CodeWarrior for MCUのトピックRe: Check bits in the CCR

Check bits in the CCR

Chris_UTS — Tue, 10 Jun 2008 19:12:54 GMT

Hi,

I'm sorry if this has been answered before but i could not find a solution.

Background, using CodeWarrior 5.7, Chip is type HC12S and programming in C.

I'm performing a 16bit x 16bit multiplication and need to check if there is an overflow. I assume i should just check the relevant bits in the CCR but i'm not sure how to do this programmically.

Any help would be much appericiated.

Thank you.
Chris

Re: Check bits in the CCR

Lundin — Thu, 29 Oct 2020 09:09:07 GMT

You will indeed have to check the CCR through inline assembler, there is no obvious way to check for overflow in C.

I suppose you could do like this:

#include <limits.h>

uint16 a;
uint16 b;
uint32 a32 = a;
uint32 b32 = b;
uint32 result = a32 * b32;

if(result > UINT_MAX)
{
  /*  a*b will give overflow  */
}

Though the assembler version will be way more efficient.

Re: Check bits in the CCR

CompilerGuru — Thu, 29 Oct 2020 09:09:09 GMT

The HW of the S12 implements a 16x16=32 multiplication, therefore there is no overflow, and no
bit in the CCS to test.

Instead try a 16*16=32 bit multiplication in C and check if the resulting long fits into 16 bit.
If performance is critical, then note that not the multiplication is the expensive operation here but everything else .

Daniel

Code:

Code:#include <limits.h>unsigned int a,b,c;unsigned char overflow;void mul() {  unsigned long ab = (unsigned long)a * b;  if (ab > UINT_MAX) {    overflow = TRUE;  } else {    c = (unsigned int)ab;      }}

Re: Check bits in the CCR

Lundin — Wed, 11 Jun 2008 13:23:36 GMT

Right... there is no bit in the CCR, but you could clear index register Y before executing EMUL, then check if Y is different from zero afterwards. It would still have to be done in assembler.