"Linux Network Development Course" 3_Introduction to IP address

Question: What is the meaning of some parameters in the network programming interface?

 sock = socket(PF_INET, SOCK_STREAM, 0);

Detailed explanation of socket parameters

int socket(int domain, int type, int protocal);

parameter	significance
domain	Protocol family information used in sockets
type	Socket data transfer type information
prorocol	Protocol information used for communication between devices

domain parameter (protocol family) in socket()

• PF_INET → IPv4 Internet Protocol Suite
• PF_INET6 → IPv6 Internet Protocol Suite
• PF_LOCAL → protocol family for local communication
• PF_PACKET → low-level data transceiver protocol
• PF_IPX → Novell Proprietary Protocol (Internet Packet Exchange Protocol)
• ...

 注意：不同协议中的地址表现形式可能不同，网络编程时地址类型必须和协议类型匹配

type and protocol parameters in socket()

• type : used to specify the protocol type

  • SOCK_STREAM : streaming data (TCP)
  • SOCK_UGRAM : Packet data (UDP)

• protocol : used to specify the specific protocol that the protocol family conforms to the type

  • Domain and type can almost uniquely determine a protocol, so this parameter is usually 0
  • That is: 0 represents the default protocol after domain and type are specified

About port numbers and IP addresses

• The port number is a 2-byte data (unsigned)
• 0 - 1024 are predefined as a specific port (assigned to a specific application)
• IP address is a 4-byte unsigned address family (can be divided into 5 types of addresses)

Deep dive into IP addresses

IP address is divided into two parts: network ID and host ID

• Network ID: Identifies the network where the network host (device) is located
• Host ID: Identify the specific address of the network host (device)

Question: An IP address is only 4 bytes, so how to distinguish the network ID from the host ID?

• IP Address and Subnet Mask Use Together to Distinguish Network ID and Host ID
• The representation of the subnet mask is also a 4-byte integer (unsigned)
• The subnet mask is used to extract the network identity (& operation) from the IP address

Understanding Subnet Masks in Depth

 设：子网掩码为 M.N.P.Q，则子网可用 IP 地址数量 n = (256 - M) * (256 - N) * (256 - P) * (256 - Q)

例：IP 地址 211.99.34.33，掩码 255.255.2555.248，因此：211.99.34.33 所在子网有 8 个 IP 地址
所在子网地址： 211.99.34.32 
广播地址：211.99.34.39
6个可分配地址：211.99.34.33 ... 211.99.34.38

注：子网地址即为网络标识

 ip 地址 211.99.34.33， 掩码 255.255.225.248
可知 211.99.34.33 所在子网有 8 个 IP 地址， 且 8 = 2^3（二的三次方），所以 Y = 32 - 3 = 29
可表示为 211.99.34.33 / 29 【简写形式】

注：29 为32位子网掩码的高位

do the math

 IP 地址 192.168.3.44，掩码 255.255.255.0

问：
    子网地址是什么？广播地址是什么？可用地址有多少？简洁表示法是什么？

答：
    子网地址：192.168.3.44
    广播地址：192.168.3.255
    可用地址：254 [256 - 0(子网地址) - 255(广播地址)]
    简洁表示：192.168.3.44 / 24

special address

• 0.0.0.0 / 0 - reserved, often used to represent "default network"
• 127.0.0.0 / 8 - loopback address, commonly used for local software testing
• 255.255.255.255/32 - broadcast address

Address Types in Network Programming

 int sock = 0;
struct sockaddr_in addr = {0};

sock = socket(PF_INET, SOCK_STREAM, 0);  // Protocol Family，协议族

if (sock == -1) {
    pritnf("socket error\n");
    return;
}

addr.sin_family = AF_INET;  // Address Family，地址族
addr.sin_addr.s_addr = inet_addr("192.168.3.241");
addr.sin_addr.port = htons(8899);

if (connect(sock, (struct sockaddr*)&addr, sizeof(addr)) == -1) {
    printf("connect error\n");
    return -1;
}

Question: Will the cast of (struct sockaddr*)&addr not be a problem?

Address data type resolution

• struct sockaddr can be understood as the top-level address type parent class, which has the same memory layout as the subclass
• connect() According to addr.sin_family to determine which address type it is and analyze accordingly

IP address related functions

#include <arpa/inet.h>

function prototype	Function description
in_addr_t inet_addr(const char* strptr);	Convert IP string to network byte order integer
int inet_aton(const char *cp, struct_addr *inp);	Convert the IP string to an integer conforming to the network byte order, returning 1 for success and 0 for failure
char *inet_ntoa(struct in_addr in)	Converts an integer address in network byte order to string form

Programming experiment: address function experiment

 #include <sys/types.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <stdio.h>
#include <unistd.h>
#include <string.h>
#include <malloc.h>

int main()
{
    unsigned int addr = inet_addr("1.2.3.4");
    struct in_addr addr1 = {0x09080706};
    struct in_addr addr2 = {0x05040302};
    char *s1 = inet_ntoa(addr1);
    char *s1_s = strcpy(malloc(32), s1);
    char *s2 = inet_ntoa(addr2);
    char *s2_s = strcpy(malloc(32), s2);

    printf("addr = %x\n", addr);

    printf("addr1 = %x\n", addr1.s_addr);
    printf("addr2 = %x\n", addr2.s_addr);

    printf("s1 = %s\n", s1);
    printf("s2 = %s\n", s2);               // 注意这里 ！！
    printf("s1 == s2 : %d\n", s1 == s2);   // 注意这里 ！！

    printf("s1_s = %s\n", s1_s);
    printf("s2_s = %s\n", s2_s);
    printf("s1_s == s2_s : %d\n", s1_s == s2_s);

    if (inet_aton("D.T.Software", &addr1)) {  // 注意这里 ！！
        printf("addr1 = %x\n", addr1.s_addr);
    }
    
    free(s1_s);
    free(s2_s);

    return 0;
}

output

 addr = 4030201
addr1 = 9080706
addr2 = 5040302
s1 = 2.3.4.5
s2 = 2.3.4.5   // 注意，转换结果被覆盖 ！！
s1 == s2 : 1   // 两次为同一个地址 ！！
s1_s = 6.7.8.9
s2_s = 2.3.4.5
s1_s == s2_s : 0

---

Remaining problem: how to enhance the server's capabilities to support multiple clients at the same time? What is multicast? What is broadcasting?