1.. SPDX-License-Identifier: GPL-2.0
2
3============================
4Tips For Running KUnit Tests
5============================
6
7Using ``kunit.py run`` ("kunit tool")
8=====================================
9
10Running from any directory
11--------------------------
12
13It can be handy to create a bash function like:
14
15.. code-block:: bash
16
17	function run_kunit() {
18	  ( cd "$(git rev-parse --show-toplevel)" && ./tools/testing/kunit/kunit.py run $@ )
19	}
20
21.. note::
22	Early versions of ``kunit.py`` (before 5.6) didn't work unless run from
23	the kernel root, hence the use of a subshell and ``cd``.
24
25Running a subset of tests
26-------------------------
27
28``kunit.py run`` accepts an optional glob argument to filter tests. Currently
29this only matches against suite names, but this may change in the future.
30
31Say that we wanted to run the sysctl tests, we could do so via:
32
33.. code-block:: bash
34
35	$ echo -e 'CONFIG_KUNIT=y\nCONFIG_KUNIT_ALL_TESTS=y' > .kunit/.kunitconfig
36	$ ./tools/testing/kunit/kunit.py run 'sysctl*'
37
38We're paying the cost of building more tests than we need this way, but it's
39easier than fiddling with ``.kunitconfig`` files or commenting out
40``kunit_suite``'s.
41
42However, if we wanted to define a set of tests in a less ad hoc way, the next
43tip is useful.
44
45Defining a set of tests
46-----------------------
47
48``kunit.py run`` (along with ``build``, and ``config``) supports a
49``--kunitconfig`` flag. So if you have a set of tests that you want to run on a
50regular basis (especially if they have other dependencies), you can create a
51specific ``.kunitconfig`` for them.
52
53E.g. kunit has one for its tests:
54
55.. code-block:: bash
56
57	$ ./tools/testing/kunit/kunit.py run --kunitconfig=lib/kunit/.kunitconfig
58
59Alternatively, if you're following the convention of naming your
60file ``.kunitconfig``, you can just pass in the dir, e.g.
61
62.. code-block:: bash
63
64	$ ./tools/testing/kunit/kunit.py run --kunitconfig=lib/kunit
65
66.. note::
67	This is a relatively new feature (5.12+) so we don't have any
68	conventions yet about on what files should be checked in versus just
69	kept around locally. It's up to you and your maintainer to decide if a
70	config is useful enough to submit (and therefore have to maintain).
71
72.. note::
73	Having ``.kunitconfig`` fragments in a parent and child directory is
74	iffy. There's discussion about adding an "import" statement in these
75	files to make it possible to have a top-level config run tests from all
76	child directories. But that would mean ``.kunitconfig`` files are no
77	longer just simple .config fragments.
78
79	One alternative would be to have kunit tool recursively combine configs
80	automagically, but tests could theoretically depend on incompatible
81	options, so handling that would be tricky.
82
83Generating code coverage reports under UML
84------------------------------------------
85
86.. note::
87	TODO(brendanhiggins@google.com): There are various issues with UML and
88	versions of gcc 7 and up. You're likely to run into missing ``.gcda``
89	files or compile errors.
90
91This is different from the "normal" way of getting coverage information that is
92documented in Documentation/dev-tools/gcov.rst.
93
94Instead of enabling ``CONFIG_GCOV_KERNEL=y``, we can set these options:
95
96.. code-block:: none
97
98	CONFIG_DEBUG_KERNEL=y
99	CONFIG_DEBUG_INFO=y
100	CONFIG_GCOV=y
101
102
103Putting it together into a copy-pastable sequence of commands:
104
105.. code-block:: bash
106
107	# Append coverage options to the current config
108	$ echo -e "CONFIG_DEBUG_KERNEL=y\nCONFIG_DEBUG_INFO=y\nCONFIG_GCOV=y" >> .kunit/.kunitconfig
109	$ ./tools/testing/kunit/kunit.py run
110	# Extract the coverage information from the build dir (.kunit/)
111	$ lcov -t "my_kunit_tests" -o coverage.info -c -d .kunit/
112
113	# From here on, it's the same process as with CONFIG_GCOV_KERNEL=y
114	# E.g. can generate an HTML report in a tmp dir like so:
115	$ genhtml -o /tmp/coverage_html coverage.info
116
117
118If your installed version of gcc doesn't work, you can tweak the steps:
119
120.. code-block:: bash
121
122	$ ./tools/testing/kunit/kunit.py run --make_options=CC=/usr/bin/gcc-6
123	$ lcov -t "my_kunit_tests" -o coverage.info -c -d .kunit/ --gcov-tool=/usr/bin/gcov-6
124
125
126Running tests manually
127======================
128
129Running tests without using ``kunit.py run`` is also an important use case.
