Why fragmented on L3 and then linearized on L2
Hi,
Today i read 2.6.18 kernel code,during data transmiting, on L3 kernel use ip_fragments() to div packet to smaller if it bigger than MTU, and soon ,on L2--function dev_queue_xmit() use skb_linearized() gather frag_list data to the first skb linear area .
why kernel does like this, i mean slice and gather so crowded?
Last edited by oscareee888; 12-15-2014 at 07:04 PM.
|
