Misra-c 10.3 violation with enum boolean only with stub variables

I have an embedded system where a variable structure is passed to my program. This underlying code is not available so to initialize these variables I'm using a stub file. I have a header file that is included in the stub (and the code being tested). A small portion of the header is:

I'm using

SOMEDATA.H

typedef enum TAG_BOOLEAN
{
   BOOLEAN_FALSE,
   BOOLEAN_TRUE 
} BOOLEAN;
typedef struct SOME_DATA_GLOBAL_TAG
{
    UI_8    FirstMessage [ 8U ] ;
   UI_8   AnotherVar;
   BOOLEAN EEPROMIsBusy ;
} SOME_DATA_GLOBAL;

typedef struct SOME_DATA_TAG
{
   SOME_DATA_GLOBAL   Global ;
} SOME_DATA ;

SOMEDATA_STUB.C

#include SOMEDATA.H
volatile SOME_DATA  m_Data ;

void Some_InitStubs ( void )
{
  m_Data.Global.FirstMessage = 0U;
       m_Data.Global.AnotherVar = 0U;
       EEPROMIsBusy = BOOLEAN_FALSE;  also tried EEPROMIsBusy = (BOOLEAN)BOOLEAN_FALSE;
    This line produces a MISRA-C:2012 10.3 violation
}
SOME_DATA * SomeData_GetpData ( void )
{
   return & m_Data ;
}

In code under test

TESTCODE.C

#include SOMEDATA.H

void CheckSomeData(void)
{
  SOME_DATA * pData ;
  BOOLEAN internalBoolean;

  pData = SomeData_GetpData();

  pData->Global.EEPROMIsBusy = BOOLEAN_TRUE;
   This line produces a MISRA-C:2012 10.3 violation
       internalBoolean = BOOLEAN_FALSE;
        This line does not produce a violation.
}

Bug Finder does not flag any assignment of BOOLEAN_TRUE or BOOLEAN_FALSE as an error except with variables used in a stub. In fact, all enum assignments to stub variables are flagged with a 10.3 violation. None of the other code is flagged even though the enum's are used extensively in the code.

How do I fix this?

[Walter Roberson]

It would probably make more sense if you changed the "This line..." to comments on the appropriate line.

Answer 1

[Sandi Southard]

Thanks, I'll do that on future posts. Is there any answer for me?

Answer 2

[Walter Roberson]

The line

EEPROMIsBusy = BOOLEAN_FALSE;

is assigning to a variable that has no type declaration in scope, so if it were permitted at all it would be an implicit int and int has a different essential type than enum does.

[Sandi Southard]

Sorry, I must have made a mistake in the original post. in SOMEDATA_STUB.C the line should read: m_DATA.Global.EEPROMIsBusy = BOOLEAN_FALSE;

This variable is in scope with a typedef of BOOLEAN.

I still don't understand why only variables of type ENUM that are defined within the stub have violations.

[Walter Roberson]

https://www.misra.org.uk/forum/viewtopic.php?t=1091 has a relevant discussion.

I would need to look it over more closely but at the moment it seems like it might be implying that the storage type of an enumeration value might be narrower than the type of an enumeration constant, which sounds a bit odd to me. Please read it over and see if you conclude the same thing: if the smallest type can be used for storage but enumeration constants are int then doesn't that imply that the constant could be wider than the destination?