As of today <zephyr/zephyr.h> is 100% equivalent to <zephyr/kernel.h>.
This patch proposes to then include <zephyr/kernel.h> instead of
<zephyr/zephyr.h> since it is more clear that you are including the
Kernel APIs and (probably) nothing else. <zephyr/zephyr.h> sounds like a
catch-all header that may be confusing. Most applications need to
include a bunch of other things to compile, e.g. driver headers or
subsystem headers like BT, logging, etc.
The idea of a catch-all header in Zephyr is probably not feasible
anyway. Reason is that Zephyr is not a library, like it could be for
example `libpython`. Zephyr provides many utilities nowadays: a kernel,
drivers, subsystems, etc and things will likely grow. A catch-all header
would be massive, difficult to keep up-to-date. It is also likely that
an application will only build a small subset. Note that subsystem-level
headers may use a catch-all approach to make things easier, though.
NOTE: This patch is **NOT** removing the header, just removing its usage
in-tree. I'd advocate for its deprecation (add a #warning on it), but I
understand many people will have concerns.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
When CONFIG_KERNEL_COHERENCE=y, it's illegal to place shared
data(like ztress_context_data in this test case) on the stack,
because that memory is incoherent with respect to other CPUs
in the system.
In this specific case, it will cause PC register to load invalid
handler function pointer and crash the Xtensa processor.
Make it static.
Signed-off-by: Ming Shao <ming.shao@intel.com>
Current ztress design supports both timer context and thread
context. The timer context always uses the first element of
several global array variables like exec_cnt[] and backoff[].
But progress_timeout() is using the last element of exec_cnt[]
to calculate the progress of timer context. It is not right.
Signed-off-by: Ming Shao <ming.shao@intel.com>
Tests with KERNEL enabled perform their cleanup logic after the suite's
after and test rules are executed. Unittests should do the same.
Signed-off-by: Yuval Peress <peress@google.com>
Updates the implementation when KERNEL is available to safely bail on
the test when the test calls fail, skip, or pass during invalid test
phases. Print a detailed message, and skip all other tests. The test
run will be marked as failed.
Signed-off-by: Yuval Peress <peress@google.com>
Move the function used for printing the phase name up so its available
for both unittest and KERNEL mode of tests.
Signed-off-by: Yuval Peress <peress@google.com>
When not running in setup, before, or in the test. Calling fail, skip,
or pass is invalid and should be considered an error. Properly handle
these cases by printing a more detailed error message and bailing the
process.
Signed-off-by: Yuval Peress <peress@google.com>
When KERNEL was not defined (unittest), the call to init_testing was
used to set a longjump target using 'stack_fail'. When triggered,
this was actually causing a segfault, because longjmp is only valid
if going directly up the stack. Since init_testing returned, it was
no longer on the stack. Instead, that logic MUST be inlined.
Signed-off-by: Yuval Peress <peress@google.com>
The current zassert macro uses implicit conversion to boolean which
has implication on analysis tools like clang-tidy-14. This add an
aditional step to create a boolean value for the evaluation instead
use the string direct which allows run analysis tool without this
warning/error.
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
It is frequent to find variable definitions like this:
```c
static const struct device *dev = DEVICE_DT_GET(...)
```
That is, module level variables that are statically initialized with a
device reference. Such value is, in most cases, never changed meaning
the variable can also be declared as const (immutable). This patch
constifies all such cases.
Signed-off-by: Gerard Marull-Paretas <gerard.marull@nordicsemi.no>
This adds the bits to pin the GCOV sections to the pinned
sections so they can be accessed during boot when demand
paging is enabled. Or else accessing them would result in
page faults.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This adds an ALIGN statement to the rodata linker snippet
for coverage. Without this, sometimes the section is not
aligned, but __init_array_start indicates an aligned
address, resulting in incorrect function pointers.
So align it.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
This commit introduces a dedicated unit testing board.
Today, a dedicated Zephyr unit testing scheme exists but is different
from how a Zephyr build generally works.
For example Kconfig is not possible, resulting on various different
hacks to pass Kconfig settings from test cases / testcase.yaml through
CMake to the code.
