Re: [PATCH net-next v2] mctp i2c: add sysfs control for mux child netdevs

[Matt Johnston <matt@codeconstruct.com.au>]

Hi Kenny,

On Tue, 2026-11-03 at 14:47 +0000, Kenny Hsieh wrote:
> Hi Matt,
> 
> On Mon, 2026-11-02 at 17:04 +0800, Matt Johnston wrote:
> > An alternative could be to create sysfs attributes under the mctp-i2c device
> > instance itself (the i2c slave).
> > 
> > Mux buses could have mctp-i2c netdevs instantiated with
> > 
> > echo "i2c-6" > /sys/bus/i2c/devices/2-1060/add_mctp_controller
> > 
> > (where i2c-6 is a child mux bus of i2c-2, and i2c-2 0x60 is a mctp-i2c i2c
> > device).
> > 
> > "echo i2c-6 > .../2-1060/remove_mctp_controller" would remove the mctp-i2c
> > netdev.
> > Setting a root node as an mctp controller would involve writing its own name
> > to the attribute file.
> > 
> > How does that look to you?
> 
> I like that a lot better. Since the attributes live on the mctp-i2c
> device, they can just go in dev_groups and the driver core creates and
> removes them for us. So I can drop the per-adapter tracking list, and
> the remove ordering and probe walk races Sashiko found go away with it.
> 
> Let me know if you'd rather have any of these differently:
> 
>   - adding a bus that already has a netdev, or removing one that
>     doesn't, just returns success

Would it be better to return EEXIST (add) or ENODEV (remove) for those cases?
That could catch mistakes. Usually there'd be a single userspace program
managing it, so it doesn't need to be idempotent?

>   - a bus that isn't in this controller's mux tree gets -EINVAL
>   - a bus name that doesn't exist, e.g. i2c-9999, gets -ENODEV

Those look good.

Cheers,
Matt