Running the full test suite takes a long time. It would be useful to
distribute tests across multiple sandbox instances to speed up testing.
Add support for running tests in parallel across multiple sandbox
instances. Each worker runs a subset of tests based on its worker ID.
Add -P<n>:<w> option to the ut command where n is the total number of
workers and w is this worker's ID (0 to n-1). Tests are distributed
by index modulo number of workers.
Series-to: u-boot
Series-cc: heinrich
Cover-letter:
test: Various improvements to unit-test infrastructure
This series adds several improvements to the unit-test infrastructure:
- Move disk images to the persistent-data directory so they don't
pollute the source tree
- Add a way to keep pytest-created artefacts for faster iteration on
C tests
- Add a helper to simplify running manual unit tests from Python
- Allow combined flags with the ut command (e.g. -Efm)
- Add a -E flag to emit machine-readable result lines
- Add a -P flag to distribute tests across parallel workers
- Add -m as an alias for -f (force manual tests)
These changes make it easier to run and debug tests, especially when
iterating on C test code.
END
Co-developed-by: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: Simon Glass <simon.glass@canonical.com>
The ut command shows test output but does not provide a machine-readable
indication of whether each individual test passed or failed. External
tools must rely on heuristics like scanning for failure patterns in the
output.
Add a -E flag that emits an explicit result line after each test:
Result: PASS: test_name: file.c
Result: FAIL: test_name: file.c
This allows tools to reliably determine per-test pass/fail status
without fragile pattern matching. The flag is optional to maintain
backward compatibility with existing scripts.
Co-developed-by: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: Simon Glass <simon.glass@canonical.com>
Add ut_get_str(), ut_get_int(), and ut_get_bool() functions with
corresponding ut_str(), ut_int(), and ut_bool() macros for accessing
test arguments with type checking.
These functions check that the argument index is within bounds and the
type matches what was requested.
The first failure for a test is reported via ut_failf() which should
make it fairly easy to debug the test.
Co-developed-by: Claude <noreply@anthropic.com>
Signed-off-by: Simon Glass <simon.glass@canonical.com>
Add support for passing key=value arguments to unit tests. The test
framework parses arguments based on definitions provided by each test
and makes them available via uts->args[]
For now the 'ut' command does not support this new feature.
Co-developed-by: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: Simon Glass <simon.glass@canonical.com>
Some tests cannot run when the console is silent.
An example is an expo test which checks text entry into a textline
object. The console must be enabled so that the characters actually
reach the putc_xy() in console_truetype, since in
scene_textline_send_key(), the lineedit restores the vidconsole state,
outputs the character and then saves the state again. If the character
is never output, then the state won't be updated and the lineedit will
be inconsistent.
Rather than having individual tests handle this manually, add an
explicit flag, in the hope that this quirk does not trip anyone else up.
Put the flag next to the existing UTF_CONSOLE flag, since they are
related.
Signed-off-by: Simon Glass <simon.glass@canonical.com>
The test.py tests pass -P to sandbox to tell it to bypass the pager.
Tests which change this value must restore it to the old value.
Fix this, so that the -P setting remains in place across test runs.
Signed-off-by: Simon Glass <sjg@chromium.org>
We generally don't want the pager to be active when running tests,
since U-Boot appears to hang forever. Perhaps we could detect when the
tests are being run interactively and use the pager in that case. But
for now, just bypass it.
Co-developed-by: Claude <noreply@anthropic.com>
Signed-off-by: Simon Glass <sjg@chromium.org>
Normally the first failed assertion causes the test to stop. Provide a
flag to allow the test to continue, as a means to discover other
problems later in the test.
Note that the utility of this depends on the test, since some tests will
be broken if any step fails. But this can be useful when there are lots
of test adjustments to make.
Signed-off-by: Simon Glass <sjg@chromium.org>
When OF_LIVE is enabled, sandbox runs tests with that and with flattree.
When debugging tests, this is often just a distraction.
Provide a -F option to skip the flattree tests in this case.
Signed-off-by: Simon Glass <sjg@chromium.org>
Enhance the ut command to accept a comma-separated list of test suites
to run. Report the summary information for these at the end.
Signed-off-by: Simon Glass <sjg@chromium.org>
Some suites need things to be set up before they can run. Add a way to
declare an init function using the UNIT_TEST_INIT() macro. The init
function is just like any other test, but is always placed first so that
it runs before all the other test functions in the suite.
Add an uninit function as well, to clean up after the test.
Signed-off-by: Simon Glass <sjg@chromium.org>
When tests are all in the same suite it is annoying to have to read all
the common text after each name. Skip this to help the user.
Signed-off-by: Simon Glass <sjg@chromium.org>
Show the average duration of a test, so we can keep track of how it is
trending. Report the suite with the longest average test to encourage
people to improve it.
Add a function to update the stats based on the results from a single
suite and another to show the summary information.
Make this optional, since sandbox's SPL tests do not have a timer driver
and people may want to print results without times.
