Optimising bitwise operations in C

Question

I had a problem in hand as this : "Exercise 2-6. Write a function setbits(x,p,n,y) that returns x with the n bits that begin at position p set to the rightmost n bits of y, leaving the other bits unchanged."

I've written a function for this as below. This is working as expected.

int func_setx(int x,int p,int n,int y)
{
    int a_t= ~0 << (p+n);
    int b_t= ~a_t >> n;
    int x_t= x& (a_t | b_t);    // a temporary x which has the bits to be changed as 0 , rest of the bits unchanged.

    int mask= (y << p) & (~(~0 << (p+n)));     // a mask which has required bits from y in positions to be set , rest all bits 0.

    printf("\nCheckpoint : mask= %x  x_t= %x\n",mask,x_t);

    int result= mask|x_t;
    return result;
}

But I somehow feel the logic is long and can be optimized, but can't think of any other way yet. Can anyone suggest any optimization to this please?

You'll probably find an answer at [Bit Twiddling Hacks](http://graphics.stanford.edu/~seander/bithacks.html) — and you should bookmark the page even if you don't find the answer there. — Jonathan Leffler, Jun 27 '13 at 05:14
Numerous duplicates, e.g. [k&r exercise confusion with bit-operations](http://stackoverflow.com/questions/2076084/kr-exercise-confusion-with-bit-operations) and [returns x with the n bits that begin at position p set to the rightmost n bits of y, leaving other bits unchanged](http://stackoverflow.com/questions/6727351/returns-x-with-the-n-bits-that-begin-at-position-p-set-to-the-rightmost-n-bits-o) and [Setting Bits in C](http://stackoverflow.com/questions/11590545/setting-bits-in-c) — Paul R, Jun 27 '13 at 08:32

Carl Norum · Accepted Answer · 2013-06-27T04:51:14.410

10

To make an n bit mask:

mask_y = (1U << n) - 1;

To start it at bit p:

mask_x = mask_y << p;

Clear the appropriate bits in x:

x &= ~mask_x;

Extract the bits from y:

y &= mask_y;

Upshift them to position p:

y <<= p;

Put it all together:

result = x | y;

Or in a more compact form:

mask = (1U << n) - 1;
result = x & ~(mask << p);
result |= (y & mask) << p;

edited Jun 27 '13 at 04:51

answered Jun 27 '13 at 04:37

Carl Norum

219,201
40
422
469

1

Assuming 32-bit integers, if `n == 32`, this solution causes undefined behaviour. That case is just `return y`, though, so it's easy to special-case. Alternately, use `1ULL` (or however you want to get a 64-bit type) instead of `1U`. – Carl Norum Jun 27 '13 at 04:43
+1 for not just giving the code but also explaining what it does step-by-step – legends2k Jun 27 '13 at 04:58
Carl, your comment should properly be part of your answer. – Jim Balter Jun 27 '13 at 05:33
It's more of an aside than part of the answer, really, which is why I left it a comment. NBD. – Carl Norum Jun 27 '13 at 16:50

Optimising bitwise operations in C

1 Answers1