130Currently it's your only option if you want to test on architectures other than
131UML.
132
133As running the tests under UML is fairly straightforward (configure and compile
134the kernel, run the ``./linux`` binary), this section will focus on testing
135non-UML architectures.
136
137
138Running built-in tests
139----------------------
140
141When setting tests to ``=y``, the tests will run as part of boot and print
142results to dmesg in TAP format. So you just need to add your tests to your
143``.config``, build and boot your kernel as normal.
144
145So if we compiled our kernel with:
146
147.. code-block:: none
148
149	CONFIG_KUNIT=y
150	CONFIG_KUNIT_EXAMPLE_TEST=y
151
152Then we'd see output like this in dmesg signaling the test ran and passed:
153
154.. code-block:: none
155
156	TAP version 14
157	1..1
158	    # Subtest: example
159	    1..1
160	    # example_simple_test: initializing
161	    ok 1 - example_simple_test
162	ok 1 - example
163
164Running tests as modules
165------------------------
166
167Depending on the tests, you can build them as loadable modules.
168
169For example, we'd change the config options from before to
170
171.. code-block:: none
172
173	CONFIG_KUNIT=y
174	CONFIG_KUNIT_EXAMPLE_TEST=m
175
176Then after booting into our kernel, we can run the test via
177
178.. code-block:: none
179
180	$ modprobe kunit-example-test
181
182This will then cause it to print TAP output to stdout.
183
184.. note::
185	The ``modprobe`` will *not* have a non-zero exit code if any test
186	failed (as of 5.13). But ``kunit.py parse`` would, see below.
187
188.. note::
189	You can set ``CONFIG_KUNIT=m`` as well, however, some features will not
190	work and thus some tests might break. Ideally tests would specify they
191	depend on ``KUNIT=y`` in their ``Kconfig``'s, but this is an edge case
192	most test authors won't think about.
193	As of 5.13, the only difference is that ``current->kunit_test`` will
194	not exist.
195
196Pretty-printing results
197-----------------------
198
199You can use ``kunit.py parse`` to parse dmesg for test output and print out
200results in the same familiar format that ``kunit.py run`` does.
201
202.. code-block:: bash
203
204	$ ./tools/testing/kunit/kunit.py parse /var/log/dmesg
205
206
207Retrieving per suite results
208----------------------------
209
210Regardless of how you're running your tests, you can enable
211``CONFIG_KUNIT_DEBUGFS`` to expose per-suite TAP-formatted results:
212
213.. code-block:: none
214
215	CONFIG_KUNIT=y
216	CONFIG_KUNIT_EXAMPLE_TEST=m
217	CONFIG_KUNIT_DEBUGFS=y
218
219The results for each suite will be exposed under
220``/sys/kernel/debug/kunit/<suite>/results``.
221So using our example config:
222
223.. code-block:: bash
224
225	$ modprobe kunit-example-test > /dev/null
226	$ cat /sys/kernel/debug/kunit/example/results
227	... <TAP output> ...
228
229	# After removing the module, the corresponding files will go away
230	$ modprobe -r kunit-example-test
231	$ cat /sys/kernel/debug/kunit/example/results
232	/sys/kernel/debug/kunit/example/results: No such file or directory
233
234Generating code coverage reports
235--------------------------------
236
237See Documentation/dev-tools/gcov.rst for details on how to do this.
238
239The only vaguely KUnit-specific advice here is that you probably want to build
240your tests as modules. That way you can isolate the coverage from tests from
241other code executed during boot, e.g.
242
243.. code-block:: bash
244
245	# Reset coverage counters before running the test.
246	$ echo 0 > /sys/kernel/debug/gcov/reset
247	$ modprobe kunit-example-test
248