Sphinx's Notes

Appearance

Linux Net 子系统深度分析 R1 ​

目录 ​

  1. Socket Layer
  2. sk_buff 数据结构
  3. Netdevice 网络设备
  4. Routing 路由子系统
  5. Netfilter Hooks
  6. TCP/UDP 协议实现
  7. 知识点关联表格

1. Socket Layer ​

1.1 struct socket 数据结构 ​

源码位置: /Users/sphinx/github/linux/include/linux/net.h:116-128

c
struct socket {
    socket_state        state;          // SS_FREE, SS_UNCONNECTED, SS_CONNECTING, SS_CONNECTED
    short               type;           // SOCK_STREAM, SOCK_DGRAM, SOCK_RAW 等
    unsigned long       flags;
    struct file         *file;          // 关联的文件结构
    struct sock         *sk;            // 关联的 inet sock
    const struct proto_ops *ops;        // 协议操作集
    struct socket_wq    *wq;            // 等待队列
};

1.2 struct sock 数据结构 ​

源码位置: /Users/sphinx/github/linux/include/net/sock.h:360-450

c
struct sock {
    struct sock_common  __sk_common;     // 公共成员
    
    /* 接收/发送队列 */
    struct sk_buff_head sk_receive_queue; // 接收数据队列
    struct sk_buff_head sk_write_queue;   // 发送数据队列
    
    /* backlog 队列 */
    struct {
        atomic_t rmem_alloc;
        int      len;
        struct sk_buff *head;
        struct sk_buff *tail;
    } sk_backlog;
    
    /* 协议特定操作 */
    const struct proto *sk_prot;        // 协议操作集
    const struct proto_ops *sk_ops;     // Socket 操作集
    
    /* 状态和选项 */
    __u8           sk_state;            // TCP: ESTABLISHED, LISTEN, CLOSE_WAIT 等
    unsigned short sk_type;            // SOCK_STREAM, SOCK_DGRAM
    int            sk_protocol;        // IPPROTO_TCP, IPPROTO_UDP
    
    /* 内存管理 */
    atomic_t       sk_drops;           // 丢包计数
    int            sk_rcvbuf;          // 接收缓冲区大小
    
    /* 定时器 */
    struct timer_list sk_timer;        // 重传定时器
};

1.3 sock_alloc() 函数 ​

源码位置: /Users/sphinx/github/linux/net/socket.c:632-650

c
struct socket *sock_alloc(void)
{
    struct inode *inode;
    struct socket *sock;

    inode = new_inode_pseudo(sock_mnt->mnt_sb);
    if (!inode)
        return NULL;

    sock = SOCKET_I(inode);
    inode->i_ino = get_next_ino();
    inode->i_mode = S_IFSOCK | S_IRWXUGO;
    inode->i_uid = current_fsuid();
    inode->i_gid = current_fsgid();
    inode->i_op = &sockfs_inode_ops;

    return sock;
}

1.4 inet_bind() 函数 ​

源码位置: /Users/sphinx/github/linux/net/ipv4/af_inet.c:473-530

c
int __inet_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len, u32 flags)
{
    struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
    struct inet_sock *inet = inet_sk(sk);
    struct net *net = sock_net(sk);
    unsigned short snum;
    
    if (addr->sin_family != AF_INET)
        return -EAFNOSUPPORT;
    
    snum = ntohs(addr->sin_port);
    
    err = inet_bind_hash(sk, inet, snum, tb_id);
    
    return err;
}

2. sk_buff 数据结构 ​

2.1 struct sk_buff 核心结构 ​

源码位置: /Users/sphinx/github/linux/include/linux/skbuff.h:885-980

c
struct sk_buff {
    /* 双向链表指针 */
    union {
        struct {
            struct sk_buff *next;      // 下一个 skb
            struct sk_buff *prev;      // 上一个 skb
            union {
                struct net_device *dev;
                unsigned long dev_scratch;
            };
        };
        struct rb_node rbnode;
        struct list_head list;
    };

    struct sock *sk;
    union {
        ktime_t tstamp;
        u64 skb_mstamp_ns;
    };

    char cb[48] __aligned(8);         // 控制缓冲区

    unsigned int len;                  // 数据总长度
    unsigned int data_len;            // 数据长度(非线性部分)
    __u16 mac_len;                  // MAC 头长度
    __u16 hdr_len;                  // 可写头长度

