probe-run

Runs embedded programs just like native ones

NOTE: Upstream probe-run 0.3.0 now supports the RP2040! This fork is no longer required - just cargo install probe-run.

probe-run is a custom Cargo runner that transparently runs Rust firmware on a remote device.

probe-run is powered by probe-rs and thus supports as many devices and probes as probe-rs does.

Features

• Acts as a Cargo runner, integrating into cargo run.
• Displays program output streamed from the device via RTT.
• Exits the firmware and prints a stack backtrace on breakpoints.

Installation

To install probe-run, use cargo install probe-run.

$ cargo install probe-run-rp --git https://github.com/rp-rs/probe-run

On Linux, you might have to install libudev and libusb from your package manager before installing probe-run.

# ubuntu
$ sudo apt install -y libusb-1.0-0-dev libudev-dev

# fedora
$ sudo dnf install -y libusbx-devel systemd-devel

Setup

1. Set the Cargo runner

The recommend way to use probe-run is to set as the Cargo runner of your application.

Add these two lines to your Cargo configuration file (.cargo/config.toml) and set the particular --chip value for your target. In this case it is nRF52840_xxAA for the nRF52840:

[target.'cfg(all(target_arch = "arm", target_os = "none"))']
runner = "probe-run --chip nRF52840_xxAA"
#                          ^^^^^^^^^^^^^

To list all supported chips run probe-run --list-chips.

1.1 Env variable

To support multiple devices, or permit overriding default behavior, you may prefer to:

1. set the ${PROBE_RUN_CHIP} environment variable, and
2. set runner (or CARGO_TARGET_${TARGET_ARCH}_RUNNER) to probe-run:

[target.'cfg(all(target_arch = "arm", target_os = "none"))']
runner = "probe-run"

1.2 Multiple probes

If you have several probes connected, you can specify which one to use by adding the --probe option to the runner or setting the ${PROBE_RUN_PROBE} environment variable with a value containing either ${VID}:${PID} or ${VID}:${PID}:${SERIAL}:

// --probe
$ probe-run --probe '0483:3748' --chip ${PROBE_RUN_CHIP}

// PROBE_RUN_PROBE
$ PROBE_RUN_PROBE='1366:0101:123456' cargo run

To list all connected probes, run probe-run --list-probes.

2. Enable debug info

Next check that debug info is enabled for all profiles. If you are using the cortex-m-quickstart template then this is already the case. If not check or add these lines to Cargo.toml.

[dependencies]
...
panic-probe = { version = "0.2", features = ["print-rtt"] }

# Cargo.toml
[profile.dev]
debug = 1 # default is `true`; not needed if not already overridden

[profile.release]
debug = 1 # default is `false`; using `true` is also OK as symbols reside on the host, not the target

3. Look out for old dependencies

The cortex-m dependency must be version 0.6.3 or newer. Older versions are not supported. Check your Cargo.lock for old versions. Run cargo update to update the cortex-m dependency if an older one appears in Cargo.lock.

4. Run

You are all set. You can now run your firmware using cargo run. For example,

use cortex_m::asm;
use cortex_m_rt::entry;
use panic_probe as _;
use rtt_target::rprintln;

#[entry]
fn main() -> ! {
    rtt_init_print!(); // You may prefer to initialize another way    
    rprintln!("Hello, world!");
    loop { asm::bkpt() }
}

would output

$ cargo run --bin hello
    Finished dev [unoptimized + debuginfo] target(s) in 0.07s
     Running `probe-run --chip nRF52840_xxAA target/thumbv7em-none-eabihf/debug/hello`
  (HOST) INFO  flashing program (30.22 KiB)
  (HOST) INFO  success!
────────────────────────────────────────────────────────────────────────────────
INFO:hello -- Hello, world!
────────────────────────────────────────────────────────────────────────────────
  (HOST) INFO  exiting because the device halted.
To see the backtrace at the exit point repeat this run with
`probe-run --chip nRF52840_xxAA target/thumbv7em-none-eabihf/debug/hello --force-backtrace`

Stack backtraces

When the device raises a hard fault exception, indicating e.g. a panic or a stack overflow, probe-run will print a backtrace and exit with a non-zero exit code.

This backtrace follows the format of the std backtraces you get from std::panic! but includes <exception entry> lines to indicate where an exception/interrupt occurred.

