Write code to implement the following function:
/*
*Generate mask indicating leftmost 1 in x. Assume w=32.
*For example 0xFF00 -> 0x8000, and 0x6600 -> 0x4000.
* If x = 0, then return 0.
*/
int leftmost_one(unsigned x);
Your function should follow the bit-level integer coding rules(page 120), except that you may assume
that data type int has w=32 bits.
Your code should contain a total of at most 15 arithmetic, bit-wise, and logical operations.
Hint: Fisrt transform x into a bit vector of the form[0...011...1].
int leftmost_one(unsigned x)
{
unsigned tmp;
//构造从原始x最高位往右所有位都为1的无符号数
tmp = x >> 1;
x |= tmp;
tmp = x >> 2;
x |= tmp;
tmp = x >> 4;
x |= tmp;
tmp = x >> 8;
x |= tmp;
tmp = x >> 16;
x |= tmp;
unsigned y,z;
y = x + 1;
z = y;
y >>= 1;
//倘若原始x最高位为31位
z == 0 && (y = 0x8000);
return y;
}
分析:
题目要求:为数A(0101)产生一个掩码M,A&M结果只保留A中最高位的1,即(0100)。所以(0100)就是A的掩码。
原理:若A为(0101 0101),若能得到B(0111 1111),则B+1得到C(1000 0000),然后C右移一位就可得到要求的掩码M(0100 0000)。
那么关键就是得到上述的B,若A为(0101 0101)(假设从右数第n位为最高位1):
A | (A >> 1) ---> B1(011* ****) , 不管*代表0还是1,现在可得到第n,n-1位为1的数B1。
B1 | (B1>>2) --->B2(0111 1***),不管*代表0还是1, 现在可得到第n,n-1,n-2,n-3位为1的数B2.
B2 | (B2>>4) ---->B3(0111 1111) , 此时,可得到B
显然,若A共w位,按上述过程,当Bi | (Bi>>w/2)一定可以得到B
有一个特殊情况,若A的最高位1恰好在w-1位上,得到的B就是(11......11)全1,这时候B+1得到C(00......00)全0,此时就用到
代码中
z == 0 && (y = 0x8000),
即特殊情况下直接返回0x8000