Em Wed, Jan 07, 2009 at 12:27:22PM +0800, shizibo escreveu:
> Hello, I am just studying the codes of DCCP, I encounter some
> questions now. I send this mail to <dccp@xxxxxxxxxxxxxxx>, but the
> yahoo mail box inform me that it can not be delivered.So sorry to
> bother you.Will you help me.
No problem, I'm replying to you and to the list, so that the answers can
help others in the future.
> in dccp_connect():
> skb = alloc_skb(sk->sk_prot->max_header, sk->sk_allocation);
> where is the sk->sk_prot->max_header set before it's used here?
Use grep:
[acme@doppio linux-2.6]$ grep "max_header.\+=" net/dccp/*.c
net/dccp/ipv4.c: .max_header = MAX_DCCP_HEADER,
net/dccp/ipv6.c: .max_header = MAX_DCCP_HEADER,
[acme@doppio linux-2.6]$
I.e. it depends on the layer 3 used. Now look at net/dccp/ipv4.c to
see exactly how it is initialized:
static struct proto dccp_v4_prot = {
<SNIP>
.max_header = MAX_DCCP_HEADER,
<SNIP>
};
And this, in turn, is referenced here:
static struct inet_protosw dccp_v4_protosw = {
.type = SOCK_DCCP,
.protocol = IPPROTO_DCCP,
.prot = &dccp_v4_prot,
.ops = &inet_dccp_ops,
.capability = -1,
.no_check = 0,
.flags = INET_PROTOSW_ICSK,
};
If you now search where dccp_v4_protosw is used:
static int __init dccp_v4_init(void)
<SNIP>
inet_register_protosw(&dccp_v4_protosw);
<SNIP>
}
inet_register_protosw will let net/ipv4/af_inet.c that when a PF_INET
socket of type SOCK_DCCP is asked for we should use dccp_v4_prot, etc.
Please take a look at a paper a wrote that explains these details:
http://ols.fedoraproject.org/OLS/Reprints-2005/melo-Reprint.pdf
Look at chapter 3.
> and, there is macro:
> #define DCCP_SKB_CB(__skb) ((struct dccp_skb_cb *)&((__skb)->cb[0]))
> It is said the skb->cb is used to hold some private control message
> per layer. For dccp, where is it initialized? What does the other
> bytes, cb[1], cb[2]..., content?
This is a scratch pad, always casted to some struct that is layer
specific, for instance, for DCCP, it is 'struct dccp_skb_cb', that is
defined in net/dccp/dccp.h:
/**
* dccp_skb_cb - DCCP per-packet control information
* @dccpd_type: one of %dccp_pkt_type (or unknown)
* @dccpd_ccval: CCVal field (5.1), see e.g. RFC 4342, 8.1
* @dccpd_reset_code: one of %dccp_reset_codes
* @dccpd_reset_data: Data1..3 fields (depend on @dccpd_reset_code)
* @dccpd_opt_len: total length of all options (5.8) in the packet
* @dccpd_seq: sequence number
* @dccpd_ack_seq: acknowledgment number subheader field value
* This is used for transmission as well as for reception.
*/
struct dccp_skb_cb {
union {
struct inet_skb_parm h4;
#if defined(CONFIG_IPV6) || defined (CONFIG_IPV6_MODULE)
struct inet6_skb_parm h6;
#endif
} header;
__u8 dccpd_type:4;
__u8 dccpd_ccval:4;
__u8 dccpd_reset_code,
dccpd_reset_data[3];
__u16 dccpd_opt_len;
__u64 dccpd_seq;
__u64 dccpd_ack_seq;
};
So we don't access, say, skb->cb[5], but instead we cast it to a
variable (or use it directly as DCCP_SKB_CB(skb)->dccpd_type), like in
dccp_v4_rcv in net/dccp/ipv4.c:
static int dccp_v4_rcv(struct sk_buff *skb)
{
const struct dccp_hdr *dh;
<SNIP>
dh = dccp_hdr(skb);
DCCP_SKB_CB(skb)->dccpd_seq = dccp_hdr_seq(dh);
DCCP_SKB_CB(skb)->dccpd_type = dh->dccph_type;
<SNIP>
So, in this case, it is initialized when we receive a packet.
Later, in the same function (dccp_v4_rcv), if we don't find a socket to
queue this packet into, we use the skb->cb area in another fashion, this
time setting dccpd_reset_code to the sensible value and calling a
routine (dccp_v4_ctl_send_reset) that will then get this value and do
something it it:
no_dccp_socket:
if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
goto discard_it;
/*
* Step 2:
* If no socket ...
* Generate Reset(No Connection) unless P.type ==
* Reset
* Drop packet and return
*/
if (dh->dccph_type != DCCP_PKT_RESET) {
DCCP_SKB_CB(skb)->dccpd_reset_code =
DCCP_RESET_CODE_NO_CONNECTION;
dccp_v4_ctl_send_reset(sk, skb);
}
- Arnaldo
--
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
