透析ICMP协议(四): 牛刀初试之二 应用篇ping(RAW Socket) (转)[@more@]

透析ICMP(四): 牛刀初试之二

 应用篇(RAW Socket)===============================这篇文章出自free/CSDN平台: VC6 XP

今晚一杯茶水之后, 让我们继续我们的ICMP讨论, 今晚介绍的是用RAW Socket 实现的ping.

原理简介:

--------  用RAW Socket实现的ping可能比上一节的应用ICMP.DLL的程序庞大些, 但是这才是我们需要关注的东西, 我的观点真正想做开发的程序员应该静下心来读读这篇文章, 相信你会从中获益颇多. 中间我也会讲解一些东西为后一章的追踪做一些铺垫.  另一个重要的要讲的东西, 宣布随时不支持上节讲的ping用到的开发接口, 但是本节的讲的是更一般的东西. 所以它不会过时, 甚至做很小的改动就可以移植到别的上去. 系统移植不是我们的讲的重点. 但是微软的长期支持足以引起我们充分的重视. 如何少作变动来使的这个程序实现追踪路由的功能, 这里只是抛砖引玉. 将ICMP包中IP包的包头该为特定的值就能得到那个路由器的IP(要求到达目的地的跳数大于你设的特定值). 这个程序需要windows2k/WindowsXP/WindowsNT平台和员的.

具体实现:

--------这段大部分来自:  /wskfaq/examples/rawping.html">http://tangentsoft.net/wskfaq/examples/rawping.html[bugfree]只做了少量修改,给出了大量的注释, 最后结合给出了自己的建议.

----------

/*

 * 程序名: rawping_.cpp * 说明:   *  程序,也是主 */#include   #include

#include "rawping.h"

#define DEFAULT_PACKET_SIZE 32  // 默认ICMP包字节数

#define DEFAULT_TTL 30  // 默认TTL值#define MAX_PING_DATA_SIZE 1024  // 最大数据块 #define MAX_PING_PACKET_SIZE (MAX_PING_DATA_SIZE + sizeof(IPHeader)) //最大ICMP包长度

/* 为 send_buf 和 recv_buf 分配

 * send_buf大小为 packet_size * recv_buf大小为 MAX_PING_PACKET_SIZE, 保证大于send_buf */int allocate_buffers(ICMPHeader*& send_buf, IPHeader*& recv_buf,  int packet_size);

///// Program entry point

int main(int argc, char* argv[])

{     int seq_no = 0;  //用在发送和接受的ICMP包头中  ICMPHeader* send_buf = 0;   IPHeader* recv_buf = 0;

  // 判断命令行是否合法

  if (argc < 2) {   cerr << "usage: " << argv[0] << " [data_size] [ttl]" <<  endl;  cerr << "tdata_size can be up to " << MAX_PING_DATA_SIZE <<  " bytes.  Default is " << DEFAULT_PACKET_SIZE << "." <<   endl;   cerr << "tttl should be 255 or lower.  Default is " <<  DEFAULT_TTL << "." << endl;  return 1;  }

  // 处理命令行参数

  int packet_size = DEFAULT_PACKET_SIZE;  int ttl = DEFAULT_TTL;  if (argc > 2) {   int temp = atoi(argv[2]);  if (temp != 0) {   packet_size = temp;  }  if (argc > 3) {   temp = atoi(argv[3]);  if ((temp >= 0) && (temp <= 255)) {   ttl = temp;  }  }  }  packet_size = max(sizeof(ICMPHeader),   min(MAX_PING_DATA_SIZE, (unsigned int)packet_size));

  // 启动 Winsock

  WSAData wsaData;  if (Wtartup(MAKE(2, 1), &wsaData) != 0) {   cerr << "Failed to find Winsock 2.1 or better." << endl;  return 1;  }

