Export cycle_internal for single stepping

[donno2048]

Using a cycles function approach was suggested in #103 as a single-stepping solution.

The exported cycle was removed however.

I think it could be useful to export such a function, maybe the cycle_internal should be having some wrapping before it can be shipped, but I think it'd beneficial overall.

[copy]

I think exposing cycle_internal is sub-optimal for two reasons:

• It doesn't really single step, e.g. the jitted code will usually execute more than one instruction
• It's a big function that is called often, so exporting it may negatively affect inlining decisions and/or code size

What you really want is probably something like this:

#[no_mangle]
unsafe fn single_step() {
    *previous_ip = *instruction_pointer;
    let opcode = match read_imm8() {
        Ok(x) => x,
        Err(()) => return,
    };
    run_instruction(opcode | (*is_32 as i32) << 8);
    *instruction_counter += 1;
}

Note that this may still execute more than one instruction in a few cases, e.g. after STI or REP prefixes (I wouldn't mind a flag to disable that, if you really need it).

[donno2048]

I think exposing cycle_internal is sub-optimal for two reasons:

• It doesn't really single step, e.g. the jitted code will usually execute more than one instruction

I guess, my use-case was more "clocking" over single-stepping so it didn't matter, but in general, it may not be a good decision...

[copy]

I guess, my use-case was more "clocking" over single-stepping so it didn't matter, but in general, it may not be a good decision...

What is the goal here? Clocking may be interesting to have directly in v86.

[donno2048]

Just to adjust the clock speed of the "cpu" to fit better the clock speed required by the program

[copy]

Just to adjust the clock speed of the "cpu" to fit better the clock speed required by the program

That sounds useful to other people, I'd suggest to upstream that feature directly rather than exposing cycle_internal. Let's close this PR.