Linux网络协议栈(一)——Socket入门(2)

ゞ浴缸里的玫瑰 2022-02-13 11:51 449阅读 0赞

3、套接字的实现  
套接字最先是在UNIX的BSD版本实现的，所以也叫做BSD套接字，它隐藏了各个协议之间的差异，并向上提供统一的接口。Linux中实现套接字的基本结构：  
![img_64b0d5a607ec5adfc5104582ef30dd50.jpeg][]  
3.1、BSD套接字  
3.1.1、核心数据结构  
为了实现BSD套接字，内核提供一个重要的数据结构struct socket，它的定义如下：

![复制代码][img_51e409b11aa51c150090697429a953ed.gif]

//BSD套接字(include/linux/net.h)  
struct socket \{  
socket\_state state; //套接字状态  
unsigned long flags;  
struct proto\_ops \*ops; //操作函数集  
struct fasync\_struct \*fasync\_list;  
struct file \*file;//每个BSD套接字都有一个inode结点,通过文件对象与其关联起来  
struct sock \*sk; //socket内部结构,与具体的协议簇(比如PF\_INET)相关  
wait\_queue\_head\_t wait;  
short type; //套接字类型：如SOCK\_STREAM, SOCK\_DGRAM, SOCK\_RAW, SOCK\_RDM, SOCK\_SEQPACKET, and SOCK\_PACKET  
unsigned char passcred;  
\};  
  
//BSD套接字操作函数集  
struct proto\_ops \{  
int family;  
struct module \*owner;  
int (\*release) (struct socket \*sock);  
int (\*bind) (struct socket \*sock,  
struct sockaddr \*myaddr,  
int sockaddr\_len);  
int (\*connect) (struct socket \*sock,  
struct sockaddr \*vaddr,  
int sockaddr\_len, int flags);  
int (\*socketpair)(struct socket \*sock1,  
struct socket \*sock2);  
int (\*accept) (struct socket \*sock,  
struct socket \*newsock, int flags);  
int (\*getname) (struct socket \*sock,  
struct sockaddr \*addr,  
int \*sockaddr\_len, int peer);  
unsigned int (\*poll) (struct file \*file, struct socket \*sock,  
struct poll\_table\_struct \*wait);  
int (\*ioctl) (struct socket \*sock, unsigned int cmd,  
unsigned long arg);  
int (\*listen) (struct socket \*sock, int len);  
int (\*shutdown) (struct socket \*sock, int flags);  
int (\*setsockopt)(struct socket \*sock, int level,  
int optname, char \_\_user \*optval, int optlen);  
int (\*getsockopt)(struct socket \*sock, int level,  
int optname, char \_\_user \*optval, int \_\_user \*optlen);  
int (\*sendmsg) (struct kiocb \*iocb, struct socket \*sock,  
struct msghdr \*m, size\_t total\_len);  
int (\*recvmsg) (struct kiocb \*iocb, struct socket \*sock,  
struct msghdr \*m, size\_t total\_len,  
int flags);  
int (\*mmap) (struct file \*file, struct socket \*sock,  
struct vm\_area\_struct \* vma);  
ssize\_t (\*sendpage) (struct socket \*sock, struct page \*page,  
int offset, size\_t size, int flags);  
\};  
//BSD套接字状态  
typedef enum \{  
SS\_FREE = 0, /\* not allocated \*/  
SS\_UNCONNECTED, /\* unconnected to any socket \*/  
SS\_CONNECTING, /\* in process of connecting \*/  
SS\_CONNECTED, /\* connected to socket \*/  
SS\_DISCONNECTING /\* in process of disconnecting \*/  
\} socket\_state;

![复制代码][img_51e409b11aa51c150090697429a953ed.gif]

3.1.2、BSD套接字初始化

![复制代码][img_51e409b11aa51c150090697429a953ed.gif]

//net/socket.c  
//BSD套接字的初始化  
void \_\_init sock\_init(void)  
\{  
int i;  
  
/\*  
\* Initialize all address (protocol) families.  
\*/  
  
for (i = 0; i < NPROTO; i++)  
net\_families\[i\] = NULL; //协议簇数组初始化  
  
/\*  
\* Initialize sock SLAB cache.  
\*/  
//分配sock缓存  
sk\_init();  
  
\#ifdef SLAB\_SKB  
/\*  
\* Initialize skbuff SLAB cache  
\*/  
skb\_init();  
\#endif  
  
/\*  
\* Initialize the protocols module.  
\*/  
  
init\_inodecache();  
  
//注册sockfs文件系统  
register\_filesystem(&sock\_fs\_type);  
//安装sockfs  
sock\_mnt = kern\_mount(&sock\_fs\_type);  
/\* The real protocol initialization is performed when  
\* do\_initcalls is run.  
\*/  
  
\#ifdef CONFIG\_NETFILTER  
netfilter\_init();  
\#endif  
\}  
  
  
//net/socket.c  
//sockfs文件系统的安装点  
static struct vfsmount \*sock\_mnt;  
//sockfs文件系统类型  
static struct file\_system\_type sock\_fs\_type = \{  
.name = "sockfs",  
.get\_sb = sockfs\_get\_sb,  
.kill\_sb = kill\_anon\_super,  
\};  
//地址簇及协议信息  
static struct net\_proto\_family \*net\_families\[NPROTO\];

