Andrew Gallatin wrote:
> Garrett D'Amore writes:
> >
> > Yes, but the API lacks the ability to express this distinction in the
> > capability bits. So far, I've only run into problems for hardware on
> > receive... for hme and eri, they have the ability to put a different
> > start offset on tx. But for rx, you don't know in advance if the frame
> > is tagged or not, so we just program in the offset for non-tagged
> > frames. If the frame shows up tagged, then we drop the partial checksum
> > data before we send the packet upstream.
>
> Why don't you just fix the checksum by "subtracting" off the last 4
> bytes of the expanded vlan header (bytes 14..17 of the frame).
>
I had considered that, but done it mostly because I am not entirely sure
that this is universally "safe". (This is due to ignorance on my part,
and I'd love to be better informed.)
There are two complexities in the checksum calculation that I'm
concerned about. The first is the inclusion of carry bits. The second
is the negation (~). I'm not entirely sure after these are done, that
the operation is reversible enough to safely subtract those bytes.
Perhaps I need to look at this a bit more closely. Getting checksum
offload on the receive side for such hardware would be nice.
-- Garrett
_______________________________________________
networking-discuss mailing list
[email protected]
