Dwgrep is a tool, an associated language (called Zwerg) and a library (libzwerg) for querying Dwarf (debuginfo) graphs. If you want to find out more about Dwarf, check out an introduction to the DWARF Debugging Format or download the Dwarf standard. But you can also pretend that Dwarf is like XML, except nodes are called DIE's. That, and perusing the output of eu-readelf -winfo, should be enough to get you started.

You can think of dwgrep expressions as instructions describing a path through a graph, with assertions about the type of nodes along the way: that a node is of given type, that it has a given attribute, etc. There are also means of expressing sub-conditions, i.e. assertions that a given node is acceptable if a separate expression matches (or doesn't match) a different path through the graph.

Apart from Dwarf objects (DIE's, Attributes and others), Zwerg expressions can work with integers, strings, and sequences of other objects.

In particular, a simple expression in dwgrep might look like this:

entry ?DW_TAG_subprogram child ?DW_TAG_formal_parameter @DW_AT_name

On a command line, you would issue it like this:

$ dwgrep /some/file/somewhere -e 'entry ?DW_TAG_subprogram ...etc....'

The query itself says: show me values of attribute DW_AT_name of DW_TAG_formal_parameter nodes that are children of DW_TAG_subprogram entries (which here means debug information entries, or DIE's). Reading forward, you get list of instructions to a matcher: take DIE's, accept all DW_TAG_subprogram's, look at their children, accept those that are DW_TAG_formal_parameter, take value of attribute DW_AT_name.

Another example comes from dwarflint:

entry ?DW_AT_decl_column !DW_AT_decl_line

This looks for DIE's that have DW_AT_decl_column, but don't have DW_AT_decl_line--a semantic violation that is worth reporting.


Conceptually, Zwerg expressions form a pipeline of functions. Much like the shell pipeline, the individual functions take input and produce output, which then becomes input of the next function in chain. Unlike shell pipelines, Zwerg expressions operate with values such as Dwarf file, DIE, attribute, abbreviation, number, or string, and in fact operate on stacks of these values.

Each function in pipeline thus takes on input a stack of values, and produces zero or more stacks on the output. Note that producing several stacks is different from producing a stack that holds a sequence. You can think of it as if any function could in theory fork (if it returns more than one stack), or terminate (if it doesn't return anything).

This may be best illustrated with an example. Among the most trivial Zwerg functions are literals, such as 5 or "Blah". For a stack given on input, they produce one stack on output, with the indicated value pushed to top:

$ dwgrep '"Hello, world!"'
Hello, world!

Another very simple function is length, which returns the input stack with top value replaced with its length:

$ dwgrep '"Hello, world!" length'

Both the literal and length yield a single stack of output for every stack of input. There are words that yield more than once however. An example is elem, which inspects elements of a sequence or a string:

$ dwgrep '"Hi!" elem'

dwgrep returned three results, and shows each of them on a single line. That's because they are one-slot stacks, and this way of formatting allows the output to be further processed by shell. This is how it looks for deeper stacks:

$ dwgrep '"Hi!" elem 5'

When using dwgrep on command line, it is typically most useful to aim for single-slot stacks. Producing deeper stacks may be useful if you use libzwerg programmatically. That way the query can return not only the primary result, but also some context.

Now would be a good time to read through the Tutorial, which gives a step-by-step account of fundamental tools of Zwerg language. You might also want to look at Syntax, where individual Zwerg forms are introduced and described.


dwgrep depends on the following software:

In addition, dwgrep can make use of the following software, if available:

To build dwgrep:

$ cmake .
$ make
$ make test     # optional, if you want to run the test suite
$ make doc      # optional, if you want to build the documentation
$ make install


The project homepage is at <>. GitHub also hosts a ticketing system. Other than that, if you wish to contact the author, send an e-mail to <>.

The source code is tracked in a GIT repository. Check it out with:

git clone

The best way to get the patches across is to create a GitHub merge request. Patches prepared with git-format-patch and sent by e-mail are also acceptable though.


dwgrep and libzwerg are dual-licensed as either GPL3+ or LGPL3+, at your option.