Re: module parameters permission

[Finn Behrens]

The module macro is a proc_macro. This is written in a way to only accept one single iteral at that position. So that is an expected behaviour with the current implementation.
If needed, this macro could be extended with functionality to deal with that case, but if octal values are preferred, I don’t see a reason to introduce that level of complexity.

Finn

> On 29. Apr 2021, at 12:25, Fabio Aiuto <fabioaiuto83@xxxxxxxxx> wrote:
> 
> On Thu, Apr 29, 2021 at 12:05:08PM +0200, Greg KH wrote:
>> On Thu, Apr 29, 2021 at 11:58:20AM +0200, Fabio Aiuto wrote:
>>> Hi all,
>>> 
>>> I'm trying to declare module parameters this way:
>>> 
>>> 
>>>   params: {
>>>        scull_major: i32 {
>>>            default: 0,
>>>            permissions: bindings::S_IRUGO as i32,
>>>            description: b"Major number",
>>>        },
>>>        scull_minor: i32 {
>>>            default: 0,
>>>            permissions: bindings::S_IRUGO as i32,
>>>            description: b"Minor number",
>>>        },
>>> 
>>> i.e. using S_IRUGO macro exposed by bindgen. But I have the
>>> following compiler error:
>>> 
>>> error: proc macro panicked
>>>  --> samples/rust/rust_scull.rs:12:1
>>>   |
>>> 12 | / module! {
>>> 13 | |     type: RustScull,
>>> 14 | |     name: b"rust_scull",
>>> 15 | |     author: b"Alessandro Rubini, Jonathan Corbet",
>>> ...  |
>>> 44 | |     },
>>> 45 | | }
>>>   | |_^
>>>   |
>>>   = help: message: Expected Literal
>>> 
>>> the same if I remove as i32 casts.
>>> 
>>> if I write permissions as in samples/rust/rust_module_parameters.rs
>>> 
>>>    params: {
>>>        my_bool: bool {
>>>            default: true,
>>>            permissions: 0,
>>>            description: b"Example of bool",
>>>        },
>>>        my_i32: i32 {
>>>            default: 42,
>>>            permissions: 0o644, <-------
>>>            description: b"Example of i32",
>>>        },
>>> 
>>> I get no error.
>>> 
>>> What's the right way to use S_I*UGO macros?
>> 
>> Not at all, use the octal values instead please.
>> 
>> That's the way that we have declared a while ago, and I think
>> checkpatch.pl will even catch if you try to do this in any new code.
>> Please don't force us to deal with S_* defines in rust code as well.
>> 
>> thanks,
>> 
>> greg k-h
> 
> thank you I didn't know that. I will use octal than.
> 
> Anyway I'd like to know what was the matter with those bindings...
> 
> fabio