sk_buff 剖析
基于内核版本2.6.37
本文主要剖析:sk_buff结构体、sk_buff操作函数、各协议层对其处理
主要源文件:linux-2.6.37/ include/ linux/ skbuff.h
linux-2.6.37/ include/ linux/ skbuff.c
==================================================================================================
一些相关数据结构
在include/linux/ktime.h中, union ktime { s64 tv64 ; #if BITS_PER_LONG != 64 && !defined(CONFIG_KTIME_SCALAR) struct { # ifdef __BIG_ENDIAN s32 sec , nsec ; #else s32 nsec , sec ; #endif } tv ; #endif } ; typedef union ktime ktime_t ; struct sk_buff_head { /* These two members must be first. */ struct sk_buff *next; struct sk_buff *prev; __u32 qlen; spinlock_t lock; }; /* 关于sk_buff_data_t */ # if BITS_PER_LONG > 32 # define NET_SKBUFF_DATA_USES_OFFSET 1 # endif # ifdef NET_SKBUFF_DATA_USES_OFFSET typedef unsigned int sk_buff_data_t ; # else typedef unsigned char *sk_buff_data_t ; #endif
==================================================================================================
sk_buff结构体
/* struct sk_buff - socket buffer */ struct sk_buff { /* These two members must be first */ struct sk_buff *next ; /* Next buffer in list */ struct sk_buff *prev ; /* Previous buffer in list */ ktime_t tstamp ; /* Time we arrived,记录接收或发送报文的时间戳*/ struct sock *sk ; /* Socket we are owned by */ /* Device we arrived on / are leaving by * 通过该设备接收或发送,记录网络接口的信息和完成操作 */ struct net_device *dev ; /* This is the control buffer. It is free to use for every * layer. Please put your private variables there. */ char cb[48] __aligned (8) ; ... /* data_len为分页数据所包含的全部报文长度 * len为某时刻的报文总长度 * 那么,线性数据的长度为:skb->len - skb->data_len */ unsigned int len , data_len ; /* 保存了下一个协议层的信息,在处理报文时由当前协议层设置 */ __be16 protocol ; ... /* head指向线性数据区的开始 * data指向驻留线性数据区中数据的起始位置 */ unsigned char *head , *data ; ... /* 协议头表示 */ sk_buff_data_t transport_header ; /* 传输层协议头 */ sk_buff_data_t network_header ; /* 网络层协议头 */ sk_buff_data_t mac_header ; /* 链路层协议头 */ sk_buff_data_t tail ; /* 指向驻留在线性数据区的最后一字节数据*/ sk_buff_data_end ; /* 指向线性数据区的结尾,确保不超出可用存储缓冲区 */ atomic_t users ; /* 引用该sk_buff的数量*/ /* 该缓冲区所分配的总内存,包括sk_buff结构大小 + 数据块大小 (应该不包括分页大小?)*/ unsigned int truesize ; } /* This data is invariant across clones and lives at * the end of the header data, ie. at skb->end. */ struct skb_shared_info { /* number of fragments belonged to this sk_buff * 此sk_buff分页段的数目,它表示frags[]数组的元素数量,该数组包含sk_buff的分页数据 */ unsigned short nr_frags; ... /* 指向其分段列表,此sk_buff的总长度为frag_list链表中每个分段长度(skb->len)的和, * 再加上原始的sk_buff的长度 * 通过此域可进行报文分段!! */ struct sk_buff *frag_list ; /* * Warning : all fields before dataref are cleared in __alloc_skb() * 此sk_buff被引用的次数 */ atomic_t dataref ; /* * must be last field * 分段的数组,包含sk_buff的分页数据 */ skb_frag_t frags[MAX_SKB_FRAGS] ; } /* To allow 64K frame to be packed as single skb without frag_list * 允许小于64K的数据不用分段,即不适用frag_list */ #define MAX_SKB_FRAGS (65536 / PAGE_SIZE + 2 ) typedef struct skb_frag_struct skb_frag_t ; struct skb_frag_struct { struct page *page ; /* 该页的虚拟地可用page_address()得到*/ #if (BITS_PER_LONG > 32) || (PAGE_SIZE >= 65536) __u32 page_offset ; __u32 size; #else __u16 page_offset ; __u16 size ; #endif };
注意:分段和分页是两个不同的概念。
分页,即使用非线性数据区,非线性区的含义是包含在sk_buff中的数据长度超过了线性数据区
所能容纳的界限(一般为一页)。包含在非线性数据区中的数据是sk_buff结构中end域所指数据
的连续,全部数据的总长度包含在线性和非线性数据区中。
sk_buff数据的总长度存储在len域,非线性数据的长度存储在sk_buff的data_len域。
分页的实现:
在skb_shared_info中,skb_frag_t frags[MAX_SKB_FRAGS]
通过分页,使得一个sk_buff最多能存:64K的数据(非线性区)+ 一页数据(线性区)。
当DMA支持物理分散页的分散-聚集操作时,才有可能存在分页数据区。如果支持,就为线性数据区
分配一页的数据,其他数据则保存在分页数据区中,随后数据的每个sk_buff分段都会分配一页的数据。
如果不支持,就尝试在线性数据区为整个sk_buff数据分配连续的物理内存。
分段,主要指IP分段的实现。当一个数据报过大时,需要分为多个。即一个sk_buff分为多个
sk_buff,这些sk_buff形成一个链表。
分段的实现:
在skb_shared_info中,struct sk_buff *frag_list
通过frag_list可以遍历分段列表。
======================================================================================================
sk_buff的操作
1. alloc_skb
static inline struct sk_buff *alloc_skb( unsigned int size , gfp_t priority) { return __alloc_skb(size , priority , 0 , NUMA_NO_NONE) ; }
size是数据包的大小。
The returned buffer has no headroom and a tail room of size bytes.