#![no_main]
#![no_std]

use cortex_m::asm;
#[entry]
fn main() -> ! {
    // trigger a hard fault exception with the UDF instruction.
    asm::udf()
}

    Finished dev [optimized + debuginfo] target(s) in 0.04s
     Running `probe-run --chip nRF52840_xxAA target/thumbv7em-none-eabihf/debug/hard-fault`
  (HOST) INFO  flashing program (30.08 KiB)
  (HOST) INFO  success!
────────────────────────────────────────────────────────────────────────────────
stack backtrace:
   0: HardFaultTrampoline
      <exception entry>
   1: __udf
   2: cortex_m::asm::udf
        at /<...>/cortex-m-0.6.4/src/asm.rs:104
   3: panic::__cortex_m_rt_main
        at src/bin/hard-fault.rs:12
   4: main
        at src/bin/hard-fault.rs:8
   5: ResetTrampoline
        at /<...>3/cortex-m-rt-0.6.13/src/lib.rs:547
   6: Reset
        at /<...>/cortex-m-rt-0.6.13/src/lib.rs:550

If we look at the return code emitted by this cargo run, we'll see that it is non-0:

$ echo $?
134

⚠️ NOTE when you run your application with probe-run, the HardFault handler (default or user-defined) will NOT be executed.

Backtrace options

--backtrace

The --backtrace flag is optional and can get passed the following values:

• --backtrace=always - forced backtrace (if you'd like to see a backtrace at the end of successful program run)
• --backtrace=never - suppresed backtrace
• --backtrace=auto - default, shows a backtrace if the program panics or the stack overflows

Run it like this (example for a forced backtrace):

$ cargo run --bin hello --backtrace=always

--backtrace-limit

The --backtrace-limit flag is optional and defaults to 50. It is possible to set any number.

--backtrace-limit=0 is accepted and means "no limit".

To show a shortened backtrace showing 5 frames, run:

$ cargo run --bin panic --backtrace-limit=5

Note: if --backtrace=never is set, setting --backtrace-limit has no effect.

Troubleshooting

probe-run --list-probes says "No devices were found."

Apart from a faulty connection between your computer and the target device, this could be caused by several things:

[Linux only] udev rules haven't been set

In order for probe-run to find the device you'd like to run your code on, your system needs permission to access the device as a non-root user.

In order to grant these permissions, you'll need to add a new set of udev rules.

To learn how to do this for the nRF52840 Development Kit, check out the installation instructions in our embedded training materials.

No external or on-board debugger present

To use probe-run you need a "probe" (also known as "debugger") that sits between your PC and the microcontroller.

Most development boards, especially the bigger ones, have a probe "on-board": If the product description of your board mentions something like a J-Link or ST-Link on-board debugger you're good to go. With these boards, all you need to do is connect your PC to the dev board using a USB cable you are all set to use probe-run!

If this is not the case for your board, check in the datasheet if it exposes exposes SWD or JTAG pins. If they are exposed, you can connect a "stand alone" probe device to the microcontroller and then connect the probe to your PC via USB. Some examples of stand alone probes are: the ST-Link and the J-Link.

Note that this may involve some soldering if your board does not come with a pre-attached header to plug your debugger into.

Error: RTT up channel 0 not found

This may instead present as Error: RTT control block not found in target memory.

Your code, or a library you're using (e.g. RTIC) might be putting your CPU to sleep when idle. You can verify that this is the problem by busy looping instead of sleeping. When using RTIC, this can be achieved by adding an idle handler to your app:

#[idle]
fn idle(_ctx: idle::Context) -> ! {
     loop {}
}

Assuming you'd like to still sleep in order to save power, you need to configure your microcontroller so that RTT can still be handled even when the CPU is sleeping. How to do this varies between microcontrollers.

On an STM32G0 running RTIC it can be done by amending your init function to set the dmaen bit on RCC.ahbenr. e.g.:

#[init]
fn init(ctx: init::Context) -> init::LateResources {
     ctx.device.RCC.ahbenr.write(|w| w.dmaen().set_bit());
     ...
}

defmt version mismatch

end-user

Follow the instructions in the error message to resolve the mismatch.

developer

If you are hacking around with probe-run, you can disable the version check by setting the PROBE_RUN_IGNORE_VERSION environment variable to true or 1 at runtime.

Developer Information

running your locally modified probe-run

For easier copy-paste-ability, here's an example how to try out your local probe_run modifications.

$ cd probe-run/
$ PROBE_RUN_IGNORE_VERSION=1 cargo run -- --chip nRF52840_xxAA --backtrace-limit=10 hello
  ˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆ                                   ˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆˆ ˆˆˆˆˆ
  environment variables                                        extra flags             binary to be
  (optional)                                                   (optional)              flashed & run

running snapshot tests

To check whether your change has altered probe-run in unexpected ways, please run the snapshot tests in tests before opening a PR if at all possible.

You can do so by connecting a nrf52840 Development Kit and running

$ cargo test -- --ignored

Support Us

probe-run is part of the Knurling project, Ferrous Systems' effort at improving tooling used to develop for embedded systems.

If you think that our work is useful, consider sponsoring it via GitHub Sponsors.

License

Licensed under either of

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)

• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be licensed as above, without any additional terms or conditions.