Redefining unsigned Char Boolean_type to True and false struct

Question

According to Autosar_SWS the boolean must be unsigned char. But I am having lots of MISRA violation in my compositions, like MISRA rule 10.1 (conversion violation), Rule 12.6 (effective boolean).

I would like to know if I redefine the BOOLEAN for my application like below:

#ifdefine BOOLEAN_T
#undefine BOOLEAN_T

typedef struct {
                 unsigned char TRUE  : 1;
                 unsigned char FALSE : 1;
               } BOOLEAN_T;    

#define TRUE 1;
#define False 0;

#endif

What will be the safety concerns and the consequences?

What is the purpose of defining BOOLEAN_T as a struct, rather than as unsigned char? — Markku K., Feb 07 '14 at 19:17
Well its functional safety software . I would like to strict the boolean value to 0,1 .. not to 0 and non-zero. and avoid any misra violations — user3285192, Feb 07 '14 at 19:42

score 1 · Answer 1 · answered Feb 07 '14 at 19:41

1

If you use a 2 bit struct to represent a single bit quantity, a random bit flip has a 50% chance of changing the true/false value to a value that is neither true nor false.

Please spend some more time studying the C language before attempting to write safety critical software.

answered Feb 07 '14 at 19:41

markgz

6,054
1
19
41

A **single** random bit flip has a 100% chance of changing a valid content to neither true nor false. Making a single bitflip detectable that way could be considered a feature... But if starting something like that, an ECC would be even more useful. – Yunnosch Jan 12 '18 at 09:45

score 1 · Answer 2 · edited Feb 22 '21 at 14:28

1

By your above method,there is a possibility of MISRA Warnings for using bifields over unsigned char.

I would suggest you to leave the definition of boolean to the platform guys of AUTOSAR (platform types) and concentrate on using boolean as an unsigned char.

edited Feb 22 '21 at 14:28

Roberto Caboni

7,252
10
25
39

answered Mar 05 '14 at 09:11

DarkKnight

131
10

score -1 · Accepted Answer · answered Feb 07 '14 at 19:39

-1

For safety you can use for example a bit pattern:

unsigned char data = 0x55 << (input_bit & 1);

switch (data):/* instead of if() */
 case 0xaa:
  /*true*/
  break;
 case 0x55:
  /*false*/
  break;
 case default:
  /*exception*/
  break;
}

answered Feb 07 '14 at 19:39

nopsoft

922
7
10

1

It would make a lot more sense to use an error correcting code to protect the data value against random bit flips. http://en.wikipedia.org/wiki/Error-correcting_code – markgz Feb 07 '14 at 19:42
It is usually true but if you don't have hardware ECC a software ECC/CRC is effective (speed of code) for data blocks - not for a single variable. – nopsoft Feb 07 '14 at 19:47
thanks .. my question not to stick to 1 bit. but more to strict to 0 and 1 #ifdefine BOOLEAN_T #undefine BOOLEAN_T typedef struct { unsigned char TRUE ; unsigned char FALSE ; } BOOLEAN_T; #define TRUE 1; #define False 0; #endif – user3285192 Feb 07 '14 at 19:47
@nopsoft: ECC can be easily implemented in software. An ECC code can be defined to protect a value of any size B bits from N errors for any values of B and N. – markgz Feb 07 '14 at 19:50
It is possible but let's notice code overhead for a single variable. It depends on your application. FEC is usually implemented in hardware (eg. DVBT). – nopsoft Feb 07 '14 at 19:53
Anyways thanks guys for the quick answers!! . There were too many different ways of defining the Boolean type in common usage, and no generally accepted single definition. Even MISRA-2004 don't have any clear definition. – user3285192 Feb 07 '14 at 20:02

Redefining unsigned Char Boolean_type to True and false struct

3 Answers3