  SOCKET sd; // RAW Socket句柄

  sockaddr_in dest, ;     // 三个任务(创建sd, 设置ttl, 初试dest的值)  if (setup_for_ping(argv[1], ttl, sd, dest) < 0) {   goto cleanup; //释放资源并退出  }  // 为send_buf和recv_buf分配内存   if (allocate_buffers(send_buf, recv_buf, packet_size) < 0) {   goto cleanup;  }  // 初试化IMCP数据包(type=8,code=0)  init_ping_packet(send_buf, packet_size, seq_no);

  // 发送ICMP数据包

  if (send_ping(sd, dest, send_buf, packet_size) >= 0) {   while (1) {   // 接受回应包  if (recv_ping(sd, source, recv_buf, MAX_PING_PACKET_SIZE) <  0) {   // Pull the sequence number out of the ICMP header.  If   // it's bad, we just complain, but otherwise we take   // off, because the read failed for some reason.  unsigned short header_len = recv_buf->h_len * 4;  ICMPHeader* icmphdr = (ICMPHeader*)  ((char*)recv_buf + header_len);  if (icmphdr->seq != seq_no) {   cerr << "bad sequence number!" << endl;  continue;  }  else {   break;  }  }  if (decode_reply(recv_buf, packet_size, &source) != -2) {   // Success or al error (as opposed to a minor error)   // so take off.  break;  }  }  }

cleanup:

  delete[]send_buf;  //释放分配的内存  delete[]recv_buf;  WSACleanup(); // 清理winsock  return 0;}

// 为send_buf 和 recv_buf的内存分配. 太简单, 我略过

int allocate_buffers(ICMPHeader*& send_buf, IPHeader*& recv_buf,  int packet_size){   // First the send buffer  send_buf = (ICMPHeader*)new char[packet_size];    if (send_buf == 0) {   cerr << "Failed to allocate output buffer." << endl;  return -1;  }

  // And then the receive buffer

  recv_buf = (IPHeader*)new char[MAX_PING_PACKET_SIZE];  if (recv_buf == 0) {   cerr << "Failed to allocate output buffer." << endl;  return -1;  }    return 0;}

/*

 * 程序名: rawping.h * 说明:   *  主要函数库头 */

#define _LEAN_AND_MEAN

#include

// ICMP 包类型, 具体参见本文的第一节

#define ICMP_ECHO_REPLY 0  #define ICMP_DEST_UNREACH 3#define ICMP_TTL_EXPIRE 11#define ICMP_ECHO_REQUEST 8

// 最小的ICMP包大小

#define ICMP_MIN 8

// 包头struct IPHeader {   BYTE h_len:4;  // Length of the header in dwords  BYTE version:4;  // Version of IP  BYTE tos;  // Type of service  USHORT total_len;  // Length of the packet in dwords  USHORT nt;  // unique identifier  USHORT flags;  // Flags  BYTE ttl;  // Time to live, 这个字段我在下一节中用来实现Tracert功能  BYTE proto;  // Protocol number (TCP, UDP etc)  USHORT checksum;  // IP checksum  ULONG source_ip;  ULONG dest_ip;};

// ICMP 包头(实际的包不包括timestamp字段,

// 作者用来计算包的回应时间,其实完全没有必要这样做)struct ICMPHeader {   BYTE type;  // ICMP packet type  BYTE code;  // Type sub code  USHORT checksum;  USHORT id;  USHORT seq;  ULONG timestamp;  // not part of ICMP, but we need it};

extern USHORT ip_checksum(USHORT* buffer, int size);extern int setup_for_ping(char* host, int ttl, SOCKET& sd,  sockaddr_in& dest);extern int send_ping(SOCKET sd, const sockaddr_in& dest, ICMPHeader* send_buf, int packet_size);extern int recv_ping(SOCKET sd, sockaddr_in& source, IPHeader* recv_buf,  int packet_size);extern int decode_reply(IPHeader* reply, int bytes, sockaddr_in* from);extern void init_ping_packet(ICMPHeader* icmp_hdr, int packet_size, int seq_no);

