Should I use packed structure to simulate memory?

[cahuila]

I am trying to simulate memory in C. I want to create a structure in C that will hold a 8bit opcode and a 32bit memory address. This would simulate a 40 bit instruction for my simulator.

I read in an old book that packed structs could be used for this to not waste space. What are the draw backs for using this? I am not worried about wasting space, but just looking for a simple way to access the memory instructions. Below is a sample of the structure that I want to use.

struct memory_area {
unsigned int opcode:8;
unsigned int address:32;
};

[weaknessforcats]

You also might try:

struct memory_area
{
    unsigned char opcode;
    unsigned char address[4];
};

In this case the 40 bit instruction is opcode for the 8-bit opcode and address is the 32 bit address.


Note that there are no character types in C or C++. A char is just an 8-bit integer. The sizeof this struct using my compiler is 5 bytes, or 40 bits.

Personally, I don't like using bitfields as being very archaic.

[donbock]

Do all instructions in your simulated processor use exactly 40 bits? If not, then your simulated memory will have to support variable sized instructions.

Yet another way to implement the 40-bit structure is with specific-size types:

struct memory_word
{
    uint8_t opcode;
    uint32_t address;
};
struct memory_word memory_area[1000];