minimult_cortex-m

Crates.iominimult_cortex-m
lib.rsminimult_cortex-m
version0.3.3
sourcesrc
created_at2019-10-20 05:04:24.395751
updated_at2019-11-20 12:38:42.87547
descriptionA minimal multitask library, or RTOS, for Cortex-M microcontrollers.
homepagehttps://github.com/convexbrain/Minimult/tree/master/minimult_cortex-m
repositoryhttps://github.com/convexbrain/Minimult
max_upload_size
id174136
size81,441
convexbrain (convexbrain)

documentation

README

minimult_cortex-m

This crate for Rust provides a minimal multitask library Minimult for Cortex-M microcontrollers.

Target

Single-core systems of

  • Cortex-M0 / M0+ / M1 (thumbv6m-none-eabi)
  • Cortex-M3 (thumbv7m-none-eabi)
  • Cortex-M4 / M7 (thumbv7em-none-eabi) with FPU (thumbv7em-none-eabihf)
  • Cortex-M23 (thumbv8m.base-none-eabi)
  • Cortex-M33 / M35P (thumbv8m.main-none-eabi) with FPU (thumbv8m.main-none-eabihf)

Features

  • Task like that of a typical RTOS
    • Minimult can take closures and register them as tasks.
    • Minimult runs into a loop to start dispatching those tasks.
      • Not supported: dynamically creating and spawning.
  • Synchronization
    • idle and kick
      • A task goes into an idle state and other tasks/interrupts wake it up by kicking.
    • MTMsgSender and MTMsgReceiver
      • Task-to-task communication by message passing.
    • MTSharedCh
      • Shared variable among tasks.
  • Priority-based dispatching
    • A higher priority task preempts lower priority tasks.
    • Round-robin dispatching within the same priority tasks.
    • dispatch can be directly requested so that timer-based preemption is also possible.
  • Static memory allocation
    • Minimult doesn't require a global allocator but reserves a bunch of memory block in advance.

Examples

Usage

// Runnable on QEMU ARM

#![no_main]
#![no_std]

use cortex_m::Peripherals;
use cortex_m_rt::entry;
use cortex_m_rt::exception;
use cortex_m_semihosting::debug;
use cortex_m_semihosting::hprintln;
use panic_semihosting as _;

use minimult_cortex_m::*;

#[entry]
fn main() -> !
{
    let mut mem = Minimult::mem::<[u8; 4096]>();
    let mut mt = Minimult::new(&mut mem, 3);

    let mut q = mt.msgq::<u32>(4);
    let (snd, rcv) = q.ch();

    let sh = mt.share::<u32>(0);
    let shch1 = sh.ch();
    let shch2 = sh.ch();

    mt.register(0/*tid*/, 1, 256, || task0(snd));
    mt.register(1/*tid*/, 1, 256, || task1(rcv, shch1));
    mt.register(2/*tid*/, 1, 256, || task2(shch2));

    // SysTick settings
    let cmperi = Peripherals::take().unwrap();
    let mut syst = cmperi.SYST;
    syst.set_clock_source(cortex_m::peripheral::syst::SystClkSource::Core);
    syst.set_reload(1_000_000);
    syst.clear_current();
    syst.enable_counter();
    syst.enable_interrupt();

    // must be error in terms of lifetime and ownership
    //drop(mem);
    //drop(q);
    //drop(snd);
    //drop(rcv);
    //drop(sh);
    //drop(shch1);
    //drop(shch2);

    hprintln!("Minimult run").unwrap();
    mt.run()
}

#[exception]
fn SysTick()
{
    Minimult::kick(0/*tid*/);
}

fn task0(mut snd: MTMsgSender<u32>)
{
    for vsnd in 0.. {
        Minimult::idle();

        hprintln!("task0 send {}", vsnd).unwrap();
        snd.send(vsnd);
    }
}

fn task1(mut rcv: MTMsgReceiver<u32>, shch: MTSharedCh<u32>)
{
    for i in 0.. {
        let vrcv = rcv.receive();

        assert_eq!(i, vrcv);
        hprintln!("task1 touch {}", vrcv).unwrap();
        let mut vtouch = shch.touch();
        *vtouch = vrcv;
    }
}

fn task2(shch: MTSharedCh<u32>)
{
    let mut j = 0;

    while j < 50 {
        let vlook = shch.look();

        assert!(j <= *vlook);
        //hprintln!("task2 look {}", *vlook).unwrap(); // many lines printed
        j = *vlook;
    }

    hprintln!("task2 exit").unwrap();
    debug::exit(debug::EXIT_SUCCESS);
}

Other Examples

You can find a specific board's example here. Currently there are very few examples, however.

Commit count: 80

cargo fmt