原始套接字的 udp 数据包分段

2022-01-22 00:00:00 sockets udp c++ fragmentation

问题跟进原始套接字的数据包碎片

如果我有一个这样实现的原始套接字:

If I have a raw socket implemented as such:

  if ((sip_socket = socket(AF_INET, SOCK_RAW, IPPROTO_RAW)) < 0)
      {
    cout << "Unable to create the SIP sockets."<< sip_socket<<" 
";
    return -3;
      }

   if ( setsockopt(sip_socket, IPPROTO_IP, IP_HDRINCL, &one, sizeof(one)) == -1)
      {
   cerr << "Unable to set option to Raw Socket.
";
   return -4;
      };

如果我的数据包大小为 1756(不包括 IP 标头)，我该如何设置 ipHdr->fragment_offset(16 位，包括 3 位标志)?
我是否需要准备两个数据包――一个大小为 1480，另一个大小为 276，然后在两个数据包上打 IP 标头?

how can I set the ipHdr->fragment_offset (16 bits including 3 bit flags) if I have a packet of size 1756 (not including the IP header)?
Do I need to prepare two packets-one of size 1480 and another of size 276, and then slap IP headers on both packets?

谁能指出一个示例代码?

Can anyone point to an example code for this?

推荐答案

是的，你需要准备两个包，每个包都有自己的IP头.

Yes, you need to prepare two packets, each with their own IP header.

如果您在第一个数据包中放入 1480 个字节的数据，在第二个数据包中放入 276 个字节，那么 IP 标头应该是相同的，除了这些字段:

If you put 1480 bytes of data in the first packet and 276 in the second, then the IP headers should be identical, except for these fields:

  • Fragment Offset:第一个包设置为0，第二个包设置为1480；
  • Total Length:设置为1480加上第一个包的头长度，276加上第二个包的头长度；
  • MF标志:在第一个包中设置为1，在第二个包中设置为0；
  • 标头校验和:根据不同的标头重新计算.
  • Fragment Offset: set to 0 in the first packet, and 1480 in the second;
  • Total Length: set to 1480 plus the header length in the first packet, and 276 plus the header length in the second packet;
  • MF flag: set to 1 in the first packet and 0 in the second;
  • Header Checksum: Recalculated over the different headers as appropriate.

相关文章