False positive candidate Rule 10.3 for float array zero-initialization

Hello,

Polyspace is raising the 10.3 MISRA violation for array initializations like the one below:

float data[3*3] = {0}

My question: is this the intended behavior of Polyspace? Shouldn't this type of zero-initialization be allowed?

My impression is that the MISRA 2012 standard allows this by exception. Other static analyzers do not raise a warning. Empty array-initialization makes the warning go away, so we may just do that.

Thanks.

Answer 1

[Anirban]

The problem is not because of the array initialization. Indeed, as you say, MISRA C:2012 standard allows an exception of rule 10.3 for aggregate initializations with {0}.

The violation is shown because you are assigning 0 to a float variable (the two have different essential types, see here for essential types). But the MISRA exception for {0} covers aggregates of all data types. So, this is indeed a false positive (and will be fixed in a future release).

[David Sosa]

It seems like:

float data[3*3] = {0};

should be accepted by exception and your answer would violate rule 9.2?

https://forum.misra.org.uk/thread-1357.html

[Anirban]

Thanks for pointing to the discussion. Using 0.0f would remove the Polyspace violation, but not address the intention behind the MISRA exception. This is indeed a false positive and will be fixed in a future release.

I edited the answer above to correct this.

[David Sosa]

Great! Thanks for you reply.

Cheers.