Why does Polyspace report "MISRA C:2012 10.3 and 10.4"

In a C project I have a "typedef unsigned char BOOL; and true and false declared as BOOL" and in a function with parameter pointer to a BOOL, I do an assign *Pointer_To_BOOL = TRUE/FALSE and I get "MISRA C:2012 10.3 (Required)   The value of an expression shall not be assigned to an object with a narrower essential type or of a different essential type category. The expression is assigned to an object with a different essential type category." And also get "MISRA C:2012 10.4 (Required)   Both operands of an operator in which the usual arithmetic conversions are performed shall have the same essential type category. The left operand of the == operator has essentially unsigned type while the right operand has essentially Boolean type." where I check *Poiter_To_BOOL against TRUE/FALSE. Why does Polyspace take an parameter *Pointer_To_BOOL as unsigned char and what can I do to stop getting this warnings (refactor/justify).

Thank you, Cristian PASCALAU

[Walter Roberson]

How have you defined TRUE and FALSE?

[Cristian PASCALAU]

TRUE and FALSE are defined as (BOOL)(1U)/ (BOOL)(0U) and I have typedefed unsigned char BOOL.

[Lorenz Mende]

Hi Christian, did you already set the -boolean-types with your specific BOOL? If you have done and working with R2016b or lower, than there might be the chance that this is related to a bug and it is fixed in R2017a.

If a update is not possible, you may justify these violations in code -> see chapter "Add Review Comments to Code" in the documentation. It works well, our review process comes too with a justification comment which is detected by doxygen additionally.

Answer 1

[Anirban]

Since R2021a, the checkers 10.x treat macros such as TRUE and FALSE that resolve to 1 and 0 as essentially Boolean. See Polyspace release notes.

Answer 2

[Walter Roberson]

"Why does Polyspace take an parameter *Pointer_To_BOOL as unsigned char"

Because you defined BOOL as unsigned char. You deference a pointer to unsigned char so the result is going to be unsigned char

[Cristian PASCALAU]

Shouldn't it be derefenced to BOOL? As it is pointer to BOOL, which is typedef unsigned char BOOL.

[Walter Roberson]

No. You are using C. typedef does not create a new type in C, it only creates a convenient name for a type. unsigned char is the essential type, and all typedef to unsigned char are completely compatible in C.

I am not sure yet why you are getting the error message. I need to see how Pointer_To_BOOL was declared and I need to see the code for the line that has the == that is being complained about

[Cristian PASCALAU]

But MISRA 10.4 tells me "Both operands of an operator in which the usual arithmetic conversions are performed shall have the same essential type category. The left operand of the == operator has essentially unsigned type while the right operand has essentially Boolean type."

Here is an example of code:

static void DoSomething(BOOL *Pointer_To_BOOL_X, ... )

{
*Pointer_To_BOOL_X = TRUE;

/** here I get warning 10.3 The value of an expression shall not be assigned to an object with a narrower essential type or of a different essential type category. The expression is assigned to an object with a different essential type category. */

    if (*Pointer_To_BOOL_X == TRUE)/* here I get warning 10.4 */

    /** do something */

    }

Where:

- typedef unsigned char BOOL;

- TRUE (#define TRUE (BOOL)(1u))

- FALSE (#define FALSE (BOOL)(0u))

and function call is: DoSomething(&BOOL_X, ... )

where:

BOOL BOOL_X;

Thank you.

[JoseT]

Any comments on this issue, please update.