gcc-callgraph-plugin

A python script to print call graphs during compilation with GCC. It uses the gcc-python-plugin architecture.

Example output for examples/random-sort:

Features

• Display the whole callgraph, considering all compilling files.
• Performs in O(V + E), where V is the number of functions and E is the number of calls.
• Possibility to restrict callgraph with sets of starting and ending functions.
• Possibility to exclude certain functions.
• Accept any output format known by dot.
• Differentiate two or more valid functions with the same name.

Dependencies

• python version >= 3.5
• PyYAML
• gcc-python-plugin
• dot (from the graphviz package)

Using

With the gcc-python-plugin installed, compile your code with:

    $ gcc -fplugin=python \
	  -fplugin-arg-python-script=<path to gcc-callgraph-plugin.py> \
	  -flto -flto-partition=none \
	  <other args>

This will generate a callgraph.svg image in the working directory, containing the program's call graph.

Note:

• Give the full path to -fplugin-arg-python-script. Don't use ~/, for example.
• When compiling and linking in two steps, don't forget to use those flags in both of them.
• If you are compilling a code with a Makefile, you can use make CC="gcc -fplugin ..."

With docker

You may also skip installing dependencies and run the plugin in a docker container.

To build the image:

    $ docker build -t callgraph .

Then, in the directory of the project you want to generate the call graph for, run:

    $ docker run -v "$PWD:/src" -it callgraph

This will run the container interactively, giving you a bash session in the directory it was invoked. Then you can compile your project with the same command shown in the previous section. Just remember to use:

-fplugin-arg-python-script=/plugin/gcc-callgraph-plugin.py

Configuring output

The plugin will read user specified settings from a .gcc-callgraph.yml file (if it exists), in the working directory or in the user's home directory (in this order). This file must be in YAML format and can contain the following attributes (none of them are required):

• start (str or str list): set of functions to start the callgraph at. Any call chain that doesn't start in one of these is excluded.
• end (str or str list): set of functions to end the callgraph at. Any call chain that does not end in one of these is excluded.
• exclude (str or str list): set of functions to be excluded from the callgraph.
• multiple_edges (boolean): whether to display each call between functions A and B as individual edges (True) or as a single edge (False). Default is False.
• abort_on_func_not_found (boolean): whether to abort imediatelly if any function specified in the config file is not found (True), or continue with a warning (False). Default is False.
• out_file (str): output file name. Extension must be one of the formats accepted by dot, e.g. .png or .svg. Default is callgraph.svg.

Functions must be specified in the format path:function_name, where path points to the file in which the function was declared, relative to the working directory of compilation. When specified, the starting nodes will be colored blue and the end ones green.

As an example, with the following we get all paths from a function A to a function B that don't contain a function C (all declared in the same file, main.c):

start: main.c:A
end: main.c:B
exclude: main.c:C

Examples

The examples directory contain a couple of toy C projects to compile with this plugin. Every subdirectory has a M̀akefile designed to generate the call graph inside the callgraph docker container.

Extra

If you want to find paths in the call graph for a particular execution, you might want to try callpath.