网络校验和计算

1. 前言
校验和计算是NAT功能和内容修改功能的基本功,这些操作进行后都需要修改数据头中的校验和。
2. 16位校验和计算
2.1 基本原理 
IP/ICMP/IGMP/TCP/UDP等协议的校验和算法都是相同的,采用的都是将数据流视为16位整数流进行重复叠加计算。为了计算检验和,首先把检验和字段置为0。然后,对有效数据范围内中每个16位进行二进制反码求和,结果存在检验和字段中,如果数据长度为奇数则补一字节0。当收到数据后,同样对有效数据范围中每个16位数进行二进制反码的求和。由于接收方在计算过程中包含了发送方存在首部中的检验和,因此,如果首部在传输过程中没有发生任何差错,那么接收方计算的结果应该为全0或全1(具体看实现了,本质一样) 。如果结果不是全0或全1,那么表示数据错误。
2.2 程序算法
2.2.1 C实现 
这是RFC1071中提供的C语言程序(源码编码不过去,做了一点修改):
unsigned short csum(unsigned char *buffer, int count)
{
	register long sum = 0;
	while( count > 1 ) {
		// This is the inner loop 
		sum +=  *(unsigned short *) buffer++;
		count -= 2;
	}		
	//Add left-over byte, if any 
	if( count > 0 )
		sum += * (unsigned char *) buffer;
	//Fold 32-bit sum to 16 bits 
 	while (sum>>16)
	sum = (sum & 0xffff) + (sum >> 16);
	return ~sum;
}
当然,如果用汇编语言实现计算速度会快得多,对于不同的CPU体系,需要各自对应编写不同的汇编,在Linux内核源代码中有各种CPU体系的IP校验和计算源代码。
2.2.2 增量式修改
如果只修改了一个字节,如只修改IP头中的TTL,重新计算校验和时是没必要将所有数据范围内数据重新计算一遍的,RFC1141中提出了增量式算法:
         ~C' = ~(C + (-m) + m') = ~C + (m - m') = ~C + m + ~m'
C'为修改后的校验和,C为修改前的校验和,m为修改前的数值,m'为修改后的数值,~为补码值。
C代码实现为:
UpdateTTL(iph,n)
      struct ip_hdr *ipptr;
      unsigned char n;
      {
          unsigned long sum;
          unsigned short old;
          old = ntohs(*(unsigned short *)&ipptr->ttl);
          ipptr->ttl -= n;
          sum = old + (~ntohs(*(unsigned short *)&ipptr->ttl) & 0xffff);
          sum += ntohs(ipptr->Checksum);
          sum = (sum & 0xffff) + (sum>>16);
          ipptr->Checksum = htons(sum + (sum>>16));
      }
 
2.3 网络应用
2.3.1 IPv4 
IPv4层中的校验和只包括IPv4头部分,不包括上层协议头和应用层数据,校验和是必须计算的。
2.3.2 IPv6 
IPv6头本身已经不包括校验和字段,只靠上层协议的校验和。
2.3.3 ICMP/IGMP
ICMP/IGMP校验和计算范围为从ICMP/IGMP开始到数据结束,不包括IP头部分,校验和是必须计算的。
2.3.4 TCP/UDP
TCP/UDP的校验和计算有点特殊,所计算的数据范围除了包括TCP/UDP头开始到数据结束外,还要包括一个IP伪头部分,所谓伪头,只有12字节数据,包括源地址(4字节)、目的地址(4字节)、协议(2字节,第一字节补0)和TCP/UDP包长(2字节)。TCP的校验和是必须的,而UDP的校验和是可选的,如果UDP中校验和字段为0,表示不进行校验和计算,因此对于UDP协议数据的修改后想偷懒的话直接将校验和设为0 就可以了。
2.3.4.1.IP和TCP校验和的计算:
in_cksum(unsigned short *addr, int len)
{
    register int nleft = len;
    register u_short *w = addr; //u_short 16bit 
    register int sum = 0;
    u_short answer =0;
    while (nleft >1)
    {
        sum += *w++;
        nleft -= 2;  //16bit=2 bytes
    }
    if (nleft == 1)   //odd时,pad zero
    {
       *(u_char *)(&answer) = *(u_char *)w;
       sum += answer;
    }
    sum = (sum >>16) + (sum & 0xffff);
    sum += (sum >> 16);
    answer = ~sum;   //求反
    return(answer);
}
 
