Re: [PATCH v1 2/3] hwmon: (emc1403) Convert to use OF bindings

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The driver probe unconditionally fails for `emc1402` devices because its chip enum value evaluates to 0, which triggers an erroneous error path.
- [Low] Directly casting `const void *` to `enum emc1403_chip` causes a `-Wpointer-to-int-cast` compiler warning on 64-bit architectures.
--

commit 8e14b2d237bb423a526fb43282f03dadf7dbb7dc
Author: Jonas Schwöbel <jonasschwoebel@yahoo.de>

hwmon: (emc1403) Convert to use OF bindings

This commit adds an OF match compatible table. This allows the driver to be 
used with Device Trees.

> diff --git a/drivers/hwmon/emc1403.c b/drivers/hwmon/emc1403.c
> --- a/drivers/hwmon/emc1403.c
> +++ b/drivers/hwmon/emc1403.c
[ ... ]
> @@ -659,14 +660,16 @@ static int emc1403_probe(struct i2c_client *client)
[ ... ]
> -	data->chip = id->driver_data;
> +	data->chip = (enum emc1403_chip)i2c_get_match_data(client);

[Severity: Low]
Does directly casting the pointer returned by i2c_get_match_data() to a 
32-bit enum cause a pointer-to-int-cast compiler warning on 64-bit 
architectures? Would it be safer to cast it to an intermediate 
pointer-sized integer first?

> +	if (!data->chip)

[Severity: High]
Since enum emc1403_chip defines emc1402 as 0, will this check evaluate 
to true for devices compatible with emc1402? It appears this might cause 
the probe function to erroneously return -ENODEV and break initialization 
for chips mapping to emc1402.

> +		return dev_err_probe(&client->dev, -ENODEV, "Unknown device!\n");
> +
>  	data->regmap = devm_regmap_init_i2c(client, &emc1403_regmap_config);
[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260603141033.111300-1-clamor95@gmail.com?part=2