Re: [PATCH 1/3] hwmon: tmp421: handle I2C errors

[Guenter Roeck]

On Wed, Sep 22, 2021 at 04:41:52PM +0300, Paul Fertser wrote:
> Function i2c_smbus_read_byte_data() can return a negative error number
> instead of the data read if I2C transaction failed for whatever reason.
> 
> I consider this fix to be stable material as lack of error checking here
> leads to serious issues on production hardware. Errors treated as
> temperatures produce spurious critical temperature-crossed-threshold
> errors in BMC logs for OCP server hardware. The patch was tested with
> Mellanox OCP Mezzanine card emulating TMP421 protocol for temperature
> sensing which sometimes leads to I2C protocol error during early boot up
> stage.
> 
> Cc: stable@xxxxxxxxxxxxxxx
> Signed-off-by: Paul Fertser <fercerpav@xxxxxxxxx>
> ---
>  drivers/hwmon/tmp421.c | 31 ++++++++++++++++++++++---------
>  1 file changed, 22 insertions(+), 9 deletions(-)
> 
> diff --git a/drivers/hwmon/tmp421.c b/drivers/hwmon/tmp421.c
> index ede66ea6a730..6175ed4b10bd 100644
> --- a/drivers/hwmon/tmp421.c
> +++ b/drivers/hwmon/tmp421.c
> @@ -93,7 +93,7 @@ struct tmp421_data {
>  	struct hwmon_channel_info temp_info;
>  	const struct hwmon_channel_info *info[2];
>  	struct hwmon_chip_info chip;
> -	char valid;
> +	int last_errno;
>  	unsigned long last_updated;
>  	unsigned long channels;
>  	u8 config;
> @@ -128,20 +128,30 @@ static struct tmp421_data *tmp421_update_device(struct device *dev)
>  	mutex_lock(&data->update_lock);
>  
>  	if (time_after(jiffies, data->last_updated + (HZ / 2)) ||
> -	    !data->valid) {
> -		data->config = i2c_smbus_read_byte_data(client,
> -			TMP421_CONFIG_REG_1);
> +	    data->last_errno) {
> +		data->last_errno = i2c_smbus_read_byte_data(client,
> +							    TMP421_CONFIG_REG_1);

No. The function should return an ERR_PTR after an error.
Something like
	int ret = 0;
	...
	ret = i2c_smbus_read_byte_data(client, TMP421_CONFIG_REG_1);
	if (ret < 0)
		goto exit;
	data->config = ret;
	...
exit:
	mutex_unlock(...);
	return ret < 0 ? ERR_PTR(ret) : data;

Or, even better, let tmp421_update_device() return an error code instead
of data, and let the caller get the data pointer.

int tmp421_update_device(struct tmp421_data *data)
{
	struct i2c_client *client = data->client;
	int ret = 0;

	...
	return ret < 0 ? ret : 0;
}
...

	struct tmp421_data *data = dev_get_drvdata(dev);

	ret = tmp421_update_device(data);
	if (ret)
		return ret;


Guenter

> +		if (data->last_errno < 0)
> +			goto exit;
> +		data->config =  data->last_errno;
>  
>  		for (i = 0; i < data->channels; i++) {
> -			data->temp[i] = i2c_smbus_read_byte_data(client,
> -				TMP421_TEMP_MSB[i]) << 8;
> -			data->temp[i] |= i2c_smbus_read_byte_data(client,
> -				TMP421_TEMP_LSB[i]);
> +			data->last_errno = i2c_smbus_read_byte_data(client,
> +								    TMP421_TEMP_MSB[i]);
> +			if (data->last_errno < 0)
> +				goto exit;
> +			data->temp[i] = data->last_errno << 8;
> +			data->last_errno = i2c_smbus_read_byte_data(client,
> +								    TMP421_TEMP_LSB[i]);
> +			if (data->last_errno < 0)
> +				goto exit;
> +			data->temp[i] |= data->last_errno;
>  		}
>  		data->last_updated = jiffies;
> -		data->valid = 1;
> +		data->last_errno = 0;
>  	}
>  
> +exit:
>  	mutex_unlock(&data->update_lock);
>  
>  	return data;
> @@ -152,6 +162,9 @@ static int tmp421_read(struct device *dev, enum hwmon_sensor_types type,
>  {
>  	struct tmp421_data *tmp421 = tmp421_update_device(dev);
>  
> +	if (tmp421->last_errno)
> +		return tmp421->last_errno;
> +
>  	switch (attr) {
>  	case hwmon_temp_input:
>  		if (tmp421->config & TMP421_CONFIG_RANGE)
> -- 
> 2.17.1
>