关于有符号和无符号的移位运算

[cpp] view plaincopy

#include <stdio.h>

int main()

{

unsigned char a = 0xA5;

unsigned char c = ~a;

unsigned char b = c >> 4;

printf("b=%d/n", b);

return 0;

}

结果为:b=5

但是写成如下形式：

[cpp] view plaincopy

#include <stdio.h>

int main()

{

unsigned char a = 0xA5;

unsigned char b = ~a >> 4;

printf("b=%d/n", b);

return 0;

}

结果为:b=245

另一种情况:

[cpp] view plaincopy

#include <stdio.h>

int main()

{

char a = 0xA5;

unsigned char b = ~a >> 4;

printf("b=%d/n", b);

return 0;

}

结果为b=5;



因为在32位系统上，位运算是按32位计算的。而且，按符号位扩展。(无符号数，在前边补0，有符号数，按当前符号位来补)

在第1种情况中，a为无符号数，

a = 0xA5; a=10100101;扩展为a=00000000 00000000 00000000 10100101

c=~a; ~a= 11111111 11111111 11111111 01011010; c = 01011010;

b = c >> 4;

c=00000000 00000000 0000000 01011010;

c>>4=00000000 0000000 00000000 00000101; b = 00000101 = 5;

第2种情况,a为无符号数

a = 0xA5; a=10100101;扩展为a=00000000 00000000 00000000 10100101

~a=11111111 11111111 11111111 01011010;

~a>>4=11111111 11111111 11111111 11110101;

b = ~a>>4 = 11110101 = 245;

第3种情况,a为有符号数

a = 0xA5; a=10100101;扩展为a=11111111 11111111 11111111 10100101

~a=00000000 00000000 00000000 01011010

~a>>4=00000000 00000000 00000000 00000101

b = ~a >> 4 = 00000101 = 5