Re: [Outreachy kernel] [PATCH] iio: adc: Call uninitialized_var for raw_sample

[Jonathan Cameron <jic23@kernel.org>]

On 19/03/15 14:05, Greg KH wrote:
> On Thu, Mar 19, 2015 at 07:04:21PM +0530, Darshana Padmadas wrote:
>> This patch calls uninitialized_var function for uninitialized
>> variable raw_sample. This error was reported by the static
>> analysis tool, cppcheck.
>>
>> Signed-off-by: Darshana Padmadas <darshanapadmadas@gmail.com>
>> ---
>>  drivers/iio/adc/ad_sigma_delta.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/iio/adc/ad_sigma_delta.c b/drivers/iio/adc/ad_sigma_delta.c
>> index d10bd0c..f0bbf3c 100644
>> --- a/drivers/iio/adc/ad_sigma_delta.c
>> +++ b/drivers/iio/adc/ad_sigma_delta.c
>> @@ -248,7 +248,8 @@ int ad_sigma_delta_single_conversion(struct iio_dev *indio_dev,
>>  	const struct iio_chan_spec *chan, int *val)
>>  {
>>  	struct ad_sigma_delta *sigma_delta = iio_device_get_drvdata(indio_dev);
>> -	unsigned int sample, raw_sample;
>> +	unsigned int sample;
>> +	unsigned int uninitialized_var(raw_sample);
> 
> I don't understand, what exactly is this "fixing"?
> 
> And note, cppcheck is known to give you lots of false warnings / errors,
> be very careful when using it.  Some kernel developers refuse to take
> patches that mention it given that it has been wrong so many times in
> the past.
> 
> thanks,
> 
> greg k-h
> 
To clarify, the only way that variable can be used uninitialized is if
we get an error from the spi transfer.  If that happens we have an error
code returned and hence don't use it anyway.

I suspect ccpcheck is either getting lost chasing through where that variable
is used, or is falling for the fact that the value of ret is not checked for
quite a few lines (so as to do some common cleanup which doesn't care if that
function succeeded or not).

Anyhow, Greg is of course correct. False positive as far as I can see.

Jonathan