From: "Constantinos Makassikis" <cmakassikis@xxxxxxxxx> Date: Mon, 15 May 2006 19:20:05 +0200 > Is there a way to predict the amount of overhead or the in-band TCP > data that can be hold ? It depends upon the machine, whether things like SMP and other networking features are enabled, how the driver allocates packets (for receive) and other variables. For each packet there is "sizeof(struct sk_buff)" for each send packet + each clone made of that packet (there is at least one clone of every transmitted packet made for TCP, so we can keep the master on the retransmit queue and send clones to the output path). Plus a "sizeof(struct skb_shared_info)" for each instance of the data. We try to make approximations in the TCP stack, see the net.ipv4.tcp_adv_win_scale sysctl and how the kernel uses that for example. - : send the line "unsubscribe linux-net" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
