net/ipv4/route.c文件分析_ipv4 route.c-优快云博客

本文主要探讨Linux内核中net/ipv4/route.c文件的内容，包括静态和动态路由表的展示，路由处理流程，以及关键的数据结构和函数详解。介绍了路由初始化、路由流程图、路由表的维护，如rt_hash_table的使用，以及路由缓存的管理等核心概念。

net/ipv4/route.c

author: elvis

目錄介紹 route.c 的背景 routing 流程相關資料結構 route.c 相關函數詳解

介紹 route.c 的背景

linux 在設定 route 有兩個機制,一個是 fib,一個是 dynamic 產生的 routing
fib 是利用 route (man 8 route) 指定來靜態 route table
而 net/ipv4/route.c 所做的則是動能產生 routing hash,以加快 route decision 的速度
範例：

靜態 route table
# route -F
Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface 192.168.1.3 * 255.255.255.255 UH 0 0 0 eth0 140.123.103.156 * 255.255.255.255 UH 0 0 0 eth1 192.168.1.0 * 255.255.255.0 U 0 0 0 eth0 140.123.0.0 * 255.255.0.0 U 0 0 0 eth1 127.0.0.0 * 255.0.0.0 U 0 0 0 lo default CSIE-C5505 0.0.0.0 UG 0 0 0 eth1

動態 route table
# route -C Kernel IP routing cache Source Destination Gateway Flags Metric Ref Use Iface infomedia.cs.cc 140.123.103.255 140.123.103.255 ri 0 6 0 eth1 CSIE-C5505 not-a-legal-add not-a-legal-add ibl 0 796989 0 lo fish.cs.ccu.edu 140.123.103.255 140.123.103.255 ri 0 2 0 eth1 didi.cs.ccu.edu 140.123.103.255 140.123.103.255 ri 0 31 0 eth1 lena.cs.ccu.edu 140.123.103.255 140.123.103.255 ri 0 2438 0 eth1 filter.cs.ccu.e 140.123.103.255 140.123.103.255 ri 0 265 0 eth1 CSIE-C5505 ALL-SYSTEMS.MCA ALL-SYSTEMS.MCA ml 0 7273 0 lo vivi.cs.ccu.edu 140.123.103.255 140.123.103.255 ri 0 7 0 eth1 gigi.cs.ccu.edu 140.123.103.255 140.123.103.255 ri 0 3 0 eth1 scoap.cs.ccu.ed 140.123.103.255 140.123.103.255 ri 0 12 0 eth1 bist.cs.ccu.edu 140.123.103.255 140.123.103.255 ri 0 86 0 eth1 macgyver.cs.ccu 140.123.103.255 140.123.103.255 ri 0 11 0 eth1

參考資料：

RFC 1122 有關 IP 層 routing 的說明這裡

table

routing 流程

	一個 skb 要決定行走的路徑(input) 時
	會呼叫如 ip_route_input(skb,...); 來決定他的 rtable
	流程大概是
		1.ip_route_input 查看是否有存在的 rtable
		2.呼叫 ip_route_input_slow 產生 rtable
		3.ip_route_input_slow 呼叫 fib_lookup
		4.fib_lookup 從 fib_rules 找出 fib_rule
		5.fib_lookup 呼叫 fib_get_table 來取得 fib_table
		6.fib_lookup 使用 fib_table 呼叫 tb->tb_lookup 取得 fib_result
		7.fib_lookup 將 fib_rule 設定給 fib_result
		8.ip_route_input_slow 利用 rt_intern_hash 設定 rtable
		9.返回 ip_route_input 此時 skb 已有設定好的 rtable還有 input/output 方法
		10.返回原呼叫者
		11.原呼叫者呼叫 dst->input() 處理

table

相關資料結構

/* IPv4 routing cache flags */

#define RTCF_DEAD	RTNH_F_DEAD
#define RTCF_ONLINK	RTNH_F_ONLINK
/* Obsolete flag. About to be deleted */
#define RTCF_NOPMTUDISC RTM_F_NOPMTUDISC