Some directly as compile definitions, some as header files with forced
inclusion on sources, some with wrapper flags which again results in
different define being enabled. There is even cases where a second
forced header inclusion undefines previous defines.
Unit test often does a manual check for the right boards, like this:
> if (NOT BOARD STREQUAL unit_testing)
> message(FATAL_ERROR "This project can only be used with...")
> endif()
Introducing a dedicated unit_testing board under `tests/root` allows
us to use Kconfig in unit test samples, and thus proper `prj.conf` and
extra Kconfig fragments.
Generation of autoconf.h so the overall architecture follows regular
Zephyr builds.
Proper and uniform error messages when invalid board is selected.
The unit_testing board and arch is located under: `subsys/testsuite` so
that it is only available when find_package(Zephyr COMPONENTS unittest)
is used, and not available for regular Zephyr builds.
Kconfig generates autoconf.h which is applied as compile flag to
test binary which means that kconfig defines placed in ztest.h can now
be removed.
Signed-off-by: Torsten Rasmussen <Torsten.Rasmussen@nordicsemi.no>
Add a detailed docstring to the zassume() macro that describes what it does
and how its usage is different from zassert.
Signed-off-by: Aaron Massey <aaronmassey@google.com>
Add duration stats at both suite level and unit test
level into test summary. The duration is at second
level. Since the new ztest fx can execute a test suite
for multiple times, the worst/longest test duration is
collected.
Note that even a skipped test can have a duration greater
than 0 because the skip operation itself is not free.
Signed-off-by: Ming Shao <ming.shao@intel.com>
Previously, unit test duration is collected within the
TC_START and Z_TC_END_RESULT macros. With existing tests,
the TC_START macro can be invoked by both the ztest fx
and the tests themselves. And the TC_START macro definition
went lengths to avoid the interference when it is invoked
within a unit test. This commit decouple the time collection
and the TC_STRAT/Z_TC_END_RESULT macros to fix this issue.
Now only the (old) ztest framework is responsible for the
test duration measure. The test duration stats of new ztest
fx is different from this btw.
Signed-off-by: Ming Shao <ming.shao@intel.com>
Add test summary after all test suites finish running.
The summary can be one-line or verbose, which is configured
with CONFIG_ZTEST_VERBOSE_SUMMARY. The one-line summary covers
overall suite stats. The verbose summary covers each test
function within the suite besides the one-line summary.
The new ztest output ultimately go through the printk. If
printk go through the logging subsystem, there may be log
messages dropped. And if log_panic is invoked, log messages
can be flushed in a mess. So several explicit log flush
are used when printing summary to ensure no content is lost
and content is in good shape.
Some macros are shared between old and new ztests. Such as
TC_START_PRINT and TC_END_PRINT. The are defined accordingly.
Signed-off-by: Ming Shao <ming.shao@intel.com>
Support verbose or one-line summary at test suite level.
Support verbose or no output at test function level.
Totally 4 combinations configurable:
- function verbose + suite verbose
- function verbose + suite oneline
- no function output + suite verbose
- no function output + suite oneline
Signed-off-by: Ming Shao <ming.shao@intel.com>
Updates the ztest_test_fail() function to allow failures in setup.
When executed, a failed assert will fail every test in the suite owning
the setup function. This was verified by adding a suite which asserts
in the setup function and has a test that should pass. During
exeuction, ztest marks the test as failing.
In order to verify exection I also added 2 new APIs:
- ZTEST_EXPECT_FAIL(suite_name, test_name)
- ZTEST_EXPECT_SKIP(suite_name, test_name)
Signed-off-by: Yuval Peress <peress@google.com>
In scenarios where test_main is overridden ztest_run_all
may be invoked multiple times leading to the verify check to
fail inadvertently.
Signed-off-by: Al Semjonovs <asemjonovs@google.com>
If the test is run with the config NO_OPTIMIZATIONS enabled then the
stack size usage increases by around 80% for ARM platforms.
Increase the stack size used in test cases that enables building with no
optimizations for ARM.
