Configuring logging

Overview

pyOCD provides extensive control over log output. It uses the standard Python logging package for all its logging. There are several ways to set log levels, both globally and with precise control.

  • Verbosity controls
  • Logger-level control
  • Advanced configuration

Log levels

There are multiple log levels, in order from least to most verbose:

  • CRITICAL
  • ERROR
  • WARNING
  • INFO
  • DEBUG

The CRITICAL level is used only by the pyocd tool for reporting fatal errors.

Each subcommand for the pyocd tool has a default logging level.

Subcommand Default level
commander WARNING
erase WARNING
flash WARNING
gdbserver INFO
json Logging fully disabled
list INFO
pack INFO
reset WARNING
rtt INFO
server INFO

Basic control

For most users, the command line --verbose/-v and --quiet/-q arguments provide sufficient control over logging. These arguments can be listed multiple times. Each use increases or decreases the logging verbosity level. For example, a single --verbose moves pyocd flash from the default level of WARNING to INFO.

Color logging

By default, log output to a tty is colorised. Control over colorised log output is possible two ways.

The command-line --color argument accepts an optional parameter that must be one of auto, always, or never. The default is auto, which will enable color only when outputting to a tty.

Another option for controlling color output is the PYOCD_COLOR environment variable. It should be set to one of the same values supported by --color. This environment variable changes the default color output setting, and is overridden by --color on the command line.

Loggers

Each module in pyOCD uses its own module-specific logger with a name matching the dotted module name, for instance pyocd.coresight.fpb. This lets you control verbosity at the module level. Even more advanced configurations, such as routing a particular module’s log output to a separate file, are also possible.

The best way to see which loggers are available is simply to look at the pyOCD source code to see its package structure.

Trace loggers

Certain modules define additional sub-module loggers that output debug trace logs. These loggers always have the suffix “.trace” and are disabled by default. This ensures the trace messages won’t be seen unless explicitly enabled by the --log-level / -L argument described in the following section.

Currently defined trace loggers:

Trace logger Trace output
pyocd.coresight.ap.trace AP memory transfers
pyocd.coresight.dap.trace AP and DP register accesses
pyocd.debug.semihost.trace Semihost file operations
pyocd.flash.flash.trace Flash algorithm operations
pyocd.probe.cmsis_dap_probe.trace CMSIS-DAP probe API calls
pyocd.probe.jlink_probe.trace Log output from JLink library
pyocd.probe.pydapaccess.dap_access_cmsis_dap.trace CMSIS-DAP packet building
pyocd.probe.pydapaccess.interface.hidapi_backend.trace CMSIS-DAP v1 hidapi backend USB transfers
pyocd.probe.pydapaccess.interface.pyusb_backend.trace CMSIS-DAP v1 pyusb backend USB transfers
pyocd.probe.pydapaccess.interface.pyusb_v2_backend.trace CMSIS-DAP v2 pyusb backend USB transfers
pyocd.probe.pydapaccess.interface.pywinusb_backend.trace CMSIS-DAP v1 pywinusb backend USB transfers
pyocd.probe.stlink.usb.trace STLink USB transfers
pyocd.probe.tcp_client_probe.trace Remote probe client requests and responses
pyocd.probe.tcp_probe_server.trace Remote probe server requests and responses
pyocd.utility.notification.trace Sent notifications

Logger-level control

The --log-level / -L command line argument makes it easy to control logging at the level of individual loggers or groups of loggers. The argument accepts a comma-separated list of logger names followed by an “=” sign and a log level name (case-insensitive).

The logger names are actually glob-style patterns, as supported by many command line shells. This allows use of these wildcards:

Pattern Meaning
* matches everything
? matches any single character
[seq] matches any character in seq
[!seq] matches any character not in seq
[*] match literal “*”
[?] match literal “?”

The --log-level argument can be used more than once on a command line. The arguments are processed in the order they appear, so later arguments can refine log level settings made by earlier arguments.

Setting the log level of parent loggers affects the level of all child loggers.

Examples:

  • -L 'pyocd.probe.*=debug': set all pyOCD debug probe modules to debug log level
  • -L pyocd.core.session,pyocd.core.options=info: set two modules to info log level
  • -L '*.trace=debug -L *.jlink=warning': enable all trace loggers, but set JLink and its trace logger to warnings only

Note that you may need to place the value used with --log-level in quotes to prevent the shell from attempting to expand wildcards as file names.

Advanced control

Fine-grained control of pyOCD log output is available through logging configuration. The logging package supports loading a configuration dictionary to control almost all aspects of log output.

The logging session option is used to specify the logging configuration. It can be set to either a logging configuration dictionary or the path to a YAML file containing a configuration dictionary. Usually it is easiest to include the configuration directly in a pyocd.yaml config file. See the configuration documentation for more on config files. The file path is most useful when passing the logging option via the command line, since you can’t provide a dictionary this way.

Controlling module log levels

A basic logging configuration to control verbosity at the module level looks like this, as shown in a pyocd.yaml config file:

logging:
  loggers:
    pyocd.flash.loader:
      level: DEBUG
    pyocd.flash.flash_builder:
      level: DEBUG

The top level logging key is the session option. Under it must be a loggers key, which has the name of each module you wish to configure as a child key. Then, under each module name, the level key specifies the log level for that module. Due to the way logging propagation works, you do not need to set the level of parent loggers to match the child levels. In fact, setting the level of a parent logger such as pyocd will set the level for all children—this is an easy way to control the log level for all of pyOCD.

Note that because the logging option is passed to and handled by the Python logging module, it does not support wildcard matching against loggers like the --log-level argument.

Full control

The full schema for the logging configuration dictionary is documented in the logging.config module documentation. The logging module’s advanced tutorial has a good introduction to the features and log output flow, so you can better understand the configuration schema.

The version key described in the schema is optional in pyOCD’s logging configuration. If not present, pyOCD will set the schema version to 1 (currently the only version). In addition, pyOCD will set the disabled_existing_loggers key to false unless it is specified in the configuration (the default is true).

Note that if you change the configuration for the root logger, you will need to define a handler and formatter in the configuration (see the example below).

Here is a much more complex example configuration that sets a custom formatter and changes several log levels:

logging:
  formatters:
    brief:
      format: '%(relativeCreated)07d - %(levelname)s - %(name)s - %(message)s'
  handlers:
    console:
      class: logging.StreamHandler
      formatter: brief   # reference to "brief" formatter above
      level: DEBUG
      stream: ext://sys.stdout
  root:
    level: INFO
    handlers:
      - console          # reference to "console" handler above
  loggers:
    pyocd:
      level: INFO        # set all pyocd loggers to INFO level
    pyocd.probe:
      level: DEBUG       # set this logger to DEBUG level

This example shows how to direct log output to a log file called pyocd_log.txt:

logging:
  root:
    handlers: [logfile]
  formatters:
    precise:
      format: "[%(relativeCreated)07d:%(levelname)s:%(module)s] %(message)s"
  handlers:
    logfile:
      class: logging.FileHandler
      formatter: precise
      filename: pyocd_log.txt
      mode: w
      delay: false