DWORD variable with low/high word and low/high byte

Question

How in C can we read and make DWORD variables with a low and high word and low and high byte?

Solution 1

In Win32 a DWORD is a 32 bit unsigned integer. In other contexts it could possibly mean something else.

Assuminng the Win32 definition (and other Win32 typedefs):

BYTE lsb = 0x11 :
BYTE next_lsb = 0x22 :
BYTE next_msb = 0x33 :
BYTE msb = 0x44 :

DWORD dword_from_bytes = (msb << 24) | (next_msb << 16) | (next_lsb << 8) | lsb ;

dword_from_bytes will have the value 0x44332211.

Similarly:

WORD lsw = 0x1111 :
WORD msw = 0x2222 :

DWORD dword_from_words = (msw << 16) | lsw ;

dword_from_words will have the value 0x22221111.

To extract say the third byte from dword_from_bytes for example:

next_msb = (dword_from_bytes >> 16) & 0xff ;

although the & 0xff is not strictly necessary in this case given the type of next_msb, but if the type of the receiver were larger than 8 bits, it will mask off the msb bits.

Solution 2

WinAPI provides macros for the manipulations of these types, such as:

• HIWORD
• LOWORD
• MAKELPARAM

Solution 3

#include <stdint.h>
#include <stdio.h>

typedef union _little_endian{
    struct _word{
        union _msw{
            struct _msw_byte{
                uint8_t MSB;
                uint8_t LSB;
            } __attribute__((__packed__)) MSW_BYTE;
            uint16_t WORD;
        } MSW;
        union _lsw{
            struct _lsw_byte{
                uint8_t MSB;
                uint8_t LSB;
            } __attribute__((__packed__)) LSW_BYTE;
            uint16_t WORD;
        } LSW;
    } __attribute__((__packed__)) WORD;
    uint32_t DWORD;
} DWORD;

int main(int argc, char *argv[]){
    DWORD test1;
    test1.WORD.MSW.MSW_BYTE.MSB = 1;
    test1.WORD.MSW.MSW_BYTE.LSB = 2;
    test1.WORD.LSW.LSW_BYTE.MSB = 3;
    test1.WORD.LSW.LSW_BYTE.LSB = 4;
    printf("test1: hex=%x uint=%u\n", test1.DWORD, test1.DWORD);
    
    DWORD test2;
    test2.DWORD = 0x08080404;
    printf("test2: hex=%x uint=%u\n", test2.DWORD, test2.DWORD);
    printf("test2.WORD.MSW.MSW_BYTE.MSB: uint=%u\n", test2.WORD.MSW.MSW_BYTE.MSB);
    printf("test2.WORD.MSW.MSW_BYTE.LSB: uint=%u\n", test2.WORD.MSW.MSW_BYTE.LSB);
    printf("test2.WORD.LSW.LSW_BYTE.MSB: uint=%u\n", test2.WORD.LSW.LSW_BYTE.MSB);
    printf("test2.WORD.LSW.LSW_BYTE.LSB: uint=%u\n", test2.WORD.LSW.LSW_BYTE.LSB);
    
    return 0;
}

I prefer to use a combination of structs and unions.

Output:

test1: hex=4030201 uint=67305985
test2: hex=8080404 uint=134743044
test2.WORD.MSW.MSW_BYTE.MSB: uint=4
test2.WORD.MSW.MSW_BYTE.LSB: uint=4
test2.WORD.LSW.LSW_BYTE.MSB: uint=8
test2.WORD.LSW.LSW_BYTE.LSB: uint=8

Comments

[Midas]

Now it starts to make sense. But what does the <<, >> and | actually mean? Thanks!

[valdo]

It's better to use the dedicated macros, such as HIWORD, LOWORD, MAKELONG and etc. Because different processors use different byte ordering (little/big endian and etc.)

[Midas]

Well, never mind. I have read the docs about bitwise operators and I understand how it works now!

[Clifford]

@valdo: While I agree that the API defined macros should be used, byte ordering is not the issue; (msw << 16) for example will always place a value in the high-order word regardless of endian-ness. There are other macros for exchanging data between machines of differing endian-ness. What they do do is encapsulate the knowledge of API defined word sizes to ensure consistency and avoid errors.

[Hanan N.]

in addition you have the HIBYTE and LOBYTE respectively

[Laurie Stearn]

MS = Most Significant, LS = Least Significant. Thanks. :)