mirror of
https://github.com/apache/nuttx.git
synced 2026-08-19 04:28:21 +00:00
ab77fb3467
Add documentation for the stackmonitor command, which starts a background daemon to monitor stack usage for all tasks and threads. Refs #11081 Signed-off-by: hanzj <your-email@example.com>
110 lines
2.8 KiB
ReStructuredText
110 lines
2.8 KiB
ReStructuredText
==============================
|
|
``stackmonitor`` Stack Monitor
|
|
==============================
|
|
|
|
Overview
|
|
========
|
|
|
|
The ``stackmonitor`` command starts a background daemon that periodically
|
|
monitors stack usage for all tasks and threads in the system. It reads the
|
|
procfs filesystem to collect stack size and usage information, helping
|
|
developers identify potential stack overflow issues and optimize memory
|
|
allocation.
|
|
|
|
The daemon runs as a separate task and outputs a table showing the PID,
|
|
stack size, stack used, and thread name (if available) for each task.
|
|
The monitoring continues until explicitly stopped with the
|
|
``stackmonitor_stop`` command.
|
|
|
|
Configuration
|
|
=============
|
|
|
|
Enable the command with ``CONFIG_SYSTEM_STACKMONITOR``. This option depends
|
|
on:
|
|
|
|
- ``CONFIG_FS_PROCFS`` - procfs filesystem support
|
|
- ``CONFIG_STACK_COLORATION`` - stack coloration feature for usage tracking
|
|
|
|
The following configuration options are available:
|
|
|
|
``CONFIG_SYSTEM_STACKMONITOR_STACKSIZE``
|
|
Stack size for the monitor daemon task. Default: 2048 bytes.
|
|
|
|
``CONFIG_SYSTEM_STACKMONITOR_PRIORITY``
|
|
Priority for the monitor daemon task. Default: 50.
|
|
|
|
``CONFIG_SYSTEM_STACKMONITOR_INTERVAL``
|
|
Monitoring interval in seconds. Default: 2 seconds.
|
|
|
|
``CONFIG_SYSTEM_STACKMONITOR_MOUNTPOINT``
|
|
procfs mount point path. Default: ``/proc``.
|
|
|
|
Usage
|
|
=====
|
|
|
|
.. code-block:: console
|
|
|
|
stackmonitor
|
|
|
|
Start the stack monitoring daemon. If the daemon is already running, display
|
|
its current status.
|
|
|
|
.. code-block:: console
|
|
|
|
stackmonitor_stop
|
|
|
|
Stop the running stack monitoring daemon.
|
|
|
|
Options
|
|
=======
|
|
|
|
The ``stackmonitor`` and ``stackmonitor_stop`` commands accept no options.
|
|
|
|
Examples
|
|
========
|
|
|
|
Start the stack monitor and observe the output:
|
|
|
|
.. code-block:: console
|
|
|
|
nsh> stackmonitor
|
|
Stack Monitor: Started: 3
|
|
|
|
The daemon will periodically output stack usage information:
|
|
|
|
.. code-block:: console
|
|
|
|
PID SIZE USED THREAD NAME
|
|
0 2048 1024 Idle_Task
|
|
1 2048 512 hpwork
|
|
2 2048 768 nsh_main
|
|
3 2048 256 Stack Monitor
|
|
|
|
Check the daemon status:
|
|
|
|
.. code-block:: console
|
|
|
|
nsh> stackmonitor
|
|
Stack Monitor: Running: 3
|
|
|
|
Stop the daemon:
|
|
|
|
.. code-block:: console
|
|
|
|
nsh> stackmonitor_stop
|
|
Stack Monitor: Stopping: 3
|
|
Stack Monitor: Stopped: 3
|
|
|
|
Notes
|
|
=====
|
|
|
|
- The daemon requires the procfs filesystem to be mounted at the configured
|
|
mount point (default ``/proc``).
|
|
- Stack usage tracking depends on the ``CONFIG_STACK_COLORATION`` feature,
|
|
which fills unused stack space with a known pattern to detect usage.
|
|
- The daemon will automatically stop after 100 consecutive errors reading
|
|
the procfs directory.
|
|
- The monitoring interval can be adjusted at compile time using
|
|
``CONFIG_SYSTEM_STACKMONITOR_INTERVAL``.
|
|
- The output format includes the thread name only when
|
|
``CONFIG_TASK_NAME_SIZE`` is greater than 0.
|