On 01/02/11 01:56, Philip Pemberton wrote:
> I've spent all day hacking away on an implementation of the
> shift-register ECC engine on the WD1100, but it isn't matching up, no
> matter what I do...

Well, isn't *this* interesting... the NSC appnote seems to be wrong.

philpem@cheetah:~/dev/mt-poly-search $ ./mt-poly-search 4
running 4 crcscan threads, using EXHAUSTIVE_SEARCH method
HIT: found polynomial match, 0x140A0445
HIT: found polynomial match, 0x4FE4A171

So the NSC appnote seems to have the wrong binary representation=20
specified for the polynomial. Let's look at these in a bitwise form:

140A0443 =3D 0b00010100 00001010 00000100 01000011 (from appnote)
140A0445 =3D 0b00010100 00001010 00000100 01000101
                                              ^^

Working backwards from the polynomial using SpeedCrunch:
x=3D2
   x=3D2
hex(x^32 + x^28 + x^26 + x^19 + x^17 + x^10 + x^6 + x^2 + 1)
   0x1140A0445

Ho hum. I've just wasted a day of debugging because an appnote author=20
couldn't count. C'est la vie!

And now the code:

uint32_t POLY =3D 0x140A0445;

uint32_t crc32(uint32_t initval, unsigned char * data, size_t len)
{
	uint32_t crc =3D initval;
	size_t i =3D 0;

	while (i < len) {
		unsigned char x =3D data[i++];
		crc ^=3D (x << (32-8));

		for (int j=3D0; j<8; j++) {
			if (crc & 0x80000000)
				crc =3D (crc << 1) ^ POLY;
			else
				crc =3D crc << 1;
		}
	}

	return crc;
}

--=20
Phil.
piclist@philpem.me.uk
http://www.philpem.me.uk/
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.