/* calculate the ip checksum */
send_tcp.ip.check = 0;   //先检验和置0
send_tcp.ip.check=in_cksum((unsigned short *)&send_tcp.ip, 20);
 
/* calculate the tcp checksum */
struct pseudo_header //tcp伪头部
{
    unsigned int source_address;
    unsigned int dest_address;
    unsigned char placeholder;
    unsigned char protocol;
    unsigned short tcp_length; //以上3*32字节为伪造的tcp首部
    struct tcphdr tcp;  //TCP头
    u_char html[bufsize];  //html数据,包括http头和数据
} pseudo_header;
/* set the pseudo header fields */
pseudo_header.source_address = send_tcp.ip.saddr;
pseudo_header.dest_address = send_tcp.ip.daddr;
pseudo_header.placeholder = 0;
pseudo_header.protocol = IPPROTO_TCP;
pseudo_header.tcp_length = htons(20);
bcopy((char *)&send_tcp.tcp, (char *)&pseudo_header.tcp, 20);
//将send_tcp的数据拷如pseudo_header中的tcp
bcopy(htmlbuf, (char *)&pseudo_header.html, sizeof(htmlbuf));
send_tcp.tcp.check = in_cksum((unsigned short *)&pseudo_header, sizeof
(pseudo_header));
 
2.3.4.2.UDP校验和计算:
UDP的CHECKSUM算法与IP包的HEADER CHECKSUM的计算方法基本一样,只是取样数据不同。UDP中,参与计算CHEKCSUM的数据包括三部分: 亚头部+UDP头部+数据部分亚头部包括:2byte源IP地址+2byte目的IP地址+0x00+1byte协议+2byte的UDP长度UDP包头:2byte源端口+2byte目的端口+2byteUDP包长+0x0000(checksum)数据部分
计算方法,以2字节为一个单位,将其顺序相加,就会产生2个字节的SUM,如果超过2字节,则将高位的值再加回低位,然后取补,得到的就是UDP的checksum.
fixCheckSumUDP(struct udphdr * const        udp,
                           struct iphdr const * const        ip,
                           void const * const                       data)
{
        uint32_t                sum;
        struct 
        {
                struct in_addr src;
                      struct in_addr dst;
                uint8_t                mbz;
                uint8_t                proto;
                uint16_t                len;
        } __attribute__((__packed__))        pseudo_hdr;
        pseudo_hdr.src.s_addr   = (ip->saddr).s_addr;
        pseudo_hdr.dst.s_addr   = (ip->daddr).s_addr;
        pseudo_hdr.mbz   = 0;
        pseudo_hdr.proto = ip->protocol;
        pseudo_hdr.len   = udp->len;
        udp->check = 0;
        sum = calculateCheckSum(&pseudo_hdr, sizeof(pseudo_hdr), 0);
        sum = calculateCheckSum(udp, sizeof(*udp), sum);
        sum = calculateCheckSum(data, ntohs(udp->len)-sizeof(*udp), sum);
        
        sum = ~ntohs(sum);
        if (sum==0) 
                sum=~sum;
        
