MISRA Discussion Forums
Rule 2.1 - Printable Version

+- MISRA Discussion Forums (https://forum.misra.org.uk)
+-- Forum: MISRA C (https://forum.misra.org.uk/forumdisplay.php?fid=4)
+--- Forum: MISRA-C: 2004 rules (https://forum.misra.org.uk/forumdisplay.php?fid=17)
+---- Forum: 6.2 Language Extensions (https://forum.misra.org.uk/forumdisplay.php?fid=29)
+---- Thread: Rule 2.1 (/showthread.php?tid=217)

Rule 2.1 - nikunj - 16-02-2006

What are the possible scenarios in which Rule 2.1 will be violated?

Rule allows for Macros, C-functions and Assembly functions.

If any assembly code is used directly it will be reported as an compliler error for undefined identifier etc.

- misra-c - 27-04-2006

C statements and assembler statements should not co-exist in a function unless the assembler is encapsulated in a macro.

The following example violates this rule.

Code:
static void foo ( void )
{
   asm { \"CLI\" };
   foobar();
   asm { \"SEI\" };
}


- nikunj - 03-05-2006

Do the following code snippets violate Rule 2.1?

1. macro encapsulation:
#pragma asm
MOV A,B
....
....
#pragma endasm

2.function call
asm(\"CLI\");

or

move(\"MOV A,B\");


Also statements like

asm{ \"CLI\" };

seem to be compiler specific.
What are the other possible ways of violating this rule (compiler specific/not specific).

Also, please can you give examples where the rule is not violated.

- lv - 05-05-2006

Quote:Also, please can you give examples where the rule is not violated.

In my understanding, there are the three solutions compliant with MISRA 2.1:

a) Assembler function:
nop.h:
Code:
extern void nop(void);

nop.asm:
Code:
NOP   PROC
   nop
   ret
NOP  ENDP

foo.c
Code:
#include \"nop.h\"

static void function(void)
{
   ... /* do somthing */
    nop();
   ... /* do somthing */

}


b) C-function encapsulation
nop.h:
Code:
extern void nop(void);


nop.c:
Code:
#include \"nop.h\"

void nop(void)
{
/* No code between bracket and pragma */
#pragma ASM
   nop
#pragma ENDASM
/* No code between pragam and bracket */
}

/* No other function in nop.c, except other ASM encapsulation. */

foo.c:
Code:
#include \"nop.h\"

static void function(void)
{
   ... /* do somthing */
    nop();
   ... /* do somthing */

}


In this solution nop.c is not MISRA but the code is isolated.
Foo.c is MISRA.


c) macro-function

nop.h:
Code:
#define nop()\   asm(\"nop\")

/* No other definition in nop.h file, except other assembleur encapsulation */

foo.c:
Code:
#include \"nop.h\"

static void function(void)
{
   ... /* do somthing */
    nop();
   ... /* do somthing */

}

The last solution could have an impact on OBJ file generation.
Some (most of?) compiler requier to compile with a specific option when assembler code is use in C file.
This last solution should compliant with MISRA 2.1 but not with MISRA 1.1
(inline assembler is not C standard).

- misra-c - 23-08-2006

MISRA-C meeting 23-8-2006

The previous reply by lv is consistent with our views.