Kcov - code coverage

Kcov is a code coverage tester based on bcov by Thomas Neumann. It allows collecting code coverage information from executables without special command-line arguments, and continuosly produces output from long-running applications. If you use and like kcov, feel free to make a small donation:

Using kcov is as simple as

  $ kcov /path/to/outdir executable [args for the executable]

So what are the similarities and differences when comparing kcov with bcov? First the similarities:

Both generate lcov-style HTML output
Both use DWARF debugging information to do coverage analysis on binaries compiled without special options
Both store data incrementally, so that you can re-run your program with different arguments to build on the coverage collection
Both are based on the same source code - that of bcov

The differences are:

kcov collects coverage information and generates HTML output in a single step, whereas bcov first collects data and then generates HTML on the output
kcov generates an updated web page every second, so long-running programs can be watched "live"
kcov allows output to be sorted, either by coverage percentage or filename
kcov allows specifying paths to include and exclude in the coverage, thereby limiting noise from system header files etc
kcov allows multiple program coverage to be kept in a single HTML output directory, and will automatically add more programs as they are run
bcov is written in C++, kcov in C

You can see some example kcov coverage output for C64-network.org and kcov itself here. Source code for Kcov is available at github and more information on the implementation can be found in this blog post. A screenshot from the report:

Download

The latest version is available here:

kcov-8.tar.gz

Author: Simon Kagstrom