#define RTCF_NOTIFY	0x00010000
#define RTCF_DIRECTDST	0x00020000
#define RTCF_REDIRECTED	0x00040000
#define RTCF_TPROXY	0x00080000
#define RTCF_FAST	0x00200000
#define RTCF_MASQ	0x00400000
#define RTCF_SNAT	0x00800000
#define RTCF_DOREDIRECT 0x01000000
#define RTCF_DIRECTSRC	0x04000000
#define RTCF_DNAT	0x08000000
#define RTCF_BROADCAST	0x10000000
#define RTCF_MULTICAST	0x20000000
#define RTCF_REJECT	0x40000000
#define RTCF_LOCAL	0x80000000
#define RTCF_NAT	(RTCF_DNAT|RTCF_SNAT)
routing type
enum
{
	RTN_UNSPEC,
	RTN_UNICAST,		/* Gateway or direct route	*/
	RTN_LOCAL,		/* Accept locally		*/
	RTN_BROADCAST,		/* Accept locally as broadcast,
				   send as broadcast */
	RTN_ANYCAST,		/* Accept locally as broadcast,
				   but send as unicast */
	RTN_MULTICAST,		/* Multicast route		*/
	RTN_BLACKHOLE,		/* Drop				*/
	RTN_UNREACHABLE,		/* Destination is unreachable   */
	RTN_PROHIBIT,		/* Administratively prohibited	*/
	RTN_THROW,		/* Not in this table		*/
	RTN_NAT,			/* Translate this address	*/
	RTN_XRESOLVE,		/* Use external resolver	*/
};
scope 的值
enum rt_scope_t
{
	RT_SCOPE_UNIVERSE=0,
/* User defined values  */
	RT_SCOPE_SITE=200,
	RT_SCOPE_LINK=253,
	RT_SCOPE_HOST=254,
	RT_SCOPE_NOWHERE=255
	};

●用來查詢 fib table 的鍵值
struct rt_key ＠include/net/route.h
{
	__u32	dst; 目的 ip 位址
	__u32	src; 來源 ip 位址
	int	iif; input interface
	int	oif; output interface
	__u8	tos; type of service
	__u8	scope; Unknown
};
●route table 最主要的結構
一些判斷位址的 macro
#define LOOPBACK(x) (((x) & htonl(0xff000000)) == htonl(0x7f000000))
	127.0.0.0/255.0.0.0 loopback
#define MULTICAST(x)    (((x) & htonl(0xf0000000)) == htonl(0xe0000000))
	class D
	1110 開頭
#define BADCLASS(x) (((x) & htonl(0xf0000000)) == htonl(0xf0000000))
	class E 保留的區段，不應該有人用到
	11110 開頭
#define ZERONET(x)  (((x) & htonl(0xff000000)) == htonl(0x00000000))
	class A 且 netid 為 0000000 (七個 0)
#define LOCAL_MCAST(x)  (((x) & htonl(0xFFFFFF00)) == htonl(0xE0000000))
	class D multicast ip
	且 只有最小的 8bit

struct rtable ＠include/net/route.h
{
	union
	{
		struct dst_entry	dst;
		struct rtable	*rt_next;
	} u;
	unsigned	rt_flags;  RTCF_MULTICAST 之類的 flag
	unsigned	rt_type;  RTN_UNICAST 之類的型態,用來表示此 route rule 是哪種型態
	__u32	rt_dst;	/* Path destination	*/
	__u32	rt_src;	/* Path source		*/
	int	rt_iif;  incoming interface
	/* Info on neighbour */
	__u32	rt_gateway;
	/* Cache lookup keys */
	struct rt_key	key;
	/* Miscellaneous cached information */
	__u32	rt_spec_dst; /* RFC1122 specific destination */
	inet_peer 的結構是個 avl tree 用來紀錄各個 peer 的資訊
	詳細 implement 在 inetpeer.h 與 inetpeer.c
	struct inet_peer	*peer; /* long-li