Fix Coverity uninitialized pointer field Thread.h

[Tharazi97]

Description

Changed _obj_mem to be initialized with creating new thread.

Pull request type

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

Reviewers

@jamesbeyond @maciejbocianski

Release Notes

[ciarmcom]

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

[0xc0170]

The current init is in start() method - it sets this to 0:

mbed-os/rtos/source/Thread.cpp

Line 109 in b1ced1b

memset(&_obj_mem, 0, sizeof(_obj_mem));

Assuming based on how it is initialized (POD type), this could be handled in the constructor as the rest of members - to be consistent however this is valid now as it is.

[kjbracey]

I'm not a fan of the existing constructor method and its in-body init - now we're C++11 you could use proper constructor delegation, and the final most-flexible constructor could do everything via its member-initializer list.

I'd also prefer if everything was consistent - all defaults in body of constructor, all defaults in member-initializer-list, or all defaults as default-member initializers.

So make it a memset in constructor, or do a more wide-ranging refactor/tidy including real constructor delegation.

[kjbracey]

I guess this is warning about get_state accessing _obj_mem.state? It can't figure out that _tid only becomes non-NULL after a memset of _obj_mem? Otherwise, not sure why it would think this is a problem.

I think this is probably fine, but you should net off the size increase a bit by removing the memset from Thread::start.

[Tharazi97]

Yes I think it is warning caused by get_state(). In constructor _tid gets asigned to 0. Maybe it should have become nullptr in constructor'' and memset'' of _obj_mem should be left in start?

[kjbracey]

Yes I think it is warning caused by get_state(). In constructor _tid gets asigned to 0. Maybe it should have become nullptr in constructor'' and memset'' of _obj_mem should be left in start?

You could certainly set _tid to nullptr as a style change, but that won't be impacting Coverity's complaint (99% certainty).

[Tharazi97]

@kjbracey-arm you were right. I've moved memset to the constructor.

[0xc0170]

started CI

[adbridge]

restarted

[mbed-ci]

Test run: FAILED

Summary: 1 of 11 test jobs failed
Build number : 2
Build artifacts

Failed test jobs:

• jenkins-ci/mbed-os-ci_greentea-test

[Tharazi97]

CI fail seems not related.
test case: 'tests-mbed_hal-reset_reason' ..................................................... FAIL in 48.35 sec

[0xc0170]

Test restarted