Update description on TEST_ARM_CORTEX_M since it was outdated and said
it was only used for a single purpose.
Fixes: #47930Fixes: #47929Fixes: #47855
Signed-off-by: Joakim Andersson <joakim.andersson@nordicsemi.no>
The signal handler needs to return gracefully. When multiple
tests assert, the first assert will raise SIGABRT and the signal
handler will run and the test will stop running.
The second assert will raise SIGABRT but the signal handler
will not be called, therefore the test PASS/FAIL status is not
updated correctly.
Signed-off-by: Al Semjonovs <asemjonovs@google.com>
A test case may invoke run_all multiple times, causing
logic for listing tests to run tests on second call.
Signed-off-by: Al Semjonovs <asemjonovs@google.com>
Make the option TEST_EXTRA_STACK_SIZE depend on Tests, otherwise it will
appear in each generated .config file.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
This config option is only used with ztest, so move it under ztest.
Otherwise it appears in .config when you build ANY application.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Add a bunch of missing "zephyr/" prefixes to #include statements in
various test and test framework files.
Signed-off-by: Fabio Baltieri <fabiobaltieri@google.com>
This commit removes the NULL check on the idle thread array
(`idle_tid`) because it is statically declared and will never be NULL.
This fixes the "comparison will always evaluate as ‘false’" warning
generated by the GCC 12.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Add test summary after all test suites finish running.
The summary can be one-line or verbose, which is configured
with CONFIG_ZTEST_VERBOSE_SUMMARY. The one-line summary covers
overall suite stats. The verbose summary covers each test
function within the suite besides the one-line summary.
Signed-off-by: Ming Shao <ming.shao@intel.com>
Support verbose or one-line summary at test suite level.
Support verbose or no output at test function level.
Totally 4 combinations configurable:
- function verbose + suite verbose
- function verbose + suite oneline
- no function output + suite verbose
- no function output + suite oneline
Signed-off-by: Ming Shao <ming.shao@intel.com>
Documentation now includes:
- How to create a test suite
- How to use predicates
- How to use setup/before/after/teardown
- How to use test rules
- Direct people to use FFF instead of ztest mock
Fixes#47420
Signed-off-by: Yuval Peress <peress@google.com>
Add macro Z_TEST_SKIP_IFNDEF which is used when you want to skip test
if configuration option is not enabled.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
The addition of the assumption API introduced a bug where a skipped test
would subsequently skip later tests within the same suite. A skipped test
was being treated as an entire skipped test suite.
Keep track of tests being skipped due to skipping in the suite setup versus
being skipped within their test function.
TEST=west build -p -b native_posix tests/kernel/condvar/condvar_api/ -t run
TEST=twister -T tests/ztest -p native_posix
Signed-off-by: Aaron Massey <aaronmassey@google.com>
Following zephyr's style guideline, all if statements, including single
line statements shall have braces.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Add an assume API which works like JUnit's. Assumptions can be made
at any point before your test returns (setup, before, and during the
test). If an assumption fails, the test will be marked as skipped.
This helps avoid a cascading affect of failed tests when a base
feature breaks. The feature is expected to have tests and the tests
which depend on it should be skipped (not failed) if that feature
is broken.
Issue #42472
Signed-off-by: Yuval Peress <peress@google.com>
When system clock is set globally for the test which is executed
on various qemu platforms it may happen that clock is set too
high compared to CPU clock frequency. In that case test may stuck.
Added check and test skipping in that case.
Signed-off-by: Krzysztof Chruscinski <krzysztof.chruscinski@nordicsemi.no>
The malloc partition may actually be empty, if so, don't try to add it to
the domain as that will generate an error.
Signed-off-by: Keith Packard <keithp@keithp.com>
On X86 platforms, the interrupt trigger method has been
changed to use APIC IPI, we don't use INT command to trigger
interrupt, so remove this unused macro.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
On X86 platforms, the interrupt trigger method has been changed
from using INT command to using APIC IPI, we need to make sure
the IPI interrupt is handled before do our check, so add some
nop operations.
Signed-off-by: Chen Peng1 <peng1.chen@intel.com>