        udp->check = sum;
}

3. 32位校验和
3.1 以太帧
以太帧校验和使用的是CRC校验,校验和为4字节32位,算法比较适合硬件实现,其计算和校验都是底层完成的,在IP栈以上时就不用考虑,即使上层直接是构造以太帧发送,也只需构造以太头部即可,发送时由底层自动添加后面的校验和。
3.2 SCTP 
在SCTP(协议号:132)协议中,校验和计算比较特殊,采用了和以太包校验和相似的CRC32算法(RFC3309),计算结果是32位而不再是16位。计算范围为从SCTP头到数据结束,不包括IP伪头。
以下是从Linux内核源码SCTP实现中摘取的CRC32算法源码:
/* Example of the crc table file */
#ifndef __crc32cr_table_h__
#define __crc32cr_table_h__ 
#define CRC32C_POLY 0x1EDC6F41
#define CRC32C(c,d) (c=(c>>8)^crc_c[(c^(d))&0xFF]) 
static unsigned long crc_c[256] =
{
0x00000000L, 0xF26B8303L, 0xE13B70F7L, 0x1350F3F4L,
0xC79A971FL, 0x35F1141CL, 0x26A1E7E8L, 0xD4CA64EBL,
0x8AD958CFL, 0x78B2DBCCL, 0x6BE22838L, 0x9989AB3BL,
0x4D43CFD0L, 0xBF284CD3L, 0xAC78BF27L, 0x5E133C24L,
0x105EC76FL, 0xE235446CL, 0xF165B798L, 0x030E349BL,
0xD7C45070L, 0x25AFD373L, 0x36FF2087L, 0xC494A384L,
0x9A879FA0L, 0x68EC1CA3L, 0x7BBCEF57L, 0x89D76C54L,
0x5D1D08BFL, 0xAF768BBCL, 0xBC267848L, 0x4E4DFB4BL,
0x20BD8EDEL, 0xD2D60DDDL, 0xC186FE29L, 0x33ED7D2AL,
0xE72719C1L, 0x154C9AC2L, 0x061C6936L, 0xF477EA35L,
0xAA64D611L, 0x580F5512L, 0x4B5FA6E6L, 0xB93425E5L,
0x6DFE410EL, 0x9F95C20DL, 0x8CC531F9L, 0x7EAEB2FAL,
0x30E349B1L, 0xC288CAB2L, 0xD1D83946L, 0x23B3BA45L,
0xF779DEAEL, 0x05125DADL, 0x1642AE59L, 0xE4292D5AL,
0xBA3A117EL, 0x4851927DL, 0x5B016189L, 0xA96AE28AL,
0x7DA08661L, 0x8FCB0562L, 0x9C9BF696L, 0x6EF07595L,
0x417B1DBCL, 0xB3109EBFL, 0xA0406D4BL, 0x522BEE48L,
0x86E18AA3L, 0x748A09A0L, 0x67DAFA54L, 0x95B17957L,
0xCBA24573L, 0x39C9C670L, 0x2A993584L, 0xD8F2B687L,
0x0C38D26CL, 0xFE53516FL, 0xED03A29BL, 0x1F682198L,
0x5125DAD3L, 0xA34E59D0L, 0xB01EAA24L, 0x42752927L,
0x96BF4DCCL, 0x64D4CECFL, 0x77843D3BL, 0x85EFBE38L,
0xDBFC821CL, 0x2997011FL, 0x3AC7F2EBL, 0xC8AC71E8L,
0x1C661503L, 0xEE0D9600L, 0xFD5D65F4L, 0x0F36E6F7L,
0x61C69362L, 0x93AD1061L, 0x80FDE395L, 0x72966096L,
0xA65C047DL, 0x5437877EL, 0x4767748AL, 0xB50CF789L,
0xEB1FCBADL, 0x197448AEL, 0x0A24BB5AL, 0xF84F3859L,
0x2C855CB2L, 0xDEEEDFB1L, 0xCDBE2C45L, 0x3FD5AF46L,
0x7198540DL, 0x83F3D70EL, 0x90A324FAL, 0x62C8A7F9L,
0xB602C312L, 0x44694011L, 0x5739B3E5L, 0xA55230E6L,
0xFB410CC2L, 0x092A8FC1L, 0x1A7A7C35L, 0xE811FF36L, 
0x3CDB9BDDL, 0xCEB018DEL, 0xDDE0EB2AL, 0x2F8B6829L,
0x82F63B78L, 0x709DB87BL, 0x63CD4B8FL, 0x91A6C88CL,
0x456CAC67L, 0xB7072F64L, 0xA457DC90L, 0x563C5F93L,
0x082F63B7L, 0xFA44E0B4L, 0xE9141340L, 0x1B7F9043L,
0xCFB5F4A8L, 0x3DDE77ABL, 0x2E8E845FL, 0xDCE5075CL,
0x92A8FC17L, 0x60C37F14L, 0x73938CE0L, 0x81F80FE3L,
0x55326B08L, 0xA759E80BL, 0xB4091BFFL, 0x466298FCL,
0x1871A4D8L, 0xEA1A27DBL, 0xF94AD42FL, 0x0B21572CL,
0xDFEB33C7L, 0x2D80B0C4L, 0x3ED04330L, 0xCCBBC033L,