![复制代码][img_51e409b11aa51c150090697429a953ed.gif]

sock\_init在系统初始化的被调用：  
![img_cfcd300a4366397c74ff6a4644afc64d.jpeg][]  
3.1.3、BSD套接字的系统调用  
实际上，Linux内核只提供了一个与套接字相关的系统调用，即sys\_socketcall，应用程序的所有套接字调用都会映射到这个系统调用上。

![复制代码][img_51e409b11aa51c150090697429a953ed.gif]

//BSD套接字调用入口(net/socket.c)  
asmlinkage long sys\_socketcall(int call, unsigned long \_\_user \*args)  
\{  
unsigned long a\[6\];  
unsigned long a0,a1;  
int err;  
  
if(call<1||call>SYS\_RECVMSG)  
return -EINVAL;  
  
/\* copy\_from\_user should be SMP safe. \*/  
if (copy\_from\_user(a, args, nargs\[call\]))//从用户区拷贝参数  
return -EFAULT;  
  
a0=a\[0\];  
a1=a\[1\];  
  
switch(call) //调用相应的函数  
\{  
case SYS\_SOCKET:  
err = sys\_socket(a0,a1,a\[2\]);  
break;  
case SYS\_BIND:  
err = sys\_bind(a0,(struct sockaddr \_\_user \*)a1, a\[2\]);  
break;  
case SYS\_CONNECT:  
err = sys\_connect(a0, (struct sockaddr \_\_user \*)a1, a\[2\]);  
break;  
case SYS\_LISTEN:  
err = sys\_listen(a0,a1);  
break;  
case SYS\_ACCEPT:  
err = sys\_accept(a0,(struct sockaddr \_\_user \*)a1, (int \_\_user \*)a\[2\]);  
break;  
case SYS\_GETSOCKNAME:  
err = sys\_getsockname(a0,(struct sockaddr \_\_user \*)a1, (int \_\_user \*)a\[2\]);  
break;  
case SYS\_GETPEERNAME:  
err = sys\_getpeername(a0, (struct sockaddr \_\_user \*)a1, (int \_\_user \*)a\[2\]);  
break;  
case SYS\_SOCKETPAIR:  
err = sys\_socketpair(a0,a1, a\[2\], (int \_\_user \*)a\[3\]);  
break;  
case SYS\_SEND:  
err = sys\_send(a0, (void \_\_user \*)a1, a\[2\], a\[3\]);  
break;  
case SYS\_SENDTO:  
err = sys\_sendto(a0,(void \_\_user \*)a1, a\[2\], a\[3\],  
(struct sockaddr \_\_user \*)a\[4\], a\[5\]);  
break;  
case SYS\_RECV:  
err = sys\_recv(a0, (void \_\_user \*)a1, a\[2\], a\[3\]);  
break;  
case SYS\_RECVFROM:  
err = sys\_recvfrom(a0, (void \_\_user \*)a1, a\[2\], a\[3\],  
(struct sockaddr \_\_user \*)a\[4\], (int \_\_user \*)a\[5\]);  
break;  
case SYS\_SHUTDOWN:  
err = sys\_shutdown(a0,a1);  
break;  
case SYS\_SETSOCKOPT:  
err = sys\_setsockopt(a0, a1, a\[2\], (char \_\_user \*)a\[3\], a\[4\]);  
break;  
case SYS\_GETSOCKOPT:  
err = sys\_getsockopt(a0, a1, a\[2\], (char \_\_user \*)a\[3\], (int \_\_user \*)a\[4\]);  
break;  
case SYS\_SENDMSG:  
err = sys\_sendmsg(a0, (struct msghdr \_\_user \*) a1, a\[2\]);  
break;  
case SYS\_RECVMSG:  
err = sys\_recvmsg(a0, (struct msghdr \_\_user \*) a1, a\[2\]);  
break;  
default:  
err = -EINVAL;  
break;  
\}  
return err;  
\}  
  
//include/asm/unistd.h  
\#define \_\_NR\_socketcall 102 //系统调用号

![复制代码][img_51e409b11aa51c150090697429a953ed.gif]

下面来看一下sys\_socket的实现：

![img_1c53668bcee393edac0d7b3b3daff1ae.gif][] Code

3.2、INET套接字  
INET套接字就是支持 Internet 地址族的套接字，它位于TCP协议之上， BSD套接字之下，如下：  
![img_8f794f414a1a29c80671a123db80c3bf.jpeg][]  
3.2.1、数据结构

![img_1c53668bcee393edac0d7b3b3daff1ae.gif][] Code

inet\_init()函数：

![img_1c53668bcee393edac0d7b3b3daff1ae.gif][] Code

sock\_register()函数：

![img_1c53668bcee393edac0d7b3b3daff1ae.gif][] Code

inet\_create()函数

![img_1c53668bcee393edac0d7b3b3daff1ae.gif][] Code

[img_64b0d5a607ec5adfc5104582ef30dd50.jpeg]: /images/20220213/db026edcbae24743b2dd456e335d0b67.png
[img_51e409b11aa51c150090697429a953ed.gif]: /images/20220213/edcc22bb183a4fb193f6ad7f7ba86c56.png
[img_cfcd300a4366397c74ff6a4644afc64d.jpeg]: /images/20220213/14573f7bb130432c96c451aa3f4776a4.png
[img_1c53668bcee393edac0d7b3b3daff1ae.gif]: /images/20220213/0fa94e06477448f1a50f8d6939ecd309.png
[img_8f794f414a1a29c80671a123db80c3bf.jpeg]: /images/20220213/ee55ab7409844f1d9828e8cfe54a45a4.png