2. skb_reserve
用来为协议头预留空间。拓展head room。
/** * skb_reserve - ajust headroom * @skb : buffer to alter * @len : bytes to move * * Increase the headroom of an empty &sk_buff by reducing the tail * room. This is only allowed for an empty buffer. */ static inline void skb_reserve( struct sk_buff *skb , int len ) { skb->data += len ; skb->tail += len ; }
此时,head room 大小为len,data room 大小0,tail room大小为原长 - len。
当构造一个报文时,要为协议头预留最大可能的空间。
如,MAX_TCP_HEADER = MAX_TCP_HEADER + MAX_IP_HEADER + LL_MAX_HEADER
3. skb_put
用来拓展data room。当要向data room增加数据时,先增加data room的可使用空间。
/** * skb_put - add data to a buffer * @skb : buffer to use * @len : amount of data to add * * This function extends the used data area of the buffer. If this would * exceed the total buffer size the kernel will panic. A pointer to the * first byte of the extra data is returned. */ unsigned char *skb_put( struct sk_buff *skb , unsigned int len ) { unsigned char *tmp = skb_tail_pointer(skb) ; /* 如果存在非线性区,即data_len > 0 ,则报bug */ SKB_LINEAR_ASSERT(skb) ; skb->tail += len ; skb->len += len ; if (unlikely(skb->tail > skb->end )) skb_over_panic(skb , len , __builtin_return_address(0)) ; return tmp ; }
4. skb_push
用来拓展data room。和skb_put不同的是,它不是向tail room扩展,而是向head room扩展。
/** * skb_push - add data to the start of a buffer * @skb : buffer to use * @len : amount of data to add * * This function extends the used data area of the buffer at the buffer * start. If this would exceed the total buffer headroom the kernel will * panic. A pointer to the first byte of the extra data is returned. */ unsigned char *skb_push( struct sk_buff *skb , unsigned int len ) { skb->data -= len ; skb->len += len ; if ( unlikely(skb->data < skb->head ) ) skb_under_panic(skb , len , __builtin_return_address(0)) ; return skb->data ; }
注意:
发送报文一般要调用alloc_skb、skb_reserve、skb_put、skb_push。
发送报文时,在不同协议层处理数据时,该数据要添加相应的协议头。
因此,最高层添加数据和自身的协议头。alloc_skb用来申请一个sk_buff。
skb_reserve用来创建头空间。skb_put用来创建用户数据空间,用户数据复制到sk->data
指向的数据区。接下来是在用户数据的前面加上协议头,使用skb_push。
5. skb_pull
在报文到达时访问协议头,接收报文时调用。使head room向data room扩展。
/** * skb_pull - remove data from the start of a buffer * @skb : buffer to use * @len : amount of data to remove * * This function removes data from the start of a buffer, returning the memory to * the headroom. A pointer to the next data in the buffer is returned. Once the * data has been pulled future pushes will overwrite the old data. */ unsigned char *skb_pull( struct sk_buff *skb , unsigned int len ) { return skb_pull_inline(skb , len ) ; } static inline unsigned char *skb_pull_inline(struct sk_buff *skb , unsigned int len) { return unlikely(len > skb->len ) ? NULL : __skb_pull(skb , len) ; } static inline unsigned char *__skb_pull(struct sk_buff *skb , unsigned int len) { skb->len -= len ; BUG_ON(skb->len < skb->data_len ) ; return skb->data += len ; }
====================================================================================================
# ifdef NET_SKBUFF_DATA_USES_OFFSET static inline unsigned char *skb_transport_header(const struct sk_buff skb) { return skb->head + skb->transport_header ; } static inline void skb_reset_transport_header(struct sk_buff *skb) { skb->transport_header = skb->data - skb->head ; } # else static inline unsigned char *skb_transport_header(const struct sk_buff skb) { return skb->transport_header ; } static inline void skb_reset_transport_header(struct sk_buff *skb) { skb->transport_header = skb->data ; } static inline struct tcphdr *tcp_hdr(const struct sk_buff *skb) { return (struct tcphdr *) skb_transport_header(skb) ; }
sk_buff中tcp协议头的表示:
sk_buff_data_t transport_header ;
用函数tcp_hdr(skb)来获取。
当tcp协议头地址有变化时,用skb_reset_transport_header(skb)来更新transport_header。
===============================================================================================
向下遍历协议层(即发送数据包)时,构建协议头
1. 添加TCP头
TCP调用tcp_transmit_skb()来为TCP数据段构建一个TCP头。
首先计算TCP头的长度,要考虑当前TCP连接所使用的选项。一旦完成该操作,就需要调用
skb_push()来为TCP头分配空间。
