Re: [PATCH v1 hwmon] hwmon: (pmbus) Fix driver info initialization in probe routine

[Guenter Roeck <linux@roeck-us.net>]

Hi,

On Sun, Feb 17, 2019 at 05:41:55PM +0000, Vadim Pasternak wrote:
> Fix tps53679_probe() by using dynamically allocated "pmbus_driver_info"
> structure instead of static. Usage of static structures causes
> overwritten of the field "vrm_version", in case the system is equipped
> with several tps53679 devices with the different "vrm_version".
> In such case the last probed device overwrites this field for all
> others.
> 
> Fixes: 610526527a13 ("hwmon: (pmbus) Add support for Texas Instruments tps53679 device")
> Signed-off-by: Vadim Pasternak <vadimp@mellanox.com>
> ---
>  drivers/hwmon/pmbus/tps53679.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/hwmon/pmbus/tps53679.c b/drivers/hwmon/pmbus/tps53679.c
> index 85b515cd9df0..45eacc504e2f 100644
> --- a/drivers/hwmon/pmbus/tps53679.c
> +++ b/drivers/hwmon/pmbus/tps53679.c
> @@ -80,7 +80,15 @@ static struct pmbus_driver_info tps53679_info = {
>  static int tps53679_probe(struct i2c_client *client,
>  			  const struct i2c_device_id *id)
>  {
> -	return pmbus_do_probe(client, id, &tps53679_info);
> +	struct pmbus_driver_info *info;
> +
> +	info = devm_kzalloc(&client->dev, sizeof(*info), GFP_KERNEL);
> +	if (!info)
> +		return -ENOMEM;
> +
> +	memcpy(info, &tps53679_info, sizeof(*info));
> +

Good find, but why not use devm_kmemdup() ?

Thanks,
Guenter

> +	return pmbus_do_probe(client, id, info);
>  }
>  
>  static const struct i2c_device_id tps53679_id[] = {