
>> Thanks!
>>
> 


-- 
Best regards,
Marek Vasut