Organize source files and add Doxygen labels

[gpsimenos]

Description

• Move source files from mbed-os/platform to mbed-os/platform/source
• Move internal header files from mbed-os/platform to mbed-os/platform/internal
• Add Doxygen groups to header files

Pull request type

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

Reviewers

@hugueskamba @evedon @kjbracey-arm

[ciarmcom]

@gpsimenos, thank you for your changes.
@kjbracey-arm @hugueskamba @evedon @ARMmbed/mbed-os-core @ARMmbed/mbed-os-maintainers please review.

[evedon]

Doxygen is missing from mbed_os_timer.h. It would be useful to add.

[gpsimenos]

Doxygen is missing from mbed_os_timer.h. It would be useful to add.

Doxygen is not generating any output for this file when added to a group:

[kjbracey]

I believe this file is not included by any public header, so can go into source. (The logic being that the source directory contains all source and headers that could be removed from a binary-only distribution, or not be on the include path at all for stuff outside the Mbed OS "source" directories).

Same for mbed_os_timer.h

Except in both cases they're being accessed from a different source directory to this one. Not sure if that forces them to be outside this source directory. Depends how we imagine future path configuration would look.

[evedon]

SysTimer.h is included by mbed_os_timer.cpp & mbed_os_timer.cpp - both in platform so no problem at all to move to platform/source

mbed_os_timer.h is not included by any public header but by a few source files in rtos and platform. I would move it under platform/source for now but indeed, we might have to consolidate platform & rtos in the future or decide if headers accessed by different source directories should be moved to internal

[kjbracey]

I thought I saw one reference over in events.

If this was Yotta, then cross-component API would have to be in platform/internal, but we can figure out if we need that restriction later as you say.

[kjbracey]

This is all implementation, so shouldn't need any Doxygen marking, I think. It's implementing stuff declared in "mbed_atomic.h", which should have Doxygen bits.

[evedon]

Doxygen is missing from mbed_os_timer.h. It would be useful to add.

As per Kevin's comment, mbed_os_timer.h will be moved to source so we do not have to add doxygen.

[gpsimenos]

This force push addresses above comments. SysTimer.h and mbed_os_timer.h moved to source, Doxygen removed from mbed_os_timer.h and mbed_atomic_impl.h.

[evedon]

As noted, we'll have to revisit later if platform/source/mbed_os_timer.h should move to platform/internal

[evedon]

@0xc0170 This should go in 5.14

[0xc0170]

@0xc0170 This should go in 5.14

5.14.0 to be specific? I tagged it to 5.14.0-rc2 , cc @adbridge

[gpsimenos]

This force push to update absolute paths to the new locations

[adbridge]

@gpsimenos @evedon Do we assume the force push has addressed all the review comments ?
@kjbracey-arm could you re-review please ?

[gpsimenos]

@gpsimenos @evedon Do we assume the force push has addressed all the review comments ?
@kjbracey-arm could you re-review please ?

I believe I have addressed everything.

[evedon]

The review comments have been addressed and the last commit is fixing absolute paths in greentea tests.

[0xc0170]

CI started

[mbed-ci]

Test run: SUCCESS

Summary: 11 of 11 test jobs passed
Build number : 1
Build artifacts