/*

 * 程序名: rawping.cpp * 说明:   *  主要函数库实现部分 */ include #include #include #include "rawping.h"  // 计算ICMP包的校验和的简单算法, 很多地方都有说明, 这里没有必要详细将 // 只是一点要提, 做校验之前, 务必将ICMP包头的checksum字段置为0USHORT ip_checksum(USHORT* buffer, int size) {   unsigned long cksum = 0;    // Sum all the words together, adding the final byte if size is odd  while (size > 1) {   cksum += *buffer++;  size -= sizeof(USHORT);  }  if (size) {   cksum += *(UCHAR*)buffer;  }

  // Do a little shuffling

  cksum = (cksum >> 16) + (cksum & 0xffff);  cksum += (cksum >> 16);    // Return the bitwise complement of the resulting mishmash  return (USHORT)(~cksum);}

//初试化RAW Socket, 设置ttl, 初试化dest

// 返回值 <0 表失败

int setup_for_ping(char* host, int ttl, SOCKET& sd, sockaddr_in& dest)

{   // Create the socket  sd = WSASocket(AF_INET, SOCK_RAW, IPPROTO_ICMP, 0, 0, 0);  if (sd == INVALID_SOCKET) {   cerr << "Failed to create raw socket: " << WSAGetLastError() <<  endl;  return -1;  }

  if (setsockopt(sd, IPPROTO_IP, IP_TTL, (const char*)&ttl,

  sizeof(ttl)) == SOCKET_ERROR) {   cerr << "TTL setsockopt failed: " << WSAGetLastError() << endl;  return -1;  }

  // Initialize the destination host info block

  memset(&dest, 0, sizeof(dest));

  // Turn first passed parameter into an IP address to ping

  unsigned int addr =_addr(host);  if (addr != INADDR_NONE) {   // It was a dotted quad number, so save result  dest.sin_addr.s_addr = addr;  dest.sin_family = AF_INET;  }  else {   // Not in dotted quad form, so try and look it up  hostent* hp = gethostbyname(host);  if (hp != 0) {   // Found an address for that host, so save it  memcpy(&(dest.sin_addr), hp->h_addr, hp->h_length);  dest.sin_family = hp->h_addrtype;  }  else {   // Not a recognized hostname either!  cerr << "Failed to resolve " << host << endl;  return -1;  }  }

  return 0;

}

 

//初试化ICMP的包头, 给data部分填充数据, 最后计算整个包的校验和

void init_ping_packet(ICMPHeader* icmp_hdr, int packet_size, int seq_no)

{   // Set up the packet's fields  icmp_hdr->type = ICMP_ECHO_REQUEST;  icmp_hdr->code = 0;  icmp_hdr->checksum = 0;  icmp_hdr->id = (USHORT)GetCurrentProcessId();  icmp_hdr->seq = seq_no;  icmp_hdr->timestamp = GetTickCount();

  // "You're dead meat now, packet!"

  const unsigned long int deadmeat = 0xDEAEEF;  char* datapart = (char*)icmp_hdr + sizeof(ICMPHeader);  int bytes_left = packet_size - sizeof(ICMPHeader);  while (bytes_left > 0) {   memcpy(datapart, &deadmeat, min(int(sizeof(deadmeat)),   bytes_left));  bytes_left -= sizeof(deadmeat);  datapart += sizeof(deadmeat);  }

  // Calculate a checksum on the result

  icmp_hdr->checksum = ip_checksum((USHORT*)icmp_hdr, packet_size);}

// 发送生成的ICMP包

// 返回值 <0 表失败

int send_ping(SOCKET sd, const sockaddr_in& dest, ICMPHeader* send_buf,

  int packet_size){   // Send the ping packet in send_buf as-is  cout << "Sending " << packet_size << " bytes to " <<   inet_ntoa(dest.sin_addr) << "..." << flush;  int bwrote = sendto(sd, (char*)send_buf, packet_size, 0,   (sockaddr*)&dest, sizeof(dest));  if (bwrote == SOCKET_ERROR) {   cerr << "send failed: " << WSAGetLastError() << endl;  return -1;  }  else if (bwrote < packet_size) {   cout << "sent " << bwrote << " bytes..." << flush;  }

  return 0;

