allow rtos idle loop to call sleep();

[maclobdell]

Description

To enable power efficiency, the rtos loop should call sleep(). To my knowledge, this can be done regardless if the debugger interface is connected (at least it can on the Kinetis MCUs). If the debugger interface is connected, the device may not achieve as great of power savings, but it still can enter low power mode. Something is better than nothing.

Entering low power mode when idle is fundamental to low power / battery powered applications.

Status

Needs CI

Migrations

NO

Related PRs

Same as #2443. Submitting this separate PR due to unknown/unrelated issue preventing the last one to pass CI.

branch	PR

N/A

Todos

N/A

Deploy notes

N/A

Steps to test or reproduce

When RTOS idle executes, it will call sleep(), which is implementation specific to the platform. In general, sleep is a very light low power mode, thus should not disrupt anything.

[bulislaw]

I think it wouldn't hurt us and can save some power even if we wake up every 1ms. We can replace it with tickles mode when we finally get our story together.

[bridadan]

/morph test

[mbed-bot]

Result: FAILURE

Your command has finished executing! Here's what you wrote!

/morph test

Output

mbed Build Number: 0

Build failed!

[bridadan]

Looks like there were some issues with the RZ_A1H and VK_RZ_A1H:

11:10:12 Building project stats (RZ_A1H, ARM)
11:10:12 Scan: ARM
11:10:12 Scan: FEATURE_LWIP
11:10:12 Scan: stats
11:10:12 Compile [100.0%]: main.cpp
11:10:12 Link: stats
11:10:14 Error: L6218E: Undefined symbol sleep (referred from rtos_idle.o).
11:10:14 Finished: 0 information, 0 warning and 1 error messages.

[bulislaw]

Ah, sleep is not mandatory. You should add #if DEVICE_SLEEP around it.

[0xc0170]

Ah, sleep is not mandatory. You should add #if DEVICE_SLEEP around it.

@maclobdell Can you respond? And rebase if possible

[sg-]

@maclobdell can you make changes based on feedback?

[sg-]

@maclobdell please reopen when feedback from comments are addressed.