Description of linuxkernel2.4.54254socket Layer

Article Title: a brief explanation of linuxkernel2.4.54254socket layer. Linux is a technology channel of the IT lab in China. Includes basic categories such as desktop applications, Linux system management, kernel research, embedded systems, and open source.
1. Create a socket
Original function:
Static int inet_create (struct socket * sock, int protocol)
In net/ipv4/af_inet.c
Detailed explanation
Static int inet_create (struct socket * sock, int protocol)
{
Struct sock * sk;
Struct proto * prot;
Sock-> state = SS_UNCONNECTED;/* set the status to not connected */
Sk = sk_alloc (PF_INET, GFP_KERNEL, 1);/* apply for memory required by sock */
/* Net/core/sock. c */
If (sk = NULL)
Goto do_oom;
Switch (sock-> type ){
Case SOCK_STREAM:/* TCP protocol */
If (protocol & protocol! = IPPROTO_TCP)
Goto free_and_noproto;
Protocol = IPPROTO_TCP;
Prot = & tcp_prot;/* tcp_prot defined in net/ipv4/tcp_ipv4.c */
Sock-> ops = & inet_stream_ops;/* socket operations on STREAM */
Break;
Case SOCK_SEQPACKET:/* not supported */
Goto free_and_badtype;
Case SOCK_DGRAM:/* UDP protocol */
If (protocol & protocol! = IPPROTO_UDP)
Goto free_and_noproto;
Protocol = IPPROTO_UDP;
Sk-> no_check = UDP_CSUM_DEFAULT;
Prot = & udp_prot;/* udp_prot defined in net/ipv4/udp. c */
Sock-> ops = & inet_dgram_ops;/* socket operation on DGRAM */
Break;
Case SOCK_RAW:/* RAW */
If (! Capable (CAP_NET_RAW)/* determine whether there is a right to establish SOCK_RAW */
Goto free_and_badperm;
If (! Protocol)/* protocol cannot be 0 */
Goto free_and_noproto;
Prot = & raw_prot;/* raw_prot defined in net/ipv4/raw. c */
Sk-> reuse = 1;/* allow address reuse */
Sk-> num = protocol;
Sock-> ops = & inet_dgram_ops;/* some features of raw are the same as those of DGRAM */
If (protocol = IPPROTO_RAW)
Sk-> protinfo. af_inet.hdrincl = 1;
/* Allow custom ip headers */
Break;
Default:
Goto free_and_badtype;
}
If (ipv4_config.no_pmtu_disc)
Sk-> protinfo. af_inet.pmtudisc = IP_PMTUDISC_DONT;
Else
Sk-> protinfo. af_inet.pmtudisc = IP_PMTUDISC_WANT;
Sk-> protinfo. af_inet.id = 0;
Sock_init_data (sock, sk);/* initialize some data */
/* Net/core/sock. c */
Sk-> destruct = inet_sock_destruct;/* Call inet_sock_destruct when the socket is destroyed */
Sk-> zapped = 0;
Sk-> family = PF_INET;
Sk-> protocol = protocol;
Sk-> prot = prot;
Sk-> backlog_rcv = prot-> backlog_rcv;/* prot-> backlog_rcv () see definitions of various types */
Sk-> protinfo. af_inet.ttl = sysctl_ip_default_ttl;/* set the default ttl */
/* Modify/proc/sys/net/ipv4/ip_default_ttl */
Sk-> protinfo. af_inet.mc_loop = 1;
Sk-> protinfo. af_inet.mc_ttl = 1;
Sk-> protinfo. af_inet.mc_index = 0;
Sk-> protinfo. af_inet.mc_list = NULL;
# Ifdef INET_REFCNT_DEBUG
Atomic_inc (& inet_sock_nr );
# Endif
If (sk-> num ){
/* It assumes that any protocol which allows
* The user to assign a number at socket
* Creation time automatically
* Shares.
