zebra的Thread机制

• 目录
• zebra 的Thread机制

zebra 的Thread机制

1．thread的四种创建方法

一个新的thread可以通过如下三种方式被创建，主要是看你需要创建的thread的类型：

1， thread_add_read：添加一个thread到read queue，该thread负责通过socket接受和读取从client端来的数据。

2， thread_add_write：添加一个thread到write queue，该thread负责通过socket向client端填充和写数据。

3， thread_add_timer function calls：添加一个thread到timer queue,该thread负责定时一个event，例如update和redistribute一个route table.

4， thread_add_event：添加一个event thread到event queue。

上面这三个函数的处理过程都差不多：

1， 创建thread。首先在unuse queue查找，如果有unuse thread，就使用它，否则重新分配空间。

2， 根据参数，对thread进行赋值。

3， 将该thread加入到相应的queue中。

2． thread的调用

1，bgp daemon不断地从event queue中取出thread并且执行它。一旦该thread被执行了，将该thread的type设置为unuse。并且将该thread添加到unuse queue中。

2，如果event queue为空时，bgp daemon 通过select函数监控读、写、异常三个描述符集。一旦有某个描述符准备就绪，则将该描述符所对应的thread加入ready queue.

而对于timer queue中的thread，只有当select函数超时后才会进入ready queue.

3．zebrad端的thread

zebrad启动后会，在read queue中会出现两个thread，一个是等待来自local client端bgpd的连接，另一个是等待来自vty client端的连接。

第1个thread

zebra_init ( )-> zebra_serv_un ( )中创建一个thread，加入read queue。该thread的处理函数为zebra_accept，监听内部client的socket。

zebra_client_create (client_sock);创建一个新的zebra client

/ Register myself. /

zebra_event (ZEBRA_SERV, accept_sock, NULL);继续监听server socket

puts("<-zebra_accept");

return 0;

}

vty_accept，加入read queue。作为vty server监听internet的vty socket.。

vty_create(vty_sock, &su);

根据vty_sock和ip地址信息su，创建一个新的vty。

puts("<-vty_accept");

return 0;

}

vty_flush。

vty_event (VTY_READ, vty_sock, vty);

根据new client vty的vty_sock创建新的VTY_READ thread，加入read queue，该thread的处理函数为vty_read。

return vty;

}

sockunion_bind(accept_sock, &su, port, NULL);

将accept_socket文件描述符与一个特定的逻辑网络连系起来。服务器端使用su->sin.sin_addr.s_addr = htonl (INADDR_ANY);表示接受任何一个主机网络接口上的连接请求。

if (ret < 0)

​ {

​ close (accept_sock); / Avoid sd leak. /

​ return;

​ }

/ Listen socket under queue 3. /

ret = listen (accept_sock, 3);

将accept_sock套接口设置成被动监听状态，用于接受连接，只能在服务器端使用。

if (ret < 0)

​ {

​ zlog (NULL, LOG_WARNING, "can't listen socket");

​ close (accept_sock); / Avoid sd leak. /

​ return;

​ }

/ Add vty server event. /

vty_event(VTY_SERV, accept_sock, NULL);

}

sockunion_bind操作

/ Bind socket to specified address. /

int　sockunion_bind (int sock, union sockunion *su, unsigned short port,

​ union sockunion *su_addr)

{

int size = 0;

int ret;

if (su->sa.sa_family == AF_INET)

​ {

​ size = sizeof (struct sockaddr_in);

​ su->sin.sin_port = htons (port);

#ifdef HAVE_SIN_LEN

​ su->sin.sin_len = size;

#endif / HAVE_SIN_LEN /

​ if (su_addr == NULL)

​ su->sin.sin_addr.s_addr = htonl (INADDR_ANY);

服务器一般将sin_addr.s_addr 字段设置为INADDR_ANY表示套接字应接收任何一个主机网络接口上的连接请求。

客户端将sin_addr.s_addr字段设置为服务器主机的IP地址。

​ }

#ifdef HAVE_IPV6

else if (su->sa.sa_family == AF_INET6)

