Running Minerva Tests

There are two classes of tests built during a Minerva build: host tests and target tests. Host tests run on the build machine, and use Lagom to build Minerva userspace libraries for the host platform. Target tests run on the Minerva machine, either emulated or bare metal.

Running Host Tests

There are two ways to build host tests: from a full build, or from a Lagom-only build. The only difference is the CMake command used to initialize the build directory.

For a full build, pass -DBUILD_LAGOM=ON to the CMake command.

cmake -GNinja -S Meta/CMake/Superbuild -B Build/superbuild-x86_64 -DBUILD_LAGOM=ON

For a Lagom-only build, pass the Lagom directory to CMake. The BUILD_LAGOM CMake option is still required.

cmake -GNinja -S Meta/Lagom -B Build/lagom -DBUILD_LAGOM=ON

In both cases, the tests can be run via ninja after doing a build. Note that test-js requires the MINERVA_SOURCE_DIR environment variable to be set to the root of the minerva source tree when running on a non-Minerva host.

# /path/to/minerva repository
export MINERVA_SOURCE_DIR=${PWD}
cd Build/lagom
ninja
ninja test

To see the stdout/stderr output of failing tests, the recommended way is to set the environment variable CTEST_OUTPUT_ON_FAILURE to 1.

CTEST_OUTPUT_ON_FAILURE=1 ninja test

# or, using ctest directly...
ctest --output-on-failure

Running with Sanitizers

CI runs host tests with Address Sanitizer and Undefined Sanitizer instrumentation enabled. These tools catch many classes of common C++ errors, including memory leaks, out of bounds access to stack and heap allocations, and signed integer overflow. For more info on the sanitizers, check out the Address Sanitizer wiki page, or the Undefined Sanitizer documentation from clang.

Note that a sanitizer build will take significantly longer than a non-sanitizer build, and will mess with caches in tools such as ccache. The sanitizers can be enabled with the -DENABLE_FOO_SANITIZER set of flags. For the Minerva target, only the Undefined Sanitizers is supported.

cmake -GNinja -S Meta/Lagom -B Build/lagom -DBUILD_LAGOM=ON -DENABLE_ADDRESS_SANITIZER=ON -DENABLE_UNDEFINED_SANITIZER=ON
cd Build/lagom
ninja
CTEST_OUTPUT_ON_FAILURE=1 MINERVA_SOURCE_DIR=${PWD}/../.. ninja test

To ensure that Undefined Sanitizer errors fail the test, the halt_on_error flag should be set to 1 in the environment variable UBSAN_OPTIONS.

UBSAN_OPTIONS=halt_on_error=1 CTEST_OUTPUT_ON_FAILURE=1 MINERVA_SOURCE_DIR=${PWD}/.. ninja test

Running Target Tests

Tests built for the Minerva target get installed either into /usr/Tests or /bin. /usr/Tests is preferred, but some system tests are installed into /bin for historical reasons.

The easiest way to run all of the known tests in the system is to use the run-tests-and-shutdown.sh script that gets installed into /home/anon/Tests. When running in CI, the environment variable $DO_SHUTDOWN_AFTER_TESTS is set, which will run shutdown -n after running all the tests.

For completeness, a basic on-target test run will need the Minerva image built and run via QEMU.

cmake -GNinja -S Meta/CMake/Superbuild -B Build/superbuild-x86_64
cmake --build Build/superbuild-x86_64
cd Build/x86_64
ninja install && ninja qemu-image && ninja run

In the initial terminal, one can easily run the test runner script:

courage ~ $ ./Tests/run-tests-and-shutdown.sh
=== Running Tests on Minerva ===
...

CI runs the tests in self-test mode, using the 'ci' run options and the TestRunner entry in /etc/SystemServer.ini to run tests automatically on startup.

The system server entry looks as below:

[TestRunner@ttyS0]
Executable=/home/anon/Tests/run-tests-and-shutdown.sh
StdIO=/dev/ttyS0
Environment=DO_SHUTDOWN_AFTER_TESTS=1 TERM=xterm PATH=/usr/local/bin:/usr/bin:/bin
User=anon
WorkingDirectory=/home/anon
SystemModes=self-test

/dev/ttyS0 is used as stdio because that serial port is connected when qemu is run with -display none and -serial stdio, and output to it will show up in the stdout of the qemu window. Separately, the CI run script redirects the serial debug output to ./debug.log so that both stdout of the tests and the dbgln from the kernel/tests can be captured.

To run with CI's TestRunner system server entry, Minerva needs booted in self-test mode. Running the following shell lines will boot Minerva in self-test mode, run tests, and exit. Note that CI also sets panic=shutdown to terminate qemu; the default value halt keeps qemu around, which allows you to inspect the state.

export MINERVA_RUN=ci
export MINERVA_KERNEL_CMDLINE="graphics_subsystem_mode=off system_mode=self-test"
ninja run