Thread: Simulating C code

I am trying to simulate the following code in excel, with cells as bits. It would be nice to just line up different rows (Data, FrameCheckSequence, GeneratorPolynomial, etc) and then do the bit operations on the cells (columnwise), shifting until one arrives at the final result. When I do this, however, I don't seem to obtain the desired answer. Trying to figure out why not.

CRC11GenerateFrameCheckSequence
/*
Inputs: ByteArrayPtr is the address of a 13 bit array holding 102 bits of data. The two most significant bits must be 0.

Outputs: 11 bit frame check sequence.
*/

Code:

Extern unsigned short CRC11GenerateFrameCheckSequence( unsigned char * ByteArrayPtr )
{ 
   unsigned short     GeneratorPolynomial = 0x0F35;
   unsigned short     FrameCheckSequence = 0x07FF;
   unsigned short     Data;
   int                        ByteIndex, Bit;

   /* Do most significant byte skipping the 2 most significant bits */
   Data =  *ByteArrayPtr << 5;
   ByteArrayPtr++

   for ( Bit = 2;  Bit < 8;  Bit++ )
  {
      if  ( (FrameCheckSequence ^ Data) & 0x400 )  
         FrameCheckSequence = (FrameCheckSequence << 1) ^ GeneratorPolynomial;
      else
          FrameCheckSequence = (FrameCheckSequence << 1);
      FrameCheckSequence &=  0x7FF;
      Data <<=1;
  }

   /* Do rest of the bytes */
   for ( ByteIndex = 1; ByteIndex < 13; ByteIndex++ )
   {
      Data = *ByteArrayPtr << 3;
      ByteArrayPtr++;
      for ( Bit = 0; Bit < 8; Bit++ )
      {
         if (  (FrameCheckSequence ^ Data) & 0x0400 )  
            FrameCheckSequence = (FrameCheckSequence << 1) ^ GeneratorPolynomial;
         else
            FrameCheckSequence = (FrameCheckSequence << 1);
         FrameCheckSequence &=  0x7FF;
         Data <<=1;
      }    
    }
    return FrameCheckSequence;

Originally Posted by tabstop

Haven't gotten very far in reading your code, but this can't be right:

Code:

 /* Do most significant byte skipping the 2 most significant bits */
   Data =  *ByteArrayPtr << 5;

Edit: Oh, wait, maybe it can if the comment doesn't have anything to do with the next bit of code. Well, in that case I would double-check your algorithm.

The C code is not mine, so I have to interpret it also. If the C code itself has problems (or even typos), that would be a start.

Hang on. What version (if any) is the "working, correct" version and what version (if any) is the version you need to fix.

Well, I don't know what you've done. I'm guessing/hoping you know what all the symbols mean? If you're doing things "by hand" you don't really have to worry (much) about the << 3 and the 0x400, as I think that's there just to get things away from the sign bit and to line things up so that you get 11 bits of answer (so you can just interpret it bit by bit as you're looking at it in the cells as long as you realize your answer is always 11 bits, I think).

Originally Posted by tabstop

Well, I don't know what you've done. I'm guessing/hoping you know what all the symbols mean? If you're doing things "by hand" you don't really have to worry (much) about the << 3 and the 0x400, as I think that's there just to get things away from the sign bit and to line things up so that you get 11 bits of answer (so you can just interpret it bit by bit as you're looking at it in the cells as long as you realize your answer is always 11 bits, I think).

Here is an example of what I did: FCS is Frame Check sequence.

Data: . . . . . . . . 0 0 0 0 0 0 0 0 1 0 1 1 0 0 0 0 1 0 1 0 1 1 0 0 0 1 1 . . . . . . . . . . . . .etc. (102 bits long)
FCS: . . . . . . . . 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 1 (16 bits long, initial value = 0x07FF)

So when it says FCS ^ Data, I do an XOR col by col for 16 cols. Not sure if this is what the c code says, or if I need the four leading zeros on the FCS to begin with. After that, it says & 0x400. That will single out one col, but is 0x400 = 010000000000 (12 bits long)?

I start out by hand just to get the logic, then write a macro to actually evaluate the final result.

Doesn't matter whether you think of it as 12 or 16 bits, I don't think, as the first four should always be zero either way.

^ is xor, right, and & is and. So you don't need to do column by column, you just need to check that one digit that you and with. If it's a 1, shift over and xor with the polynomial thing; if it's a zero, just shift over.

Originally Posted by tabstop

Doesn't matter whether you think of it as 12 or 16 bits, I don't think, as the first four should always be zero either way.

^ is xor, right, and & is and. So you don't need to do column by column, you just need to check that one digit that you and with. If it's a 1, shift over and xor with the polynomial thing; if it's a zero, just shift over.

Agreed. So you think it can look like this:

Data: . . . . . . . . 0 0 0 0 0 0 0 0 1 0 1 1 0 0 0 0 1 0 1 0 1 1 0 0 0 1 1 . . . . . . . . . . . . .etc. (102 bits long)
FCS: . . . . . . . . 0 1 1 1 1 1 1 1 1 1 1 1 (12 bits long, initial value)
0x400 . . . . . . . . 0 1 0 0 0 0 0 0 0 0 0 0 (12 bits long)

and the one col I check is the 2nd col? And when I shift and xor with the generator, I should do a 12 col xor?

Generator = 1 1 1 1 0 0 1 1 0 1 0 1 = F35?

Originally Posted by penrose

Agreed. So you think it can look like this:

Data: . . . . . . . . 0 0 0 0 0 0 0 0 1 0 1 1 0 0 0 0 1 0 1 0 1 1 0 0 0 1 1 . . . . . . . . . . . . .etc. (102 bits long)
FCS: . . . . . . . . 0 1 1 1 1 1 1 1 1 1 1 1 (12 bits long, initial value)
0x400 . . . . . . . . 0 1 0 0 0 0 0 0 0 0 0 0 (12 bits long)

and the one col I check is the 2nd col? And when I shift and xor with the generator, I should do a 12 col xor?

Generator = 1 1 1 1 0 0 1 1 0 1 0 1 = F35?

Still can't make my answers agree with the "official" answer. Somehow my variables are not correctly aligned or not the right length or something else is going on.