​ {

​ size = sizeof (struct sockaddr_in6);

​ su->sin6.sin6_port = htons (port);

#ifdef SIN6_LEN

​ su->sin6.sin6_len = size;

#endif / SIN6_LEN /

​ if (su_addr == NULL)

​ {

#if defined(LINUX_IPV6) || defined(NRL)

​ bzero (&su->sin6.sin6_addr, sizeof (struct in6_addr));

#else

​ su->sin6.sin6_addr = in6addr_any;

#endif / LINUX_IPV6 /

​ }

​ }

#endif / HAVE_IPV6 /

ret = bind (sock, (struct sockaddr *)su, size);

if (ret < 0)

​ zlog (NULL, LOG_WARNING, "can't bind socket : %s", strerror (errno));

return ret;

}

vty_event操作

/ struct thread_master master; */

static void　vty_event (enum event event, int sock, struct vty *vty)

{

struct thread *vty_serv_thread;

switch (event)

​ {

​ case VTY_SERV:

​ vty_serv_thread = thread_add_read (master, vty_accept, vty, sock);

​ vector_set_index (Vvty_serv_thread, sock, vty_serv_thread);

​ break;

​ case VTY_READ:

​ vty->t_read = thread_add_read (master, vty_read, vty, sock);

​ / Time out treatment. /

​ if (vty->v_timeout)

​ {

​ if (vty->t_timeout)

​ thread_cancel (vty->t_timeout);

​ vty->t_timeout =

​ thread_add_timer (master, vty_timeout, vty, vty->v_timeout);

​ }

​ break;

​ case VTY_WRITE:

​ if (! vty->t_write)

​ vty->t_write = thread_add_write (master, vty_flush, vty, sock);

​ break;

​ case VTY_TIMEOUT_RESET:

​ if (vty->t_timeout)

​ {

​ thread_cancel (vty->t_timeout);

​ vty->t_timeout = NULL;

​ }

​ if (vty->v_timeout)

​ {

​ vty->t_timeout =

​ thread_add_timer (master, vty_timeout, vty, vty->v_timeout);

​ }

​ break;

​ }

}

第2 个thread

bgp_serv_sock( )-> bgp_serv_sock_family( )中创建一个thread，不是通过event的方式添加的。该thread的处理函数为bgp_accept。作为bgp_server，接受来自 internet上的bgp连接。

bgp_serv_sock_family操作

port = 179 family = AF_INET

/ Make bgpd's server socket. /

void　bgp_serv_sock_family (unsigned short port, int family)

{

int ret;

int bgp_sock;

union sockunion su;

bzero (&su, sizeof (union sockunion));

/ Specify address family. /

su.sa.sa_family = family;

bgp_sock = sockunion_stream_socket (&su); 产生一个BGP socket

sockopt_reuseaddr (bgp_sock);

sockopt_reuseport (bgp_sock);

ret = sockunion_bind (bgp_sock, &su, port, NULL);

ret = listen (bgp_sock, 3);

if (ret < 0)

​ {

​ zlog (NULL, LOG_INFO, "Can't listen bgp server socket : %s",

​ strerror (errno));

​ return;

​ }

thread_add_read (master, bgp_accept, NULL, bgp_sock); 添加一个thread 到readlist中。

}

VTY server和BGP server 在使用accept操作的方法如下：

他们均是通过创建一个THREAD_READ 类型的thread，加到Master的readlist的队列后面，thread 的处理函数会执行accept操作。

VTY accept：

vty_serv_thread = thread_add_read (master, vty_accept, vty, sock);

