Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!necntc!ames!sdcsvax!ucbvax!imagen.UUCP!geof
From: geof@imagen.UUCP (Geof Cooper)
Newsgroups: comp.protocols.tcp-ip
Subject: TCP checksum unrolling
Message-ID: <8710062308.AA00107@apolling.imagen.uucp>
Date: Tue, 6-Oct-87 19:08:57 EDT
Article-I.D.: apolling.8710062308.AA00107
Posted: Tue Oct  6 19:08:57 1987
Date-Received: Fri, 9-Oct-87 23:51:58 EDT
Sender: daemon@ucbvax.BERKELEY.EDU
Reply-To: imagen!geof@decwrl.dec.com
Distribution: world
Organization: The ARPA Internet
Lines: 62

Here is one, undebugged, that illustrates the concept.  It
also uses the trick that (if you have it) you can use 32-bit
two's complement addition and add all the carries in at the
end (another trick that is sometimes faster is to generate
a 32-bit one's complement sum and then add the top and bottom
halves together to get the 16-bit sum).  Some C compilers
won't accept the wierd syntax below; or maybe I should point
out, as you wretch on the floor, that there is at least ONE
c compiler that DOES accept this syntax.

It is trivial to code it for all C compilers -- but
what you really want to do is code the exact intent of the
following into assembly language.  That makes it a lot faster
to add the two halves of a 32-bit word.

These tricks don't work for XNS checksums.  Our experience is
that this difference alone makes our XNS implementation a little
slower than our TCP implementation on a 68000.

- Geof

checksum(p, n)
    unsigned short *p;
    short n;
{
    short nloop;
    short nrem;
    unsigned long sum;

    sum = 0;
    if ( n > 0 ) {
        nloop = (n >> 3) + 1;
        nrem  = n & 7;

        switch ( nloop ) {

            do {
                    sum += *p++;
                case 7:
                    sum += *p++;
                case 6:
                    sum += *p++;
                case 5:
                    sum += *p++;
                case 4:
                    sum += *p++;
                case 3:
                    sum += *p++;
                case 2:
                    sum += *p++;
                case 1:
                    sum += *p++;
                case 0:
            } while ( --nloop > 0 );
        }
    }

    sum = (sum >> 16) + (sum & 0xffff);
    sum = (sum >> 16) + (sum & 0xffff);

    return ( sum );
}