From: Brendan Johnson <johnsobm@clarkson.edu>
Date: 05 Dec 2002 12:04:58 -0500
Some explicit questions I have are: After being processed by the IP
layer, what TCP function does it get handed up to? what forum does it
get handed up in, a struct? If so what are the fields?
I'm insanely pissed off when I see questions like this which show
that the person asking them has made absolutely no reasonable effort
on their part to figure out the answer on their own and wants other
people to do the work for you.
I don't think you're really reading any of the source code
at all if you still have questions like this. In fact I'm going
to demonstrate how little work is needed to answer your question.
If you did read the code, you'd see that ip_input calls
the handler indexed by protocal type in the array inet_protos[].
So you're next task would be to figure out how this inet_protos[]
array gets initialized, let's run grep on net/ipv4/*.c to see
where that happens:
[davem@nuts sparc-2.5]$ egrep inet_protos net/ipv4/*.c
net/ipv4/af_inet.c: struct inet_protosw *answer;
net/ipv4/af_inet.c: answer = list_entry(p, struct inet_protosw, list);
net/ipv4/af_inet.c:static struct inet_protosw inetsw_array[] =
net/ipv4/af_inet.c:#define INETSW_ARRAY_LEN (sizeof(inetsw_array) / sizeof(struct inet_p\rotosw))
net/ipv4/af_inet.c:void inet_register_protosw(struct inet_protosw *p)
net/ipv4/af_inet.c: struct inet_protosw *answer;
net/ipv4/af_inet.c: answer = list_entry(lh, struct inet_protosw, list);
net/ipv4/af_inet.c:void inet_unregister_protosw(struct inet_protosw *p)
net/ipv4/af_inet.c: struct inet_protosw *q;
net/ipv4/icmp.c: ipprot = inet_protos[hash];
net/ipv4/ip_input.c: if ((ipprot = inet_protos[hash]) != NULL) {
net/ipv4/protocol.c:struct inet_protocol *inet_protos[MAX_INET_PROTOS];
net/ipv4/protocol.c: if (inet_protos[hash]) {
net/ipv4/protocol.c: inet_protos[hash] = prot;
net/ipv4/protocol.c: if (inet_protos[hash] == prot) {
net/ipv4/protocol.c: inet_protos[hash] = NULL;
[davem@nuts sparc-2.5]$
It seems the only place this gets assigned is in net/ipv4/protocol.c
in the function "inet_add_protocol()" Let's see where that is invoked
for TCP, and we'd discover that this occurs during inet_init() as we
traverse the inet_protocol_base list, one of which is tcp_protocol in
net/ipv4/protocol.c which has "handler" member "tcp_v4_rcv()".
And you'd simply find tcp_v4_rcv in net/ipv4/tcp_ipv4.c, and you'd
thus see the arguments to that function and thus how the packets are
passed in by TCP input.
So I am going to ask you blatantly why this was such a difficult thing
to figure out on your own? This didn't require any specific kernel
knowledge. Rather it took a few minutes, using a few simple text
searching tools, some common sense and above all very basic C
programming knowledge.
Simply put, the answer is that you're too damn lazy to do the work to
figure this stuff out yourself and you want other people to do the
work for you. I simply can't respect such lazy people who leach off
mailing lists to get answers they can full well answer yourself.
-
: 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