BGP server accept:

thread_add_read (master, bgp_accept, NULL, bgp_sock);

bgpd和zebrad间通信

bgp和zebra是通过zebra message进行通信，格式如下：

报文头3字节 （前两字节length，后1字节为command type）

报文体长度不定。

/ Zebra message types. /

#define ZEBRA_INTERFACE_ADD 1

#define ZEBRA_INTERFACE_DELETE 2

#define ZEBRA_INTERFACE_ADDRESS_ADD 3

#define ZEBRA_INTERFACE_ADDRESS_DELETE 4

#define ZEBRA_INTERFACE_UP 5

#define ZEBRA_INTERFACE_DOWN 6

#define ZEBRA_IPV4_ROUTE_ADD 7

#define ZEBRA_IPV4_ROUTE_DELETE 8

#define ZEBRA_IPV6_ROUTE_ADD 9

#define ZEBRA_IPV6_ROUTE_DELETE 10

#define ZEBRA_REDISTRIBUTE_ADD 11

#define ZEBRA_REDISTRIBUTE_DELETE 12

#define ZEBRA_REDISTRIBUTE_DEFAULT_ADD 13

#define ZEBRA_REDISTRIBUTE_DEFAULT_DELETE 14

#define ZEBRA_IPV4_NEXTHOP_LOOKUP 15

#define ZEBRA_IPV6_NEXTHOP_LOOKUP 16

bgpd和zebrad之间的api接口

bgp端接受到message后，会执行相应的bgp action:

bgp action func message type

int (interface_add) (int, struct zclient , zebra_size_t); ZEBRA_INTERFACE_ADD

int (interface_delete) (int, struct zclient , zebra_size_t); ZEBRA_INTERFACE_DELETE

int (interface_up) (int, struct zclient , zebra_size_t); ZEBRA_INTERFACE_UP

int (interface_down) (int, struct zclient , zebra_size_t); ZEBRA_INTERFACE_DOWN

int (interface_address_add) (int, struct zclient , zebra_size_t); ZEBRA_INTERFACE_ADDRESS_ADD

int (interface_address_delete) (int, struct zclient , zebra_size_t); ZEBRA_INTERFACE_ADDRESS_DELETE

int (ipv4_route_add) (int, struct zclient , zebra_size_t); ZEBRA_IPV4_ROUTE_ADD

int (ipv4_route_delete) (int, struct zclient , zebra_size_t); ZEBRA_IPV4_ROUTE_DELETE

int (ipv6_route_add) (int, struct zclient , zebra_size_t); ZEBRA_IPV6_ROUTE_ADD

int (ipv6_route_delete) (int, struct zclient , zebra_size_t); ZEBRA_IPV6_ROUTE_DELETE

zebra 端接受到message后，会执行相应的zebra action:

zebra action func message type

void zread_interface_add (struct zserv *client, u_short length) ZEBRA_INTERFACE_ADD

void zread_interface_delete (struct zserv *client, u_short length) ZEBRA_INTERFACE_DELETE

void zread_ipv4_add (struct zserv *client, u_short length) ZEBRA_IPV4_ROUTE_ADD

void zread_ipv4_delete (struct zserv *client, u_short length) ZEBRA_IPV4_ROUTE_DELETE

void zread_ipv6_add (struct zserv *client, u_short length) ZEBRA_IPV6_ROUTE_ADD

void zread_ipv6_delete (struct zserv *client, u_short length) ZEBRA_IPV6_ROUTE_DELETE

void zebra_redistribute_add (int command, struct zserv *client, int length) ZEBRA_REDISTRIBUTE_ADD

void zebra_redistribute_delete (int command, struct zserv *client, int length) ZEBRA_REDISTRIBUTE_DELETE

voidzebra_redistribute_default_add (int command,

struct zserv *client, int length) ZEBRA_REDISTRIBUTE_DEFAULT_ADD

void zebra_redistribute_default_delete (int command,

struct zserv *client, int length) ZEBRA_REDISTRIBUTE_DEFAULT_DELETE

void zread_ipv4_nexthop_lookup (struct zserv *client, u_short length) ZEBRA_IPV4_NEXTHOP_LOOKUP

void zread_ipv6_nexthop_lookup (struct zserv *client, u_short length) ZEBRA_IPV6_NEXTHOP_LOOKUP

bgp action：将local_client_socket中数据，写入bgp数据库。

zebra action：将zebra数据库中的信息写入local_server_subsocket，让local client端进行读取。

bgp peer间通信

bgp_accept操作

/ Accept bgp connection. /

int bgp_accept (struct thread *thread)

