我有一个函数来解析20个字节长的IP头缓冲区:
void parseIp(struct ipHeader *ip, const void *buffer)
{
uint8_t* b = buffer;
// memcpy(b,buffer,20);
ip->version = (b[0] & 0xf0) >> 4;
ip->ihl = (b[0] & 0x0f);
ip->dscp = (b[1] & 0xfC)>>2;
ip->ecn = (b[1] & 0x3);
unsigned short l = (b[2] << 8) | b[3];
printf("%d\n",l);
ip->length = l;
ip->identification = (b[4] << 0xFF) | b[5];
}
struct ipHeader:
struct ipHeader {
int version;
int ihl;
int dscp;
int ecn;
unsigned short length;
unsigned short identification;
int flags;
int fragment_offset;
int time_to_live;
int protocol;
unsigned short header_checksum;
unsigned char source_ip[4];
unsigned char destination_ip[4];
};
现在代码将l打印为467,这是正确的,但是当这个l被分配给struct字段长度时它会变为54017.我根本不明白发生了什么。我添加了变量l以确保不会发生溢出或类型转换错误,但它仍然会发生变化。
这是学校工作的一部分,所以我无法改变结构。
EDIT 完整代码:
#include <stdio.h>
#include <arpa/inet.h>
#include "ipheader.h"
/* Parses the given buffer into an IP header structure.
*
* Parameters:
* ip: pointer to the IP header structure that will be filled based
* on the data in the buffer
* buffer: buffer of 20 bytes that contain the IP header. */
void parseIp(struct ipHeader *ip, const void *buffer)
{
uint8_t* b = buffer;
// memcpy(b,buffer,20);
ip->version = (b[0] & 0xf0) >> 4;
ip->ihl = (b[0] & 0x0f);
ip->dscp = (b[1] & 0xfC)>>2;
ip->ecn = (b[1] & 0x3);
unsigned short l = (b[2] << 8) | b[3];
printf("%d\n",l);
ip->length = l;
ip->identification = (b[4] << 8) | b[5];
}
/* Builds a 20-byte byte stream based on the given IP header structure
*
* Parameters:
* buffer: pointer to the 20-byte buffer to which the header is constructed
* ip: IP header structure that will be packed to the buffer */
void sendIp(void *buffer, const struct ipHeader *ip)
{
}
/* Prints the given IP header structure */
void printIp(const struct ipHeader *ip)
{
/* Note: ntohs below is for converting numbers from network byte order
to host byte order. You can ignore them for now
To be discussed further in Network Programming course... */
printf("version: %d ihl: %d dscp: %d ecn: %d\n",
ip->version, ip->ihl, ip->dscp, ip->ecn);
printf("length: %d id: %d flags: %d offset: %d\n",
ntohs(ip->length), ntohs(ip->identification), ip->flags, ip->fragment_offset);
printf("time to live: %d protocol: %d checksum: 0x%04x\n",
ip->time_to_live, ip->protocol, ntohs(ip->header_checksum));
printf("source ip: %d.%d.%d.%d\n", ip->source_ip[0], ip->source_ip[1],
ip->source_ip[2], ip->source_ip[3]);
printf("destination ip: %d.%d.%d.%d\n", ip->destination_ip[0],
ip->destination_ip[1],
ip->destination_ip[2], ip->destination_ip[3]);
}
/* Shows hexdump of given data buffer */
void hexdump(const void *buffer, unsigned int length)
{
const unsigned char *cbuf = buffer;
unsigned int i;
for (i = 0; i < length; ) {
printf("%02x ", cbuf[i]);
i++;
if (!(i % 8))
printf("\n");
}
}
struct ipHeader {
int version;
int ihl;
int dscp;
int ecn;
unsigned short length;
unsigned short identification;
int flags;
int fragment_offset;
