Re: The TCP and UDP checksum algorithm may soon need updating

[Joseph Touch <touch@xxxxxxxxxxxxxx>]

+1

On Jun 7, 2020, at 11:37 AM, Phillip Hallam-Baker <phill@xxxxxxxxxxxxxxx> wrote:

Rather than trying to make TCP/IP a flawless transport, we have to apply the same principle of making a lossy channel robust at the higher level.

My other concern is also with offload mechanisms and middleboxes. There’s the old adage of whether we’re corruption errors (line, memory) or manipulation errors (incorrect code or assumptions about the inputs to that code).

Overall, I’d feel a lot better about upending transport checksums if we had evidence that the checksum wasn’t catching errors. If the checksum is correct because it’s being constantly recomputed without being checked, a new alg won’t fix the issue.

And if this were truly a 1 in 100 problem, wouldn’t we be seeing a lot more of it?

If somethings fishy here, it’s not clear at all yet what it is.

Joe