[RFC] i2c_check_functionality and error code

[RFC] i2c_check_functionality and error code

[Matt Ranostay]

Jonathan et all,

Has anyone noticed that there is no clear consensus on which error
code to return when a i2c_check_functionality() check fails within the
probe function. I've seen so far ENODEV, ENOTSUPP, EOPNOTSUPP, EIO,
and ENOSYS in drivers/iio

Shouldn't these be made a standard value like -ENOTSUPP?

Thanks,

Matt

Re: [RFC] i2c_check_functionality and error code

[Jonathan Cameron]

On 14/02/16 23:09, Matt Ranostay wrote:
> Jonathan et all,
> 
> Has anyone noticed that there is no clear consensus on which error
> code to return when a i2c_check_functionality() check fails within the
> probe function. I've seen so far ENODEV, ENOTSUPP, EOPNOTSUPP, EIO,
> and ENOSYS in drivers/iio
> 
> Shouldn't these be made a standard value like -ENOTSUPP?
Would make sense - but is this the right choice.

Thought I'd grep HWMON as a possible source of a consensus on this and
got no clear answer.  The most common in there looks to be -ENODEV though
(From the first few pages of results anyway ;)

Hohum. Wolfram what do you think?

Worth cleaning this up?  Perhaps even kernel wise would lead to some
consistency. I've never been that sharp on this in IIO so I can't
really talk ;)

Jonathan


> 
> Thanks,
> 
> Matt
> 

Re: [RFC] i2c_check_functionality and error code

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 197 bytes --]

> Hohum. Wolfram what do you think?

I'd go for "EOPNOTSUPP" since it is most descriptive. I wouldn't say we
really need to fix it up in the kernel tree. However, I neither would
vote against it.


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: [RFC] i2c_check_functionality and error code

[Jonathan Cameron]

On 21/02/16 22:56, Wolfram Sang wrote:
>> Hohum. Wolfram what do you think?
> 
> I'd go for "EOPNOTSUPP" since it is most descriptive. I wouldn't say we
> really need to fix it up in the kernel tree. However, I neither would
> vote against it.
> 
I'm certainly keen on cleaning this up in IIO at the least.
A bit of consistency is always nice and once we have done it once
it should be reasonably pain free.

Anyhow, patches welcome ;)

Jonathan