Signed-off-by: Simon Glass <sjg@chromium.org>
Show the time taken by each test suite with 'ut all' and the total time
for all suites.
Take care to remove any sandbox time-offset from the values.
Fix the comment-format on timer_test_add_offset() while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a function to show the stats, so we can decide when to print it.
This slightly adjusts the output, so that any 'test not found' message
appears on its own line after all other output.
The 'failures' message now appears in lower case so update pytest
accordingly.
Signed-off-by: Simon Glass <sjg@chromium.org>
This is useful information and is not always the same as the 'count' arg
to ut_run_list() so add it as a separate stat.
Signed-off-by: Simon Glass <sjg@chromium.org>
Use a struct to hold the stats, since we also want to have the same
stats for all runs as we have for each suite.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a Python test which runs 'ut all' and then checks that the expected
suites are present and all tests in each suite are run.
This can help to check that nothing is missing.
Update 'ut info' to ignore the 'all' suite when counting the number of
suites, since that is really just a combination of all the other suites.
Adjust the message for skipped tests so that appears even if no
particular test was selected. This helps the new 'test_suite' test see
what is going on.
Signed-off-by: Simon Glass <sjg@chromium.org>
Move these operations into separate functions so that it is clearer what
is needed. These functions can also be called from somewhere other than
ut_run_list().
Signed-off-by: Simon Glass <sjg@chromium.org>
When mentioning a test name, add single quotes to make it easier to see.
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Tom Rini <trini@konsulko.com> # rpi_3, rpi_4, rpi_arm64, am64x_evm_a53, am64-sk
This should show the test name, not the selected name, since the user
may be running all tests, in which case 'select_name' is NULL
Fix it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Tested-by: Tom Rini <trini@konsulko.com> # rpi_3, rpi_4, rpi_arm64, am64x_evm_a53, am64-sk
Tests which create a new bloblist overwrite the existing one in sandbox.
Provide a flag for tests to declare this behaviour. Save and restore the
bloblist pointer so that other tests remain unaffected.
Note that when sandbox is running normally, the bloblist has been
relocated to high in memory. The existing bloblist tests create a new
bloblist low in memory, so they do not conflict.
Correct a build error on coreboot by using accessors for gd->bloblist:
Signed-off-by: Simon Glass <sjg@chromium.org>
This function assumes that all tests in a suite are being run. This
means that it can sometimes call dm_test_restore() when it should not.
The impact of this is that it is not possible, for example, to run
'ut bootstd bootflow_cros' and then check the state of bootstd
afterwards, since all devices are removed and recreated.
Update the function to take account of any selected test, to avoid this
problem.
Add a comment for test_insert while we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
A comment in test-main.c was not updated with the recent rename. Fix it.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
When a test returns -EAGAIN this should not be considered a failure.
Fix what seems to be a problem case, where the pytests see a failure
when a test has merely been skipped.
We cannot squash the -EAGAIN error in ut_run_test() since the failure
count is incremented by its caller, ut_run_test_live_flat()
The specific example here is on snow, where a test is compiled into the
image but cannot run, so returns -EAGAIN to skip:
test/py/tests/test_ut.py sssnow # ut bdinfo bdinfo_test_eth
Test: bdinfo_test_eth: bdinfo.c
Skipping: Console recording disabled
test/test-main.c:486, ut_run_test_live_flat(): 0 == ut_run_test(uts,
test, test->name): Expected 0x0 (0), got 0xfffffff5 (-11)
Test bdinfo_test_eth failed 1 times
Skipped: 1, Failures: 1
snow # F+u-boot-test-reset snow snow
The fix is simply to respect the return code from ut_run_test(), so do
that.
Signed-off-by: Simon Glass <sjg@chromium.org>
Complete this rename for all directories outside arch/ board/ drivers/
and include/
Use the new symbol to refer to any 'SPL' build, including TPL and VPL
Signed-off-by: Simon Glass <sjg@chromium.org>
Clear any USB-keyboard devices before running a unit test, to avoid
using a stale udevice pointer in stdio. Add a long comment to explain
this situation and why this solution seems best, at least for now.
Signed-off-by: Simon Glass <sjg@chromium.org>
The _REC suffix doesn't add much. Really what we want to know is whether
the test uses the console, so rename this flag.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
The UT_TESTF_ macros read as 'unit test test flags' which is not right.
Rename to UTF ('unit test flags').
This has the benefit of being shorter, which helps keep UNIT_TEST()
declarations on a single line.
Give the enum a name and reference it from the UNIT_TEST() macros while
we are here.
Signed-off-by: Simon Glass <sjg@chromium.org>
As part of bringing the master branch back in to next, we need to allow
for all of these changes to exist here.
Reported-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Tom Rini <trini@konsulko.com>
When bringing in the series 'arm: dts: am62-beagleplay: Fix Beagleplay
Ethernet"' I failed to notice that b4 noticed it was based on next and
so took that as the base commit and merged that part of next to master.
This reverts commit c8ffd1356d, reversing
changes made to 2ee6f3a5f7.