int time_to_live;
int protocol;
unsigned short header_checksum;
unsigned char source_ip[4];
unsigned char destination_ip[4];
};
void parseIp(struct ipHeader *ip, const void *buffer);
void sendIp(void *buffer, const struct ipHeader *ip);
void printIp(const struct ipHeader *ip);
void hexdump(const void *buffer, unsigned int length);
#include <arpa/inet.h>
#include "ipheader.h"
int main()
{
/* Feel free to modify this function to test different things */
unsigned char bytes[] = {
0x45, 0x00, 0x01, 0xd3, 0xda, 0x8d, 0x40, 0x00,
0x40, 0x06, 0x8c, 0xd5, 0xc0, 0xa8, 0x01, 0x46,
0x6c, 0xa0, 0xa3, 0x33 };
struct ipHeader ip;
parseIp(&ip, bytes);
printIp(&ip);
struct ipHeader ipfields = {
4, // version
28, // ihl
4, // dscp
0, // ecn
htons(1500), // length
htons(1234), // id
1, // flags
1024, // offset
15, // time_to_live
33, // protocol
htons(0x1234), // checksum (invalid)
{1, 2, 3, 4}, // source IP
{5, 6, 7, 8} // destination IP
};
unsigned char sendbuf[20];
sendIp(sendbuf, &ipfields);
hexdump(sendbuf, sizeof(sendbuf));
}
答案 0 :(得分:1)
对于给定的输入:
unsigned char bytes[] = { 0x45, 0x00, 0x01, 0xd3, 0xda,
然后是代码:
unsigned short l = (b[2] << 8) | b[3];
生成l
,其值为467
。
你在问题中说,&#34;因为这个l被分配给结构字段长度,所以它变为54017。&#34;。但事实并非如此。如果您在现有ip->length = l;
之后立即添加一行:
printf("%d\n", ip->length);
您仍会看到467
。
我猜你提到的问题是你的printIp
函数打印54017
。这是因为该函数不会打印ip->length
。它打印ntohs(ip->length)
。 ntohs
宏将值从567
更改为54017
。
要解决此问题,请更改printIp
功能以打印ip->length
,而不是ntohs(ip->length)
。
从该功能中删除其他ntohs
来电,并从htons
的定义中删除ipfields
。整数应该存储在struct ipHeader
内的 h ost顺序(即本机顺序)中,并存储在 n etwork order(即big-endian)中在unsigned char
缓冲区时。
可移植性注1:从技术上讲,您应该在这两个%hu
语句中使用printf
作为格式说明符,因为参数类型为unsigned short
。
可移植性注2: l == 467
无论int
大小,都与目前为止的一些评论/答案中的建议相反。但是,要在具有16位b[2]
的系统上运行时支持0x7F
大于int
的值,您应该编写((unsigned)b[2] << 8) | b[3]
。
可移植性注3:最好使用uint16_t
代替unsigned short
,因为现在系统有32位unsigned short
。如果这样做,printf格式说明符为"%"PRI16u
,您可能需要#include <inttypes.h>
答案 1 :(得分:-1)
不确定您是否知道Endianness(Big endian / Little endian) 请参阅:https://en.wikipedia.org/wiki/Endianness
基本上,Little-endian格式反转顺序并将最低有效字节存储在较低的存储器地址,最高有效字节存储在最高存储器地址。
因此,当您指定I(467 = 0x1d3)时,它会以小端格式存储,具体取决于您的机器字节顺序(0xd301 = 54017)。
如果要分配正确的值,请使用htons。
答案 2 :(得分:-2)
正如其他人所提到的那样,看起来你可能会遇到一些转移问题,因为你的数据类型超出了宽度,导致......? (我不知道,似乎有关于这是否是定义的行为的争论)。在我的机器上,你的代码导致467这是正确的,幸运的是。然而,显式地定义引用定义了你想要的东西。
unsigned short l = (((unsigned short)b[2]) << 8) | ((unsigned short)b[3]);
此外,你需要担心endianess(你是),因为网络头代码应该总是大端,我个人不会担心如果我没有必要的位移。对于多字节和字节边界的标题部分,我会做这样的事情:
ip->length = ntohs(((unsigned short*)b)[1]);
ip->identification = ntohs(((unsigned short*)b)[2]);
ip->header_checksum = ntohs(((unsigned short*)b[5]);
/*
unsigned int sourceIpAddr = ntohl(((unsigned int*)b)[3]);
unsigned int destIpAddr = ntohl(((unsigned int*)b[4]);
Not sure what endianess you want for the source and destination IPs since those are just byte arrays
*/
注意,当您将b
强制转换为其他指针类型时,索引会发生变化。
如果我完全控制结构,我会使用位字段创建整个事物,那么你根本不必担心转移,但你说结构是为你定义的。