Online CRC Calculation

Be careful: there are several ways to realize a CRC. They differ (at least) in the way which bit is shifted in first and also in the initialization of the flipflops.

A typical hardware implementation (LFSR - Linear Feedback Shift Register) is shown here:

Dr.-Ing. K. Gorontzi, 2005

The input bits are shifted into the very left XOR gate. The MSB (leftmost bit) of each byte is shifted in first.

Each flipflop represents a single CRC output bit. The leftmost flipflop is the MSB of the CRC. This implementation doesn't need to augment the serial input message with zeros.

Note that in our case the flipflops are cleared to zeros at the beginning of each CRC calculation.

A simple VERILOG implementation of the above polynom is shown here. You can directly copy the source snippet to your code (distributed under LGPL):

// ==========================================================================
// CRC Generation Unit - Linear Feedback Shift Register implementation
// (c) Kay Gorontzi, GHSi.de, distributed under the terms of LGPL
// ==========================================================================
module CRC_Unit(BITVAL, BITSTRB, CLEAR, CRC);
   input         BITVAL;                            // Next input bit
   input         BITSTRB;                           // Current bit valid (Clock)
   input         CLEAR;                             // Init CRC value
   output [7:0] CRC;                               // Current output CRC value

   reg    [7:0] CRC;                               // We need output registers
   wire          inv;
   
   assign inv = BITVAL ^ CRC[7];                   // XOR required?
   
   always @(posedge BITSTRB or posedge CLEAR) begin
      if (CLEAR) begin
         CRC = 0;                                  // Init before calculation
         end
      else begin
         CRC[7] = CRC[6];
         CRC[6] = CRC[5];
         CRC[5] = CRC[4];
         CRC[4] = CRC[3] ^ inv;
         CRC[3] = CRC[2] ^ inv;
         CRC[2] = CRC[1] ^ inv;
         CRC[1] = CRC[0];
         CRC[0] = inv;
         end
      end
   
endmodule

A simple C implementation of the above polynom is shown in the following code. Again, you can directly copy the source snippet to your code (distributed under LGPL):

// ==========================================================================
// CRC Generation Unit - Linear Feedback Shift Register implementation
// (c) Kay Gorontzi, GHSi.de, distributed under the terms of LGPL
// ==========================================================================
char *MakeCRC(char *BitString)
   {
   static char Res[9];                                 // CRC Result
   char CRC[8];
   int  i;
   char DoInvert;
   
   for (i=0; i<8; ++i)  CRC[i] = 0;                    // Init before calculation
   
   for (i=0; i<strlen(BitString); ++i)
      {
      DoInvert = ('1'==BitString[i]) ^ CRC[7];         // XOR required?

      CRC[7] = CRC[6];
      CRC[6] = CRC[5];
      CRC[5] = CRC[4];
      CRC[4] = CRC[3] ^ DoInvert;
      CRC[3] = CRC[2] ^ DoInvert;
      CRC[2] = CRC[1] ^ DoInvert;
      CRC[1] = CRC[0];
      CRC[0] = DoInvert;
      }
      
   for (i=0; i<8; ++i)  Res[7-i] = CRC[i] ? '1' : '0'; // Convert binary to ASCII
   Res[8] = 0;                                         // Set string terminator

   return(Res);
   }

// A simple test driver:

#include <stdio.h>

int main()
   {
   char *Data, *Result;                                       // Declare two strings

   Data = "1101000101000111";
   Result = MakeCRC(Data);                                    // Calculate CRC
   
   printf("CRC of [%s] is [%s] with P=[100011101]\n", Data, Result);
   
   return(0);
   }

Of course, the software is provided here 'as is' with no expressed or implied warranties at all.

Do you consider this page somewhat useful? Please drop me a note via

$ fd	(hexadecimal)
% 11111101	(binary, see calculation details here)
! 253	(decimal)