Reported-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Tom Rini <trini@konsulko.com>
Sandbox unit tests in U-Boot proper load a test device tree to have some
devices to work with. In order to do the same in SPL, we must enable
SPL_OF_REAL. However, we already have SPL_OF_PLATDATA enabled. When
generating platdata from a devicetree, it is expected that we will not need
devicetree access functions (even though SPL_OF_CONTROL is enabled). This
expectation does not hold for sandbox, so allow user control of
SPL_OF_REAL.
There are several places in the tree where conditions involving OF_PLATDATA
or OF_REAL no longer function correctly when both of these options can be
selected at the same time. Adjust these conditions accordingly.
Signed-off-by: Sean Anderson <seanga2@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
SPL doesn't have OF_LIVE enabled, so we can only run tests with a flat
tree. Don't skip them even if they don't use the devicetree.
Fixes: 6ec5178c0e ("test: Skip flat-tree tests if devicetree is not used")
Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Many tests don't actually use the devicetree at all so there is no point
in running the tests both with livetree and flat tree. Check for this and
skip the flat tree test in that case.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
dm_test_restore() is called after dm unit test is run.
But this function does not scan any nodes under /firmware since
it calls dm_scan_fdt().
This causes an issue. For instance, scmi_sandbox_agent device
will disappear after running 'ut dm scmi_sandbox_agent'.
So call dm_extended_scan() instead. This change will be coherent
with what dm_scan() and test_pre_run() does.
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
This is pretty messy at present since it relies on a SPL_SANDBOX option
that does not exist. Use the normal options instead, so that it will work
with split config.
Signed-off-by: Simon Glass <sjg@chromium.org>
Most tests don't want these and they can create a lot of noise. Add a way
to disable them. Use that in tests, with a flag provided to enable them
for tests that need this feature.
Signed-off-by: Simon Glass <sjg@chromium.org>
Most tests don't want these and can create a lot of noise. Add a way to
disable them. Use that in tests, with a flag provided to enable them for
tests that need this feature.
Signed-off-by: Simon Glass <sjg@chromium.org>
We must call dm_scan_other() after devices from the device tree have been
created, since that function behaves differently if there is no bootstd
device.
Adjust the logic to achieve this.
Also fix the bootflow_system() test which was relying on this broken
behaviour.
Signed-off-by: Simon Glass <sjg@chromium.org>
When running unit tests, some may have side effects which cause a
subsequent test to break. This can sometimes be seen when using 'ut dm'
or similar.
Add a new argument which allows a particular (failing) test to be run
immediately after a certain number of tests have run. This allows the
test causing the failure to be determined.
Update the documentation also.
Signed-off-by: Simon Glass <sjg@chromium.org>
Some tests access data in block devices and so cause the cache to fill
up. This results in memory being allocated.
Some tests check the malloc usage at the beginning and then again at the
end, to ensure there is no memory leak caused by the test. The block cache
makes this difficult, since the any test may cause entries to be allocated
or even freed, if the cache becomes full.
It is simpler to clear the block cache after each test. This ensures that
it will not introduce noise in tests which check malloc usage.
Add the logic to clear the cache, using the existing blkcache_invalidate()
function. Drop the duplicate code at the same time.
Signed-off-by: Simon Glass <sjg@chromium.org>
Currently, we must call cyclic_init() at some point before
cyclic_register() becomes possible. That turns out to be somewhat
awkward, especially with SPL, and has resulted in a watchdog callback
not being registered, thus causing the board to prematurely reset.
We already rely on gd->cyclic reliably being set to NULL by the asm
code that clears all of gd. Now that the cyclic list is a hlist, and
thus an empty list is represented by a NULL head pointer, and struct
cyclic_drv has no other members, we can just as well drop a level of
indirection and put the hlist_head directly in struct
global_data. This doesn't increase the size of struct global_data,
gets rid of an early malloc(), and generates slightly smaller code.
But primarily, this avoids having to call cyclic_init() early; the cyclic
infrastructure is simply ready to register callbacks as soon as we
enter C code.
We can still end up with schedule() being called from asm very early,
so we still need to check that gd itself has been properly initialized
[*], but once it has, gd->cyclic_list is perfectly fine to access, and
will just be an empty list.
As for cyclic_uninit(), it was never really the opposite of
cyclic_init() since it didn't free the struct cyclic_drv nor set
gd->cyclic to NULL. Rename it to cyclic_unregister_all() and use that
in test/, and also insert a call at the end of the board_init_f
sequence so that gd->cyclic_list is a fresh empty list before we enter
board_init_r().
A small piece of ugliness is that I had to add a cast in
cyclic_get_list() to silence a "discards 'volatile' qualifier"
warning, but that is completely equivalent to the existing handling of
the uclass_root_s list_head member.
[*] I'm not really sure where we guarantee that the register used for
gd contains 0 until it gets explicitly initialized, but that must be
the case, otherwise testing gd for being NULL would not make much sense.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Tested-by: Tim Harvey <tharvey@gateworks.com> # imx8mm-venice-*