/* This routine actually transmit TCP packets queued in by tcp_do_sendmsg(). * This is used by both the initial transmission and possible later retransmissions. * All SKB's seen here are completely headerless. It is our job to build the TCP * header, and pass the packet down to IP so it can do the same plus pass the * packet off to the device. * * We are working here with either a clone of the original SKB, or a fresh unique * copy made by the retransmit engine. */ static int tcp_transmit_skb(struct sock *sk , struct sk_buff *skb , int clone_it , gfp_t gfp_mask) { ... struct inet_sock *inet = inet_sk(sk) ; unsigned tcp_option_size, tcp_header_size ; struct tcphdr *th ; ... tcp_header_size = tcp_option_size + sizeof(struct tcphdr) ; ... skb_push(skb , tcp_header_size) ; skb_reset_transport_header(skb) ; ... /* Build TCP header and checksum it. */ th = tcp_hdr(skb) ; th->source = inet->inet_sport ; th->dest = inet->inet_dport ; ... }
2. 添加IP头
ip_build_and_send_pkt()构造报文的IP头,并发送给链路层。
/* * Add an ip header to a sk_buff and sent it out. */ int ip_build_and_sent_pkt(struct sk_buff *skb , struct sock *sk , __be32 saddr , __be32 daddr , struct ip_options *opt) { struct inet_sock *inet = inet_sk(sk) ; ... struct iphdr *iph ; /* Build the IP header. */ skb_push(skb , sizeof(struct iphdr) + (opt ? opt->optlen : 0) ) ; skb_reset_network_header(skb) ; iph = ip_hdr(skb) ; iph->version = 4 ; iph->ihl = 5 ; iph->tos = inet->tos ; ... }
3. 添加链路层头
eth_header构造以太网帧协议头。
#define ETH_HLEN 14 /** * eth_header - create the Ethernet header * @skb : buffer to alter * @dev : source device * @type : Ethernet type field * @daddr : destination address * @saddr : source address * @len : packet length (<= skb->len) * * Set the protocal type. For a packet of type ETH_P_802_3/2 we put * the length in here instead. */ int eth_header(struct sk_buff *skb , struct net_device *dev , unsigned short type , const void *daddr , const void *saddr, unsigned len) { struct ethhdr *eth = (struct ethhdr *) skb_push(skb , ETH_HLEN) ; ... }
=======================================================================================================
向上遍历协议层(接收数据包)时,解析协议头
1. 解析以太网头
当新报文到达时,要为新报文分配一个新的sk_buff,其大小等于报文的长度。sk_buff
的data域指向报文的起始位置(以太网头)。使用skb_pull来提取不同的协议层头。
该例程在sk_buff到IP backlog队列排队之前完成。
/** * eth_type_trans - determine the packet's protocol ID. * @skb : received socket data * @dev : receiving network device * * The rule here is that we * assume 802.3 if the type field is short enough to be a length. * This is normal practice and works for any 'now in use' protocol. */ __be16 eth_type_trans(struct sk_buff *skb , struct net_device *dev ) { struct ethhdr *eth ; skb->dev = dev ; skb_reset_mac_header(skb) ; /* 更新mac_header */ skb_pull_inline(skb , ETH_HLEN) ; /* 此后data指向IP头 */ eth = eth_hdr(skb) ; ... }
2. 解析IP头
现在sk_buff处于IP backlog队列中,由netif_receive_skb()负责处理,该函数将sk_buff
从backlog队列中取出。
netif_receive_skb() 接收数据包得主要处理函数。
/** * netif_receive_skb - process receive buffer from network * @skb : buffer to process * netif_receive_skb() is the main receive data processing function. * It always succeeds. The buffer may be dropped during processing * for congestion control or by the protocol layers. * * This function may only be called from softirq context and interrupts * should be enabled. * * Return values (usually ignored) : * NET_RX_SUCCESS : no congestion * NET_RX_DROP : packet was dropped */ int netif_receive_skb(struct sk_buff *skb)
3. 解析tcp头
网络层处理完报文,在将data指针指向传输层起始位置,并更新transport_header后,
将报文递给传输层,这些工作有ip_local_deliver_finish()来完成。
static int ip_local_deliver_finish(struct sk_buff *skb) { ... __skb_pull(skb , ip_hdrlen(skb)) ; skb_reset_transport_header(skb) ; ... } static inline unsigned int ip_hdrlen(const struct sk_buff *skb) { return ip_hdr(skb)->ihl * 4 ; }
传输层调用tcp_v4_do_rcv()处理传输层头报文。如果连接已建立,并且TCP报文中有数据,
就调用skb_copy_datagram_iovec()将从skb->data偏移tcp_header_len开始的数据复制给
用户应用程序。如果由于某些原因不能复制数据给用户应用程序,就将sk_buff的data指针
向前移动tcp_header_len,再将其发往套接字的接受队列排队。