Portable way of bit masking in C (or C++)

[abhishekm71]

In my book, I have an exercise that asks me to take an int x (like 0x12345678) and return another int with all but the least significant byte to be changed to FF (like 0xFFFFFF78).

So far so good. I implemented it as follows and it works:

#include <stdio.h>

int main()
{
    int x = 0x12345678;
    printf( " %.2x\n", x | 0xFFFFFF00 );
    return 0;
}

The catch here is that it says your code should not assume any 'word size'.

Although our examples assume a 32-bit word size, your code should work for any word size >= 8

How can I incorporate this 'portability' issue?

Thanks!

[helios]

x | ~0xFF

By the way, you should be careful when doing bit twiddling with signed values, and do it with unsigned values whenever possible.

[abhishekm71]

Aah Thanks helios

Was struggling with this!

And thanks for the suggestion on signed vs unsigned values as well!

[Catfish666]

A somewhat ugly solution would be casting the int to char* then use a for() loop, or std::fill(), or std::fill_n() if you want to be fancy.

Have you learned about endianness yet?
http://en.wikipedia.org/wiki/Endianness

http://www.cplusplus.com/reference/algorithm/fill/
http://www.cplusplus.com/reference/algorithm/fill_n/

int x = 0x12345678;
char *tp = reinterpret_cast<char *> (x);

if (/* Little-Endian machine */)
{
    std::fill_n(tp + 1, sizeof x - 1, 0xFF);
}
else /* Big-Endian machine */
{
    std::fill_n(tp, sizeof x - 1, 0xFF);
}

Disclaimer: code above was not tested.