[][src]Crate panic_semihosting

Report panic messages to the host stderr using semihosting

This crate contains an implementation of panic_fmt that logs panic messages to the host stderr using cortex-m-semihosting. Before logging the message the panic handler disables (masks) the device specific interrupts. After logging the message the panic handler trigger a breakpoint and then goes into an infinite loop.

Currently, this crate only supports the ARM Cortex-M architecture.


This example is not tested

extern crate panic_semihosting;

fn main() {
(gdb) monitor arm semihosting enable
(gdb) continue
Program received signal SIGTRAP, Trace/breakpoint trap.
rust_begin_unwind (args=..., file=..., line=8, col=5)
    at $CRATE/src/lib.rs:69
69          asm::bkpt();
$ openocd -f (..)
panicked at 'FOO', src/main.rs:6:5

Optional features


When this feature is enabled the panic handler performs an exit semihosting call after logging the panic message. This is useful when emulating the program on QEMU as it causes the QEMU process to exit with a non-zero exit code; thus it can be used to implement Cortex-M tests that run on the host.

We discourage using this feature when the program will run on hardware as the exit call can leave the hardware debugger in an inconsistent state.


When this feature is enabled semihosting is implemented using inline assembly (asm!) and compiling this crate requires nightly.

When this feature is disabled semihosting is implemented using FFI calls into an external assembly file and compiling this crate works on stable and beta.