{

int bgp_sock;

int accept_sock;

union sockunion su;

struct peer *peer;

struct peer *peer1;

char buf[SU_ADDRSTRLEN];

/ Regiser accept thread. /

accept_sock = THREAD_FD (thread);

​ printf("->bgp_accept [%d]\n",accept_sock);

thread_add_read (master, bgp_accept, NULL, accept_sock);

/ Accept client connection. /

bgp_sock = sockunion_accept (accept_sock, &su);

if (bgp_sock < 0)

​ {

​ zlog_err ("[Error] BGP socket accept failed (%s)", strerror (errno));

​ printf("[Error] BGP socket accept failed (%s)", strerror (errno));

​ puts("<-bgp_accept 2");

​ return -1;

​ }

if (BGP_DEBUG (events, EVENTS))

​ zlog_info ("[Event] BGP connection from host %s", inet_sutop (&su, buf));

printf("[Event] BGP connection from host %s", inet_sutop (&su, buf));

/ Check remote IP address /

peer1 = peer_lookup_by_su (&su);

if (! peer1 || peer1->status == Idle)

​ {

​ if (BGP_DEBUG (events, EVENTS))

​ {

​ if (! peer1)

​ zlog_info ("[Event] BGP connection IP address %s is not configured",

​ inet_sutop (&su, buf));

​ else

​ zlog_info ("[Event] BGP connection IP address %s is Idle state",

​ inet_sutop (&su, buf));

​ }

​ close (bgp_sock);

​ puts("<-bgp_accept 2");

​ return -1;

​ }

/ Make dummy peer until read Open packet. /

if (BGP_DEBUG (events, EVENTS))

​ zlog_info ("[Event] Make dummy peer structure until read Open packet");

printf("[Event] Make dummy peer structure until read Open packet\n");

{

​ char buf[SU_ADDRSTRLEN + 1];

​ peer = peer_create_accept ();

​ SET_FLAG (peer->sflags, PEER_STATUS_ACCEPT_PEER);

​ peer->su = su;

​ peer->fd = bgp_sock;

​ peer->status = Active;

​ / Make peer's address string. /

​ sockunion2str (&su, buf, SU_ADDRSTRLEN);

​ peer->host = strdup (buf);

}

BGP_EVENT_ADD (peer, TCP_connection_open); 创建一个event thread执行bgp_event函数

puts("<-bgp_accept 0");

return 0;

}

bgp_event操作

/ Execute event process. /

int bgp_event (struct thread *thread)

{

int ret;

int event;

int next;

struct peer *peer;

peer = THREAD_ARG (thread); // get peer

event = THREAD_VAL (thread); // get FSM event eg.TCP_connection_open

puts("->bgp_event");

/ Logging this event. /

next = FSM [peer->status -1][event - 1].next_state; next为下一个状态

if (BGP_DEBUG (fsm, FSM))

​ plog_info (peer->log, "%s [FSM] %s (%s->%s)", peer->host,

​ bgp_event_str[event],

​ LOOKUP (bgp_status_msg, peer->status),

​ LOOKUP (bgp_status_msg, next));

printf("%s [FSM] %s (%s->%s)", peer->host,

​ bgp_event_str[event],

​ LOOKUP (bgp_status_msg, peer->status),

​ LOOKUP (bgp_status_msg, next));

/ Call function. /

ret = (*(FSM [peer->status - 1][event - 1].func))(peer); 执行本状态处理函数

/ When function do not want proceed next job return -1. /

if (ret < 0)

​ {

​ puts("<-bgp_event 1");

​ return ret;

​ }

/ If status is changed. /

if (next != peer->status) 判断状态是否需要转变

​ fsm_change_status (peer, next);

/ Make sure timer is set. /

bgp_timer_set (peer);

puts("<-bgp_event 0");

return 0;

}