HRT common code ( and derivatives like Tunes library implementation HRT uses)

[davids5]

Now that there are 4 chip architectures stm32, stm32F7, Kinetis and samv7

There is a need to isolate the stm32 arch specific code from the arch agnostic code for drivers (hrt in this case) and derivatives. Such as Tunes library implementation .

[PX4BuildBot]

Hey, this issue has been closed because the label status/STALE is set and there were no updates for 30 days. Feel free to reopen this issue if you deem it appropriate.

(This is an automated comment from GitMate.io.)

[simonegu]

This still an issue.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[davids5]

We need a pass on the htr to do some clean up and document all the interrupt states in the call tree. then build the unified driver with arch specifics like the tone_alarm.

fyi: @mcsauder see https://github.com/PX4/Firmware/pull/11351 for the 32 bit timer version.

I think this needs a critical section

hrt_call_delay(struct hrt_call *entry, hrt_abstime delay)
{
	entry->deadline = hrt_absolute_time() + delay;
}

There is also this that makes no sense

/**
 * Store the absolute time in an interrupt-safe fashion
 */
hrt_abstime
hrt_store_absolute_time(volatile hrt_abstime *now)
{
	irqstate_t flags = px4_enter_critical_section();

	hrt_abstime ts = hrt_absolute_time();

	px4_leave_critical_section(flags);

	return ts;
}

@LorenzMeier, @bkueng, @dagar how would you feel about changing the new hrt_elapsed_time_atomic to be done differently or removed because hrt_absolute_time already has has a critical section. then: hrt_absolute_time(void) should become _hrt_absolute_time(const volatile hrt_abstime *offset) hrt_absolute_time(void) wraps and calls _hrt_absolute_time(nullptr)

Then hrt_elapsed_time_atomic(const volatile hrt_abstime *then) will use _hrt_absolute_time(then) and it its critical section if (my_offset~==nullptr) do the math and return that value.

The additional overhead is 1 pointer null check and one register load for the call with null ptr and for the call with an offset it is just the math as opposed to 2 cs prologues and epilogues which is many more instructions that add not value.

[bkueng]

@davids5 given that hrt_elapsed_time_atomic is almost never used, I don't think it's worth optimizing. On the other hand hrt_absolute_time is used all over the place, called at high frequency, and therefore worth optimizing every bit.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.

[junwoo091400]

Is it still relevant @davids5 ?

[davids5]

Yes. It is harder to separate it out