On Mon, 2017-09-18 at 14:41 +1000, Daniel Axtens wrote: > Hi Eric, ... > So I've been experimenting with this and reading through the core > networking code. If my understanding is correct, disabling GSO will > cause the packet to be segmented, but it will be segemented into > gso_size+header length packets. So in this case (~10kB gso_size) the > resultant packets will still be too big - although at least they don't > cause a crash in that case.
You describe a bug in core networking stack then. When we perform software segmentation, we must do the check against route mtu, and drop the offending frame, and send an ICMP back eventually.