V2621. MISRA. Tag names should be unique across all name spaces.

This diagnostic rule is based on the MISRA (Motor Industry Software Reliability Association) manual for software development.

This rule only applies to C. The names of structures, enumerations, and unions must be unique for all namespaces and blocks. Reusing name tags can confuse the developer.

Look at the example:

int foo()
{
  {
    struct MyStuct
    {
      unsigned char data;                      // (1)
    };

    struct MyStuct sample  = { .data = 250 };  // ok 
  }

  // ....

  {
    struct MyStruct
    {
      signed char data;                        // (2)
    };

    struct MyStruct sample = { .data = 250 };  // error
  }
}

The 'unsigned char' type allows values from 0 to 255, and the 'signed char' type allows values from -128 to 127. After working with the first 'MyStruct' structure, the developer may get used to the fact that the member of the 'data' structure has the 'unsigned char' type. After that, the developer can easily make a mistake in the second block by writing the 'sample.data' value. This leads to the signed integer overflow.

Fixed example:

int foo()
{
  {
    struct MyStuctUnsigned
    {
      unsigned char data;                              // (1)
    };

    struct MyStuctUnsigned sample  = { .data = 250 };  // ok 
  }

  // ....

  {
    struct MyStructSigned
    {
      signed char data;                              // (2)
    };

    struct MyStructSigned sample = { .data = 127 };  // ok
  }
}

Here the names of the structures are different. Thus, it's more difficult to make a mistake.

The type alias declared via 'typedef' can duplicate the name when the developer declares 'struct', 'union' or 'enum' if they are associated with this 'typedef':

typedef struct list
{
  struct list* next;
  int element;
} list;               // ok

This diagnostic is classified as: