[RFC][i2c-tools] Accept device-node path as alternative to bus number

[RFC][i2c-tools] Accept device-node path as alternative to bus number

[Gero Schwäricke]

I'm working with multiple hotpluggable I2C adapters and have the
following proposal:

Allow all i2c-tools to accept a device-node path (e.g., /dev/i2c-5) as
the i2cbus argument.

Example:

  # pre RFC:
  i2cget 5 0x48

  # post RFC:
  i2cget /dev/i2c-5 0x48

Motivation:

- Paths can be made stable via udev symlinks even when bus numbers vary
  between boots, hotplugging, or across machines.
- Scripts often already have a device node path. Accepting the path
  removes the need to parse out the numeric bus ID.

Backward compatibility:

- No behavior change for existing invocations.
- The first non-option argv is treated as today unless it begins with
  '/', in which case the tool treats it as a path and opens that exact
  node.

Sketch of changes:

- For each tool’s argument parsing, detect a leading '/' for the bus
  argument.
- If a path is given: use that path; skip bus resolution by number/name.
- Adjust control logic to work with paths instead of bus number.
- Update help/man text: add e.g., "i2cbus indicates ..., or a device
  node path like /dev/i2c-5."
 
Questions for maintainers:

1. Is accepting a device-node path an acceptable extension to the CLI?
2. Currently the bus number/name argument is parsed to an integer and
then only stores as such in struct i2c_adap. When calling open_i2c_dev()
the number is resolved to a /dev/i2c path. Is it acceptable to instead
resolve a given number/name to a path during argument parsing and store
the path in struct i2c_adap instead of the number?

Thank you,
Gero

Re: [RFC][i2c-tools] Accept device-node path as alternative to bus number

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 244 bytes --]

On Mon, Nov 10, 2025 at 03:57:34PM +0100, Gero Schwäricke wrote:
> I'm working with multiple hotpluggable I2C adapters and have the
> following proposal:

From my side, this all sounds reasonable. But we need Jean's opinion
here, too.


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [RFC][i2c-tools] Accept device-node path as alternative to bus number

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 274 bytes --]


> > I'm working with multiple hotpluggable I2C adapters and have the
> > following proposal:
> 
> From my side, this all sounds reasonable. But we need Jean's opinion
> here, too.

Seems Jean is busy. So, from my side, let's go. I can apply it if all is
well...


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [RFC][i2c-tools] Accept device-node path as alternative to bus number

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 437 bytes --]

Hi Gero,

On Mon, Dec 15, 2025 at 10:58:24PM +0900, Wolfram Sang wrote:
> 
> > > I'm working with multiple hotpluggable I2C adapters and have the
> > > following proposal:
> > 
> > From my side, this all sounds reasonable. But we need Jean's opinion
> > here, too.
> 
> Seems Jean is busy. So, from my side, let's go. I can apply it if all is
> well...

Did you have time to develop this?

Happy hacking,

   Wolfram


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [RFC][i2c-tools] Accept device-node path as alternative to bus number

[Gero Schwäricke]

Hi Wolfram,

On Tue Jan 13, 2026 at 6:38 PM CET, Wolfram Sang wrote:
> Hi Gero,
>
> On Mon, Dec 15, 2025 at 10:58:24PM +0900, Wolfram Sang wrote:
>> 
>> > > I'm working with multiple hotpluggable I2C adapters and have the
>> > > following proposal:
>> > 
>> > From my side, this all sounds reasonable. But we need Jean's opinion
>> > here, too.
>> 
>> Seems Jean is busy. So, from my side, let's go. I can apply it if all is
>> well...
>
> Did you have time to develop this?

thanks for following up on this and sorry for the late response, I'm
currently occupied with other work, but I have this still on my TODO
list. I intend to tackle this once I find some uninterrupted time.

Best,
Gero

>
> Happy hacking,
>
>    Wolfram

Re: [RFC][i2c-tools] Accept device-node path as alternative to bus number

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 323 bytes --]

Hi Gero,

> thanks for following up on this and sorry for the late response, I'm
> currently occupied with other work, but I have this still on my TODO
> list. I intend to tackle this once I find some uninterrupted time.

Sure thing. Thanks for the heads up. I will stop pinging and just wait.

Happy hacking,

   Wolfram


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: [RFC][i2c-tools] Accept device-node path as alternative to bus number

[Jean Delvare]

Hi Gero, Wolfram,

On Mon, 10 Nov 2025 17:15:44 +0100, Wolfram Sang wrote:
> On Mon, Nov 10, 2025 at 03:57:34PM +0100, Gero Schwäricke wrote:
> > I'm working with multiple hotpluggable I2C adapters and have the
> > following proposal:  
> 
> From my side, this all sounds reasonable. But we need Jean's opinion
> here, too.

Sorry for the delay. The proposal sounds perfectly reasonable to me,
and I'd be happy to see and review a candidate implementation thereof.

Please make sure that any piece of code which is common to all tools
ends up in i2cbusses.c and does not get duplicated.

Thanks,
-- 
Jean Delvare
SUSE L3 Support