}

// 接受ICMP包// 返回值 <0 表失败int recv_ping(SOCKET sd, sockaddr_in& source, IPHeader* recv_buf,   int packet_size){   // Wait for the ping reply  int fromlen = sizeof(source);  int bread = recvfrom(sd, (char*)recv_buf,   packet_size + sizeof(IPHeader), 0,  (sockaddr*)&source, &fromlen);  if (bread == SOCKET_ERROR) {   cerr << "read failed: ";  if (WSAGetLastError() == WSAEMSGSIZE) {   cerr << "buffer too small" << endl;  }  else {   cerr << "error #" << WSAGetLastError() << endl;  }  return -1;  }

  return 0;

}

// 对收到的ICMP解码// 返回值 -2表忽略, -1 表失败, 0 成功int decode_reply(IPHeader* reply, int bytes, sockaddr_in* from) {   // 跳过IP包头, 找到ICMP的包头  unsigned short header_len = reply->h_len * 4;  ICMPHeader* icmphdr = (ICMPHeader*)((char*)reply + header_len);

  // 包的长度合法, header_len + ICMP_MIN为最小ICMP包的长度

  if (bytes < header_len + ICMP_MIN) {   cerr << "too few bytes from " << inet_ntoa(from->sin_addr) <<   endl;  return -1;  }  // 下面的包类型详细参见我的第一部分 "透析ICMP协议(一): 协议原理"  else if (icmphdr->type != ICMP_ECHO_REPLY) {  //非正常回复  if (icmphdr->type != ICMP_TTL_EXPIRE) {  //ttl减为零  if (icmphdr->type == ICMP_DEST_UNREACH) { //主机不可达  cerr << "Destination unreachable" << endl;  }  else {  //的ICMP包类型  cerr << "Unknown ICMP packet type " << int(icmphdr->type) <<  " received" << endl;  }  return -1;  }  }  else if (icmphdr->id != (USHORT)GetCurrentProcessId()) {  //不是本进程发的包, 可能是同机的其它ping进程发的  return -2;  }

  // 指出包传递了多远

  // [bugfree]我认为作者这里有问题, 因为有些系统的ttl初值为128如,   // 有些为256如我的211.97.168.129, 作者假设为256有点武断,   // 可以一起探讨这个问题, 回e:zhangliangsd@.com  int nHops = int(256 - reply->ttl);  if (nHops == 192) {   // TTL came back 64, so ping was probably to a host on the  // LAN -- call it a single hop.  nHops = 1;  }  else if (nHops == 128) {   // Probably localhost  nHops = 0;  }

  // 所有工作结束,打印信息

  cout << endl << bytes << " bytes from " <<   inet_ntoa(from->sin_addr) << ", icmp_seq " <<   icmphdr->seq << ", ";  if (icmphdr->type == ICMP_TTL_EXPIRE) {   cout << "TTL expired." << endl;  }  else {   cout << nHops << " hop" << (nHops == 1 ? "" : "s");  cout << ", time: " << (GetTickCount() - icmphdr->timestamp) <<  " ms." << endl;  }

  return 0;

}

总结和建议:-----------  bugfree建议其中的这些方面需要改进:   1. 头文件iostream.h 改为 iostream, 后者是标准C++的头文件    同时添加对std::cout 和 std::endl;的引用    对于cerr 建议都改为std::cout(因为后者头文件不支持)   2. 程序的发送和接受采用了同步的方式, 这使得如果出现网络问题recv_ping将陷入持续等待.    这是我们不想看到的.     这三种技术可以达到目的:    - 使用多线程, 将ping封装进线程, 在主程序中对它的超时进行处理    - 使用()函数来实现    - 使用windows的 WSAAsyncSelect()    这里对这些方法不作具体讨论, 留给读者自已完成. 

 

链接:

-------我的其它文章,<>, 和其它文章参见:

联系方式:

------- 

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/10748419/viewspace-975029/，如需转载，请注明出处，否则将追究法律责任。