On Sun, Aug 11, 2013 at 08:41:41PM +0900, Joe Stringer wrote:
> On Sat, Aug 10, 2013 at 5:09 AM, Ben Pfaff <b...@nicira.com> wrote:
>
> > On Wed, Jul 31, 2013 at 09:31:55AM +0900, Joe Stringer wrote:
> > > +ovs_be32
> > > +crc32c(const uint8_t *data, size_t size)
> > > +{
> > > + uint32_t crc = 0xffffffffL;
> > > +
> > > + while (size--) {
> > > + crc = crc32Table[(crc ^ *data++) & 0xff] ^ (crc >> 8);
> > > + }
> > > +
> > > + /* The result of this CRC calculation provides us a value in the
> > reverse
> > > + * byte-order as compared with our architecture. On big-endian
> > systems,
> > > + * this is opposite to our return type. So, to return a big-endian
> > > + * value, we must swap the byte-order. */
> > > +#if defined(WORDS_BIGENDIAN)
> > > + crc = get_unaligned_u32(&crc);
> > > +#endif
> >
> > I don't see how this does any kind of byteswapping (as the comment
> > claims). Just treating 'crc' as unaligned (&crc should in fact be
> > properly aligned) has no effect. As far as I can tell it is
> > equivalent to "crc = crc;"
>
>
> Ah, right you are. I think I got a bit lost on my way to a byteswapping
> function. I'll fix this up. Unfortunately I don't have a system to test
> this on.
That's OK. I'm more concerned about letting in obvious bugs than
testing, in this case.
Thanks,
Ben.
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