*/
Sk-> sport = htons (sk-> num);/* set the local port */
/* Add to protocol hash chains .*/
Sk-> prot-> hash (sk );
}
If (sk-> prot-> init ){
Int err = sk-> prot-> init (sk);/* initialize socket */
If (err! = 0 ){
Inet_sock_release (sk );
Return (err );
}
}
Return (0 );
Free_and_badtype:
Sk_free (sk);/* release memory */
Return-ESOCKTNOSUPPORT;
Free_and_badperm:
Sk_free (sk );
Return-EPERM;
Free_and_noproto:
Sk_free (sk );
Return-EPROTONOSUPPORT;
Do_oom:
Return-ENOBUFS;
}
In net/core/sock. c
Void sock_init_data (struct socket * sock, struct sock * sk)
{
Skb_queue_head_init (& sk-> receive_queue);/* initialize three queues for receiving, sending, and Error */
Skb_queue_head_init (& sk-> write_queue );
Skb_queue_head_init (& sk-> error_queue );
Init_timer (& sk-> timer);/* initialize timer */
　　
Sk-> allocation = GFP_KERNEL;
Sk-> rcvbuf = sysctl_rmem_default;
Sk-> sndbuf = sysctl_wmem_default;
Sk-> state = TCP_CLOSE;
Sk-> zapped = 1;
Sk-> socket = sock;
If (sock)
{
Sk-> type = sock-> type;
Sk-> sleep = & sock-> wait;
Sock-> sk = sk;
} Else
Sk-> sleep = NULL;
Sk-> dst_lock = RW_LOCK_UNLOCKED;
Sk-> callback_lock = RW_LOCK_UNLOCKED;
/* Sock_def_wakeup (), sock_def_readable (),
Sock_def_write_space (), sock_def_error_report (),
Sock_def_destruct () in net/core/sock. c */
Sk-> state_change = sock_def_wakeup;
Sk-> data_ready = sock_def_readable;
Sk-> write_space = sock_def_write_space;
Sk-> error_report = sock_def_error_report;
Sk-> destruct = sock_def_destruct;
Sk-> peercred. pid = 0;
Sk-> peercred. uid =-1;
Sk-> peercred. gid =-1;
Sk-> rcvlowat = 1;
Sk-> rcvtimeo = MAX_SCHEDULE_TIMEOUT;/* set accept and send timeout */
Sk-> sndtimeo = MAX_SCHEDULE_TIMEOUT;
Atomic_set (& sk-> refcnt, 1 );
}
1.1 SOCK_STREAM Initialization
In net/ipv4/tcp_ipv4.c
Static int tcp_v4_init_sock (struct sock * sk)
{
Struct tcp_opt * tp = & (sk-> tp_pinfo.af_tcp );
Skb_queue_head_init (& tp-> out_of_order_queue );
Tcp_init_xmit_timers (sk );
Tcp_prequeue_init (tp );
Tp-> rto = TCP_TIMEOUT_INIT;
Tp-> mdev = TCP_TIMEOUT_INIT;
　　　　　
/* So many TCP implementations out there (incorrectly) count
* Initial SYN frame in their delayed-ACK and congestion control
* Algorithms that we must have the following bandaid to talk
* Efficiently to them.-DaveM
*/
Tp-> snd_cwnd = 2;
/* See draft-stevens-tcpca-spec-01 for discussion of
* Initialization of these values.
*/
Tp-> snd_ssthresh = 0x7fffffff;/* Infinity */
Tp-> snd_cwnd_clamp = ~ 0;
Tp-> mss_cache = 536;
Tp-> reordering = sysctl_tcp_reordering;
Sk-> state = TCP_CLOSE;
Sk-> write_space = tcp_write_space;/* tcp_write_space () in net/ipv4/tcp. c */
Sk-> use_write_queue = 1;
Sk-> tp_pinfo.af_tcp.af_specific = & apos;
/* Ipv4_specific in net/ipv4/tcp_ipv4.c */
Sk-> sndbuf = sysctl_tcp_wmem [1];/* set the buffer size for sending and receiving */
Sk-> rcvbuf = sysctl_tcp_rmem [1];/* sysctl_tcp _ * in net/ipv4/tcp. c */
Atomic_inc (& tcp_sockets_allocated);/* tcp_sockets_allocated indicates the current number of TCP sockets */
Return 0;
}
SOCK_DGRAM is not initialized
1.2 SOCK_RAW Initialization
In net/ipv4/raw. c
Static int raw_init (struct sock * sk)
{
Struct raw_opt * tp = & (sk-> tp_pinfo.tp_raw4 );
If (sk-> num = IPPROTO_ICMP)
Memset (& tp-> filter, 0, sizeof (tp-> filter ));
Return 0;
}
2. Server
2.1 bind
Static int inet_bind (struct socket * sock, struct sockaddr * uaddr, int addr_len)
{
Struct sockaddr_in * addr = (struct sockaddr_in *) uaddr;
Struct sock * sk = sock-> sk;
Unsigned short snum;
Int chk_addr_ret;
Int err;
/* If the socket has its own bind function then use it. (RAW )*/
If (sk-> prot-> bind)
Return sk-> prot-> bind (sk, uaddr, addr_len );
/* Only SOCK_RAW defines its own bind function */
If (addr_len <sizeof (struct sockaddr_in ))
Return-EINVAL;
Chk_addr_ret = inet_addr_type (addr-> sin_addr.s_addr );
/* Inet_addr_type: Type of the returned address */
/* In net/ipv4/fib_frontend.c */
/* Not specified by any standard per-se, however it breaks too
* Applications when removed. It is unfortunate since
* Allowing applications to make a non-local bind solv