0xA24BB5A6L, 0x502036A5L, 0x4370C551L, 0xB11B4652L,
0x65D122B9L, 0x97BAA1BAL, 0x84EA524EL, 0x7681D14DL,
0x2892ED69L, 0xDAF96E6AL, 0xC9A99D9EL, 0x3BC21E9DL,
0xEF087A76L, 0x1D63F975L, 0x0E330A81L, 0xFC588982L,
0xB21572C9L, 0x407EF1CAL, 0x532E023EL, 0xA145813DL,
0x758FE5D6L, 0x87E466D5L, 0x94B49521L, 0x66DF1622L,
0x38CC2A06L, 0xCAA7A905L, 0xD9F75AF1L, 0x2B9CD9F2L,
0xFF56BD19L, 0x0D3D3E1AL, 0x1E6DCDEEL, 0xEC064EEDL,
0xC38D26C4L, 0x31E6A5C7L, 0x22B65633L, 0xD0DDD530L,
0x0417B1DBL, 0xF67C32D8L, 0xE52CC12CL, 0x1747422FL,
0x49547E0BL, 0xBB3FFD08L, 0xA86F0EFCL, 0x5A048DFFL,
0x8ECEE914L, 0x7CA56A17L, 0x6FF599E3L, 0x9D9E1AE0L,
0xD3D3E1ABL, 0x21B862A8L, 0x32E8915CL, 0xC083125FL,
0x144976B4L, 0xE622F5B7L, 0xF5720643L, 0x07198540L,
0x590AB964L, 0xAB613A67L, 0xB831C993L, 0x4A5A4A90L,
0x9E902E7BL, 0x6CFBAD78L, 0x7FAB5E8CL, 0x8DC0DD8FL,
0xE330A81AL, 0x115B2B19L, 0x020BD8EDL, 0xF0605BEEL,
0x24AA3F05L, 0xD6C1BC06L, 0xC5914FF2L, 0x37FACCF1L,
0x69E9F0D5L, 0x9B8273D6L, 0x88D28022L, 0x7AB90321L,
0xAE7367CAL, 0x5C18E4C9L, 0x4F48173DL, 0xBD23943EL,
0xF36E6F75L, 0x0105EC76L, 0x12551F82L, 0xE03E9C81L,
0x34F4F86AL, 0xC69F7B69L, 0xD5CF889DL, 0x27A40B9EL,
0x79B737BAL, 0x8BDCB4B9L, 0x988C474DL, 0x6AE7C44EL,
0xBE2DA0A5L, 0x4C4623A6L, 0x5F16D052L, 0xAD7D5351L,
};
#endif 
u_int32_t
crc32c(unsigned char *buffer, unsigned int length)
{
unsigned int i;
unsigned long crc32 = ~0L;
unsigned long result;
unsigned char byte0,byte1,byte2,byte3; 
for (i = 0; i < length; i++){
      CRC32C(crc32, buffer[i]);
}
result = ~crc32; 
/* result now holds the negated polynomial remainder;
   * since the table and algorithm is "reflected" [williams95].
   * That is, result has the same value as if we mapped the message
   * to a polynomial, computed the host-bit-order polynomial
   * remainder, performed final negation, then did an end-for-end
   * bit-reversal.
   * Note that a 32-bit bit-reversal is identical to four inplace
   * 8-bit reversals followed by an end-for-end byteswap.
   * In other words, the bytes of each bit are in the right order,
   * but the bytes have been byteswapped. So we now do an explicit
   * byteswap. On a little-endian machine, this byteswap and
   * the final ntohl cancel out and could be elided.
   */ 
byte0 = result & 0xff;
byte1 = (result>>8) & 0xff;
byte2 = (result>>16) & 0xff;
byte3 = (result>>24) & 0xff; 
crc32 = ((byte0 << 24) |
           (byte1 << 16) |
           (byte2 << 8) |
           byte3);
return ( crc32 );
}
 
4. 结论 
Linux内核网络编程中经常会遇到重新计算校验和的问题,这个基本功一定是要掌握的,其实内核中已经提供了大量的校验和计算函数供使用,要尽量使用这些已有函数而不必自己重新编写。

你可能感兴趣的:(网络校验和计算)