    /* 克隆状态 */
    __u8 cloned:1;
    __u8 nohdr:1;
    __u8 fclone:2;
};

2.2 skb_put / skb_push 函数 ​

c
// skb_put - 在尾部添加数据
static inline void *__skb_put(struct sk_buff *skb, unsigned int len)
{
    void *tmp = skb_tail_pointer(skb);
    skb->tail += len;
    skb->len  += len;
    return tmp;
}

// skb_push - 在头部添加数据
static inline void *__skb_push(struct sk_buff *skb, unsigned int len)
{
    skb->data -= len;
    skb->len  += len;
    return skb->data;
}

2.3 skb_shared_info 结构 ​

源码位置: /Users/sphinx/github/linux/include/linux/skbuff.h:593-630

c
struct skb_shared_info {
    __u8        flags;
    __u8        nr_frags;           // 碎片数量
    unsigned short gso_size;        // GSO 大小
    unsigned short gso_segs;        // GSO 段数
    struct sk_buff *frag_list;      // 碎片列表
    
    atomic_t dataref;               // 数据引用计数
};

3. Netdevice 网络设备 ​

3.1 struct net_device 数据结构 ​

源码位置: /Users/sphinx/github/linux/include/linux/netdevice.h:2109-2200

c
struct net_device {
    /* TX 热路径 */
    unsigned long priv_flags:32;
    const struct net_device_ops *netdev_ops;
    struct netdev_queue *_tx;
    unsigned int real_num_tx_queues;
    unsigned int mtu;
    unsigned short hard_header_len;
    
    /* 统计信息 */
    union {
        struct pcpu_lstats __percpu *lstats;
        struct pcpu_sw_netstats __percpu *tstats;
    };
    
    unsigned int flags;
    struct netdev_rx_queue *_rx;
    rx_handler_func_t __rcu *rx_handler;
    
    char name[IFNAMSIZ];
    int ifindex;
};

3.2 dev_queue_xmit() 发送流程 ​

源码位置: /Users/sphinx/github/linux/net/core/dev.c:4760-4893

c
int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev)
{
    struct net_device *dev = skb->dev;
    struct netdev_queue *txq;
    struct Qdisc *q;
    
    skb_reset_mac_header(skb);
    txq = skb_get_tx_queue(dev, skb);
    
    rcu_read_lock_bh();
    
    q = rcu_dereference(txq->qdisc);
    
    if (q->enqueue == NULL) {
        rc = dev_hard_start_xmit(skb, dev, txq);
    } else {
        rc = q->enqueue(skb, q, &to_free) & &rc;
        if (to_free)
            kfree_skb_list(to_free);
        qdisc_run(txq);
    }
    
    rcu_read_unlock_bh();
    return rc;
}

4. Routing 路由子系统 ​

4.1 struct rtable 数据结构 ​

源码位置: /Users/sphinx/github/linux/include/net/route.h:57-78

c
struct rtable {
    struct dst_entry dst;
    
    int rt_genid;
    unsigned int rt_flags;
    __u16 rt_type;
    __u8 rt_is_input:1;
    __u8 rt_uses_gateway:1;
    
    int rt_iif;
    
    union {
        __be32 rt_gw4;
        struct in6_addr rt_gw6;
    };
};

4.2 fib_lookup() 路由查找 ​

源码位置: /Users/sphinx/github/linux/net/ipv4/fib_frontend.c:280-308

c
int fib_lookup(struct net *net, const struct flowi4 *flp,
               struct fib_result *res, unsigned int flags)
{
    struct fib_table *tb;
    
    tb = fib_get_table(net, RT_TABLE_MAIN);
    if (!tb)
        return -ENOENT;
    
    return fib_table_lookup(tb, flp, res, flags);
}

5. Netfilter Hooks ​

5.1 struct nf_hook_ops 数据结构 ​

源码位置: /Users/sphinx/github/linux/include/linux/netfilter.h:98-111

c
struct nf_hook_ops {
    struct list_head list;
    struct rcu_head rcu;
    
    nf_hookfn *hook;
    struct net_device *dev;
    void *priv;
    u8 pf;
    unsigned int hooknum;
    int priority;
};

