kernel-services

Zephyr Kernel Services

Move beyond basic threading and logging to build modular, event-driven, and robust Zephyr applications.

Core Workflows

1. Event-Driven Communication (Zbus)

Decouple your modules using a lightweight publish-and-subscribe bus.

• Reference: zbus.md
• Key Tools: ZBUS_CHAN_DEFINE, ZBUS_SUBSCRIBER_DEFINE, zbus_chan_pub.

2. Behavioral Logic (SMF)

Manage complex system states and transitions using the State Machine Framework.

• Reference: smf.md
• Key Tools: smf_set_state, SMF_CREATE_STATE, Hierarchical states.

3. Background Processing (Work Queues)

Defer long-running or non-critical tasks to prevent blocking interrupts or high-priority threads.

• Reference: settings_workqueue.md
• Key Tools: k_work_submit, k_work_delayable, Custom work queues.

4. Persistence (Settings)

Save and restore configuration data and state across reboots.

• Reference: settings_workqueue.md
• Key Tools: settings_load, settings_save_one, NVS backends.

Quick Start (Zbus)

// Define a channel for sensor data
ZBUS_CHAN_DEFINE(sensor_data_chan, struct sensor_msg, NULL, NULL, ZBUS_OBSERVERS_EMPTY, ZBUS_CHAN_DEFAULTS);

// Publish from a thread
zbus_chan_pub(&sensor_data_chan, &msg, K_NO_WAIT);

Professional Patterns (Asset Tracker Style)

• Modularity: Use Zbus as the backbone for inter-module communication.
• Predictability: Use SMF to define clear lifecycle states for each module (e.g., Uninitialized -> Ready -> Active -> Error).
• Responsiveness: Use custom work queues for sensor data ingestion to keep the main thread responsive for cloud communication.
• Sensor Integration: For sensor data ingestion patterns, see the hardware-io skill.

Automation Tools

• zbus_channel_lint.py: Detect duplicate ZBUS_CHAN_DEFINE names across source files.

Examples & Templates

• smf_state_table_template.c: Starter SMF state table and lifecycle wiring.

Validation Checklist

• Zbus publishers and subscribers exchange messages without deadlock or missed updates.
• SMF transitions follow expected state graph under normal and error conditions.
• Deferred work executes on intended queue context with bounded execution time.
• Settings values persist across reboot and reload successfully at startup.

Resources

• References:
  • zbus.md: Publish/Subscribe patterns and subscriber types.
  • smf.md: Finite and Hierarchical state machine implementation.
  • settings_workqueue.md: Background work and persistent storage.
• Scripts:
  • zbus_channel_lint.py: Zbus channel name collision checker.
• Assets:
  • smf_state_table_template.c: State-machine template.