|
Size: 820
Comment:
|
Size: 926
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 7: | Line 7: |
| Given a buffer that is a multiple of 16 bits such as an IP packet header, create a sum and add to it the buffer 16 bits at a time. After each add, check to see if we have carried beyond 16 bits. If we have chop off the carry and increment the sum. After you have add all the 16 bit parts of the buffer, make sure the result is only 16 bits and return its complement. | The IpCheckSum uses the OnesComplement representation of positive and negative numbers. Given a buffer that is a multiple of 16 bits such as an IP packet header, create a OnesComplement sum of each of the 16 bit words. After you have added all the 16 bit words of the buffer, make sure the result is only 16 bits and ''return its complement''. This is the value added to the checksum field. Now when you add all the pieces, you should get -0 in 1s complement parlance. |
Back to ComputerTerms
IP Checksum
Process
The IpCheckSum uses the OnesComplement representation of positive and negative numbers.
Given a buffer that is a multiple of 16 bits such as an IP packet header, create a OnesComplement sum of each of the 16 bit words. After you have added all the 16 bit words of the buffer, make sure the result is only 16 bits and return its complement. This is the value added to the checksum field. Now when you add all the pieces, you should get -0 in 1s complement parlance.
The Algorithm
cksum(u_short *buf, int count) {
register u_long sum = 0;
while(count--) {
sum += *buf++ /* this adds the 16 bit value of *buf and then increments the pointer */
if (sum & 0xFFFF0000) { /* if we carried a bit over the 16 bit limit */
sum &= 0xFFFF;
sum++;
}
}
return ~(sum & 0xFFFF);Back to ComputerTerms