5.2 Hook 点定义 ​

c
enum nf_INET_hooks {
    NF_INET_PRE_ROUTING,    // 0 - 接收后,路由查找前
    NF_INET_LOCAL_IN,       // 1 - 目的地是本机
    NF_INET_FORWARD,        // 2 - 转发
    NF_INET_LOCAL_OUT,     // 3 - 本机发送
    NF_INET_POST_ROUTING,   // 4 - 发送前
    NF_INET_NUMHOOKS
};

6. TCP/UDP 协议实现 ​

6.1 TCP 状态机 ​

c
enum {
    TCP_ESTABLISHED = 1,
    TCP_SYN_SENT,
    TCP_SYN_RECV,
    TCP_FIN_WAIT1,
    TCP_FIN_WAIT2,
    TCP_TIME_WAIT,
    TCP_CLOSE,
    TCP_CLOSE_WAIT,
    TCP_LAST_ACK,
    TCP_LISTEN,
    TCP_CLOSING,
    TCP_NEW_SYN_RECV,
};

6.2 struct udp_sock ​

源码位置: /Users/sphinx/github/linux/include/net/udp.h:40-50

c
struct udp_sock {
    struct sock sk;
    struct udp_hslot *udp_table;
    int bind_portaddr_hash;
    bool no_check6_tx;
    bool no_check6_rx;
    atomic_t hchgs;
};

6.3 UDP 校验和计算 ​

c
static inline __wsum udp_csum_outgoing(struct sock *sk, struct sk_buff *skb)
{
    __wsum csum = csum_partial(skb_transport_header(skb),
                               sizeof(struct udphdr), 0);
    return csum;
}

static inline __sum16 udp_v4_check(int len, __be32 saddr,
                                   __be32 daddr, __wsum base)
{
    return csum_tcpudp_magic(saddr, daddr, len, IPPROTO_UDP, base);
}

7. 知识点关联表格 ​

7.1 数据结构关联表 ​

结构体文件位置关联结构
struct socketinclude/linux/net.h:116struct sock
struct sockinclude/net/sock.h:360struct sk_buff, struct inet_sock
struct sk_buffinclude/linux/skbuff.h:885struct net_device, struct sock
struct net_deviceinclude/linux/netdevice.h:2109struct netdev_queue, struct Qdisc
struct rtableinclude/net/route.h:57struct dst_entry

7.2 核心函数调用链 ​

功能路径函数调用链
Socket 创建socket() → sock_alloc() → inet_create()
数据发送sendmsg() → inet_sendmsg() → tcp_sendmsg()/udp_sendmsg()
包发送设备dev_queue_xmit() → __dev_queue_xmit() → sch_handle_egress()
包接收设备netif_receive_skb() → __netif_receive_skb() → rx_handler()
路由查找fib_lookup() → fib_table_lookup()

7.3 TCP/UDP 核心函数对比 ​

功能TCPUDP
发送函数tcp_sendmsg()udp_sendmsg()
接收函数tcp_recvmsg()udp_recvmsg()
连接建立tcp_v4_connect()不需要连接
监听inet_listen()不支持
校验和tcp_v4_check()udp_v4_check()

总结 ​

Linux Net 子系统是一个复杂的分层架构:

  1. Socket Layer 提供统一的 BSD Socket 接口,支持多种协议
  2. sk_buff 实现高效的包缓冲管理,支持克隆和复制
  3. Netdevice 提供网络设备的抽象,支持发送和接收队列
  4. Routing 实现灵活的路由查找和输出路径
  5. Netfilter 提供强大的包过滤框架
  6. TCP/UDP 实现可靠的传输协议

文档版本: R1
分析源码版本: Linux Kernel (latest)
生成时间: 2026-04-27

基于 VitePress 构建

Copyright © 2024-present Sphinx