🅃🅁ace 🄸d’s 🄲 🄴mbedded

github.io/trice/

Log in (a) trice (S>G)

even inside ↯ interrupts in less than 1 µs❗

About

• Replace printf or log in C-code for getting:
  • Speed, to be usable also inside interrupts,
  • Space, to reduce needed FLASH memory size,
  • Features delighting the developers heart.
• Main idea: Logging strings not into an embedded device to display them later on a PC but keep usage comfortable and simple.
• Trice consists of 2 parts:
  1. C code trice macros, similar usable like printf, generating tiny & super-fast embedded device real-time trace/log code.
  2. Tool trice for managing and visualization.
     • Written in Go and therefore usable on all platforms Go supports.
     • You can also use your own environment to receive the Trice packages, exchange the carried IDs with the format string and print out.
• Trice User guide

Possible Use Cases

• Using Trice not only for printf debugging but also as logging technique is possible and gives the advantage to have very short messages (no strings) for transmission, but keep in mind that the file til.json is the key to read all output if your devices in the field for 10 or more years.
  • Optionally add til.json as a (compressed) resource to your target image. One possibility is using SRecord. Or simply provide a download link.
• You can see Trice also as a kind of data compression what could be interesting for IoT things, especially NB-IoT, where you have very low data rates.
• Storing Trice messages in FLASH memory for later log analysis saves memory because a typical TRICE occupies only 4 bytes independently of the format string length plus optional values.
• Also, it is possible to encrypt the Trice transfer packets to get a reasonable protection for many cases.
  • This way you can deliver firmware images with encrypted Trice output, only readable with the appropriate key and til.json.
  • XTEA is implemented as one option.
• You can even translate the til.json file in different languages, so changing a language is just changing the til.json file without touching the target binary.
• With Trice it is easy to do timing analysis on distributed embedded systems. Host and target timestamps are supported.

How it approximately works (UART example)

This slightly simplified view is explained here

Data Transfer

• Implemented:
  • UART connectable to virtual UART over USB
  • RTT over J-Link and RTT over ST-Link
• A small separate microcontroller is always usable as interfaces bridge to GPIO, I²C, SPI, CAN, LIN, …
• With a Chip from FTDI, check for example Adafruit FT232H Breakout, GPIO, I2C and SPI are easy accessable.

  Display server option

Start trice ds inside a console, option: third_party/alacritty, locally or on a remote PC and connect with several trice tool instances like with trice log -p COM15 -ds for example.

Documentation

• https://interrupt.memfault.com/blog/trice
• Trice User Guide
• Check the docs folder. No need to read all this stuff - it is just for help and reference.
• Check issues and discussions including the closed items.

Debugging using VS-Code and Clang for a Trice-instrumented Project in Direct-Out Mode over SEGGER-RTT

(See folder ./examples for more details.)

Trice Cache

Starting with Trice version 0.70.0 it is possible to use additionally the Trice -cache CLI switch for the commands trice insert and trice clean. This switch will have only effect when the user creates in his home directory the .trice/cache folder. The Trice cache is considered as experimental, even it it thoroughly tested. (Trice Cache Spec)

When to use it

When you use trice i as a pre-compile step and trice c as a post-compile step to have the IDs not in the project source code when you work on it, therefore only during compilation time, and wish to speed-up the whole thing.

How it works

The Trice cache keeps copies of all to trice i or trice c passed files after processing them, to avoid repeatedly ID inserting and cleaning. The copies are used to get the same results as with trice i or trice c for all files not edited inbetween. Edited files are processed normally and the cache is updated afterwards. Because the file copies are done without changing the file modification time, a build system does not process unchanged files again even the IDs have been temorarily removed from the files.

Hint

Special care is needed, when the build system modifies source files as well!

For example an auto-formatter should get active before the tice insert command.

Wich mode to use?

• For the development direct mode with SEGGER_RTT is recommended.
• Most use cases are coverable in deferred mode with TRICE_BUFFER == TRICE_RING_BUFFER (less RAM needs) in TRICE_MULTI_PACK_MODE (less transfer data).

Project Status

Trice is full usable and there are no known bugs (see issues).

Future

The documentation could get improved, for example by adding a quick start help. Additional features like remote procedure calls could be described and a separate tlog, maybe written in C, would allow logging on any platforms - not only Go supported ones.

Support?

Yes please - or simply :star: it. ☺

Become a Sponsor with your Github Account

OR

OR

Cloning the repo:

git clone https://github.com/rokath/trice.git

Similar projects

• baical.net (C)
• call stack logger function instrumentation (a way to trace programs flow of execution)
• Debugging with Dynamic Printf Breakpoints (Eclipse IDE option)
• defmt (Rust)
• Diagnostic Log and Trace (AUTOSAR)
• elog (Embedded logger with minimal footprint and memory usage)
• J-Link System View (SEGGER)
• Logging with symbols - The Embedonomicon
• MCUViewer
• Memfault Compact Log Library
• Metal Serial Library
• Minimal Structured Logging for Autonomous Vehikles (C++, closed source talk)
• NanoLog (Linux C++)
• Percepio Tracealyzer (Visual Trace Diagnostics)
• Pigweed Trace Tokenized
• Postform (Postponed formatting experiments with string interning in C++. Inspired by the defmt Rust crate )
• qpspy (C/C++)
• Serial-Studio (Data visualisation)
• Traces (API tracing framework for Linux C/C++ applications
• Zepyr Dictionary Based Logging