[p2] fixes INTERRUPTS_01_isisr_willpreempt_servicedirqn test

[scott-brust]

Problem

INTERRUPTS_01_isisr_willpreempt_servicedirqn test fails on P2 due to unimplemented HAL

Solution

Use an array to backup any overridden ISR handlers. When override is removed, restore the backup.

There are other parts of the HAL that are still unimplemented, namely HAL_Core_Restore_Interrupt() and the link_interrupt_vectors_location RAM table in general during HAL_Core_Config().

This PR is still a WIP to facilitate discussion on how to approach these problems.

Steps to Test

Run wiring/no_fixture, specifically INTERRUPTS_01_isisr_willpreempt_servicedirqn

Example App

Run wiring/no_fixture

References

Links to the Community, Docs, Other Issues, etc..

---

Completeness

• User is totes amazing for contributing!
• Contributor has signed CLA (Info here)
• Problem and Solution clearly stated
• Run unit/integration/application tests on device
• Added documentation
• Added to CHANGELOG.md after merging (add links to docs and issues)

[avtolstoy]

This is a bit fragile in the fact that you can call this with handler=nullptr and HAL_INTERRUPT_DIRECT_FLAG_RESTORE when there is no backed-up handler and it will nuke the existing entry in VTOR. Consider a few approaches:

1. Back up entire VTOR on first use if `!memcmp(hal_interrupts_handler_backup, 0, sizeof(hal_interrupts_handler_backup))
2. Perhaps use some struct for backup instead and store the state along with backed-up ISR function

[XuGuohui]

Or considering use std::pair(irqn, handler) along with vector().

[avtolstoy]

I'd avoid using vector here due to dynamic allocations. VTOR size is known and negligible, so we can keep the whole thing preallocated.

[scott-brust]

Is there ever a valid use case to have a NULL interrupt handler? Can I just check if the backed up handler is not null, and only restore it if something has been saved?

[scott-brust]

It might actually simplify the logic if I use goto to bail out rather than have to check for error conditions? I cant return here because I need to re-enable interrupts.

[avtolstoy]

Use CHECK_TRUE/CHECK_FALSE then, much better than goto, so:

CHECK_FALSE(hal_interrupts_handler_backup[IRQN_TO_IDX(irqn)], SYSTEM_ERROR_ALREADY_EXISTS);

[scott-brust]

Ah yes, those macros are handy. Unfortunately I need to re-enable interrupts before exiting. ill just do that and return SYSTEM_ERROR_ALREADY_EXISTS. I think thats cleaner than adding a return variable and checking it, etc.

[avtolstoy]

Use CHECK_TRUE/CHECK_FALSE then, much better than goto, so:

CHECK_FALSE(hal_interrupts_handler_backup[IRQN_TO_IDX(irqn)], SYSTEM_ERROR_ALREADY_EXISTS);