Add Unittest equeue tests

[Tharazi97]

Description

Ported events queue test to the Greentea standard:
https://github.com/ARMmbed/mbed-os/blob/master/events/equeue/tests/tests.c

Pull request type

[ ] Fix
[ ] Refactor
[ ] Target update
[ ] Functionality change
[ ] Docs update
[x] Test update
[ ] Breaking change

Reviewers

@jamesbeyond @maciejbocianski

Release Notes

[ciarmcom]

@Tharazi97, thank you for your changes.
@maciejbocianski @jamesbeyond @ARMmbed/mbed-os-core @ARMmbed/mbed-os-test @ARMmbed/mbed-os-maintainers please review.

[SeppoTakalo]

Sounds a lot like unit testing.
Should most of the functionality be tested on unit/module testing framework in host machine, instead of spending time on running on embedded devices?

[jamesbeyond]

Sounds a lot like unit testing.
Should most of the functionality be tested on unit/module testing framework in host machine, instead of spending time on running on embedded devices?

It is a valid point that Greentea tests adding CI executing time. How much effort would it be to converting these test to the UNITTEST ? @Tharazi97 @maciejbocianski

[bulislaw]

It's a good point, we are trying to push the testing time down so all the not hw specific tests should be offloaded to unittest framework.

[Tharazi97]

Right I will convert them to gtest.

[Tharazi97]

I have converted them to googletests, but they don't work properly on Windows. Does it have to pass all the test on Windows, or is Windows not supported already?

[SeppoTakalo]

Not passing on Windows, and passing on Linux, means that you might have just bug that is affected by uninitialised variable.
It is just a difference of compiler and operating system, whether uninitialised are zero, or random values.

Try running your tests using Valgrind to see where the problem is.

[jamesbeyond]

I have converted them to googletests, but they don't work properly on Windows. Does it have to pass all the test on Windows, or is Windows not supported already?

@Tharazi97
NOT PASS Do you mean the new equeue test? or do you mean the existing UNITTEST ?
all existing UNITTEST should be all passed on Windows, if you git some issue that could mean your environments would have some issues

[Tharazi97]

Other tests pass. I think that it could be caused by using pthreads library on MinGW, or by wrong reads of ticker function. I'm working on it.

[Tharazi97]

I've commited some changes. Unit tests should work on Windows and Linux. But I suggest to leave greentea test as they are, or make them optional to compile, because Equeue is a real time solution and I think it should be able to test it with a ticker and threads available on targets.

@SeppoTakalo @jamesbeyond @maciejbocianski @bulislaw

[maciejbocianski]

I'm not sure if it's good idea to remove this test file as it's part of equeue library and still runs on CI (https://github.com/ARMmbed/mbed-os/blob/master/.travis.yml#L230). It should be left as is and only disable on travis script.

[maciejbocianski]

This test file also should stay, as it's part of equeue library and still runs on CI (https://github.com/ARMmbed/mbed-os/blob/master/.travis.yml#L232).

[maciejbocianski]

@bulislaw @SeppoTakalo @jamesbeyond
I think this is good idea to to leave greentea test as is, and make it optional.
It adds valuable documentation.

What do you think about adding global mechanism/flag to make our test/example codes be aware of running on CI (e.g. MBED_CI_TEST_RUN). That could easily allow to disable specific test/example code when running on CI

Similar flag SKIP_TIME_DRIFT_TESTS was introduced to skip some time drift tests on CI(86dab2f)

#ifndef MBED_CI_TEST_RUN
// when running on CI this test is disabled due to ...

// test code here

#endif 

[jamesbeyond]

@maciejbocianski @Tharazi97
What is the benefits to keep the GreenTea test?

[OPpuolitaival]

If we have some benefits of Greentea tests then we can leave those using MBED_EXTENDED_TESTS flag to be make easier to run all tests if wanted. But if there are no benefits of those then those can be removed. Good option is to leave one or two greentea tests as hardware smoke test and remove the rests.

[maciejbocianski]

What is the benefits to keep the GreenTea test?

The benefit is that we can test whole queue API on target hardware, not only partially as tests-events-queue does

[jamesbeyond]

What is the benefits to keep the GreenTea test?

The benefit is that we can test whole queue API on target hardware, not only partially as tests-events-queue does

OK, please add the macro as O-P suggested

[0xc0170]

@Tharazi97 Would it make sense to squash (clean the history) - to make a history release ready

[0xc0170]

CI started

[mbed-ci]

Test run: FAILED

Summary: 1 of 4 test jobs failed
Build number : 1
Build artifacts

Failed test jobs:

• jenkins-ci/mbed-os-ci_greentea-test

[Tharazi97]

CI fails seems not related.
| LPC55S69_S-ARMC6 | LPC55S69_S | mbed-os-tests-psa-crypto_access_control | TIMEOUT | 1800.7 | default |

[0xc0170]

Test was restarted and is all good. I had to restart merge jenkins error, should also be fine (internal CI issue fixed now).

[0xc0170]

Waiting for pr-merge fix, @ARMmbed/mbed-os-test please review

Converted to unittests

[0xc0170]

The CI status fixed, this is ready for integration

[jamesbeyond]

@ARMmbed/mbed-os-maintainers Can this be merged?

[adbridge]

This is on top of 5.14 changes so moved