#modbus #server #hardware #nostd


Fast and platform-independent Modbus server framework

21 unstable releases (3 breaking)

new 0.4.0 Sep 27, 2020
0.3.9 Sep 19, 2020
0.2.2 Sep 13, 2020
0.1.7 Sep 13, 2020

#25 in Hardware support

Download history 91/week @ 2020-09-08 157/week @ 2020-09-15

90 downloads per month


2.5K SLoC

rmodbus - Modbus for Rust

A framework to build fast and reliable Modbus-powered applications.

Cargo crate: https://crates.io/crates/rmodbus

What is rmodbus

rmodbus is not a yet another Modbus server. rmodbus is a set of tools to quickly build Modbus-powered applications.

Why yet another Modbus lib?

  • rmodbus is transport and protocol independent

  • rmodbus is platform independent (no_std is fully supported!)

  • can be easily used in blocking and async (non-blocking) applications

  • tuned for speed and reliability

  • provides a set of tools to easily work with Modbus context

  • supports server frame processing for Modbus TCP/UDP, RTU and ASCII.

  • server context can be easily, managed, imported and exported

So the server isn't included?

Yes, there's no server included. You build the server by your own. You choose protocol, technology and everything else. rmodbus just process frames and works with Modbus context.

Here's an example of a simple TCP blocking server:

use std::io::{Read, Write};
use std::net::TcpListener;
use std::thread;

use lazy_static::lazy_static;

use std::sync::RwLock;

use rmodbus::server::{context::ModbusContext, ModbusFrame, ModbusFrameBuf, ModbusProto};

lazy_static! {
    pub static ref CONTEXT: RwLock<ModbusContext> = RwLock::new(ModbusContext::new());

pub fn tcpserver(unit: u8, listen: &str) {
    let listener = TcpListener::bind(listen).unwrap();
    println!("listening started, ready to accept");
    for stream in listener.incoming() {
        thread::spawn(move || {
            println!("client connected");
            let mut stream = stream.unwrap();
            loop {
                let mut buf: ModbusFrameBuf = [0; 256];
                let mut response = Vec::new(); // for nostd use FixedVec with alloc [u8;256]
                if stream.read(&mut buf).unwrap_or(0) == 0 {
                let mut frame = ModbusFrame::new(unit, &buf, ModbusProto::TcpUdp, &mut response);
                if frame.parse().is_err() {
                    println!("server error");
                if frame.processing_required {
                    let result = match frame.readonly {
                        true => frame.process_read(&CONTEXT.read().unwrap()),
                        false => frame.process_write(&mut CONTEXT.write().unwrap()),
                    if result.is_err() {
                        println!("frame processing error");
                if frame.response_required {
                    println!("{:x?}", response.as_slice());
                    if stream.write(response.as_slice()).is_err() {

There are also examples for Serial-RTU, Serial-ASCII and UDP in examples folder (if you're reading this text somewhere else, visit rmodbus project repository.

Running examples:

cargo run --example app --features std
cargo run --example tcpserver --features std

Modbus context

The rule is simple: one standard Modbus context per application. 10k+10k 16-bit registers and 10k+10k coils are usually more than enough. This takes about 43Kbytes of RAM, but if you need to reduce context size, download library source and change CONTEXT_SIZE constant in "context.rs".

rmodbus server context is thread-safe, easy to use and has a lot of functions.

The context is created automatically, as soon as the library is imported. No additional action is required.

Every time Modbus context is accessed, a context mutex must be locked. This slows down a performance, but guarantees that the context always has valid data after bulk-sets or after 32-bit data types were written. So make sure your application locks context only when required and only for a short period time.

There are two groups of context functions:

  • High-level API: simple functions like coil_get automatically lock the context but do this every time when called. Use this for testing or if the speed is not important.

  • Advanced way is to use low-level API, lock the context manually and then call proper functions, like set, set_f32 etc.

Take a look at simple PLC example:

use std::fs::File;
use std::io::Write;

use lazy_static::lazy_static;

use std::sync::RwLock;

use rmodbus::server::context::ModbusContext;

lazy_static! {
    pub static ref CONTEXT: RwLock<ModbusContext> = RwLock::new(ModbusContext::new());

fn looping() {
    println!("Loop started");
    loop {
        let ctx = CONTEXT.read().unwrap();
        let _param1 = ctx.get_holding(1000).unwrap();
        let _param2 = ctx.get_holdings_as_f32(1100).unwrap(); // ieee754 f32
        let _param3 = ctx.get_holdings_as_u32(1200).unwrap(); // u32
        let cmd = ctx.get_holding(1500).unwrap();
        if cmd != 0 {
            println!("got command code {}", cmd);
            let mut ctx = CONTEXT.write().unwrap();
            ctx.set_holding(1500, 0).unwrap();
            match cmd {
                1 => {
                    println!("saving memory context");
                    let _ = save("/tmp/plc1.dat", &ctx).map_err(|_| {
                        eprintln!("unable to save context!");
                _ => println!("command not implemented"),
        // ==============================================
        // DO SOME JOB
        // ..........
        // WRITE RESULTS
        let mut ctx = CONTEXT.write().unwrap();
        ctx.set_coil(0, true).unwrap();
        ctx.set_holdings_bulk(10, &(vec![10, 20])).unwrap();
        ctx.set_inputs_from_f32(20, 935.77).unwrap();

fn save(fname: &str, ctx: &ModbusContext) -> Result<(), std::io::Error> {
    let mut file = match File::create(fname) {
        Ok(v) => v,
        Err(e) => return Err(e),
    for i in ctx.iter() {
        match file.write(&[i]) {
            Ok(_) => {}
            Err(e) => return Err(e),
    match file.sync_all() {
        Ok(_) => {}
        Err(e) => return Err(e),
    return Ok(());

To let the above program communicate with outer world, Modbus server must be up and running in the separate thread, asynchronously or whatever is preferred.


rmodbus supports no_std mode. Most of the library code is written the way to support both std and no_std.

Set dependency as:

rmodbus = { version = "*", features = ["nostd"] }

Small context

Default Modbus context has 10000 registers of each type, which requires 42500 bytes total. For systems with small RAM amount it's possible to reduce the context size to the 1000 registers of each type (4250 bytes) with the following feature:

rmodbus = { version = "*", features = ["nostd", "smallcontext"] }


Some of rmodbus functions use vectors to store result. In std mode, either standard std::vec::Vec or FixedVec can be used. In nostd mode, only FixedVec is supported.



  • Modbus context is no longer created automatically and no mutex guard is provided by default. Use ModbusContext::new() to create context object and then use it as you wish - protect with any kind of Mutex, with RwLock or just put into UnsafeCell.

  • Context SDK changes: all functions moved inside context, removed unnecessary ones, function args optimized.

  • FixedVec support included by default, both in std and nostd.

  • Added support of 64-bit integers

Modbus client