On Thu, Nov 13, 2003 at 08:54:28AM +0100, francois donzet wrote: > > The problem is i found nowhere in the code the clue > that skb->csum is filled with the checksum on > tcpheader+data. (i know that after it is complemented > with pseudo header checksum). > Yes, you 're right. There's no mention in the comments in the driver source that *only* the trarsport-layer-PDU (TCP header + data) is checksumed by the hardware. But a sum of the *complete* packet contents (IP header + TCP header + data) sounds meaningless to me, as this is *neither* an IP checksum, *nor* a TCP one! I guess that someone who is more familirial with the e100 hardware will be kind enough to clarify things a bit, and give a more authoritative answer to this: What part of an *incoming* frame carrying a TCP packet is really checksumed by the hardware?? If all of it does (minus the ethernet header), as seems to be suggested in the driver comments, then how can this checksum be anything meaningfull since it is neither an IP nor a TCP checksum? Until an answer is given, though, or more corroborating evidence are collected, I would stand by my explanation, that only the transport-layer PDU is checksumed by the hardware. /npat -- The human mind ordinarily operates at only ten percent of its capacity---the rest is overhead for the operating system. - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
