Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Full support for entering and exiting DORMANT mode #701

Merged
merged 13 commits into from
Mar 2, 2024

Conversation

andrewh42
Copy link
Contributor

I noticed that there were the beginnings of some support for the RP2040 dormant lowest-power sleep but that several loose ends remained.

I've taken care not to make any breaking changes in this pull request to maximise its prospects. However, I think an interesting future change would be to add an addiitional pll::State to represent a powered down PLL. This would streamline the process of restarting the PLLs following dormancy.

Looking forward to your feedback and hopefully an approved pull request. I'm keen to hear of any areas that don't meet rp-rs standards or where I've done something that isn't idiomatic - apologies in advance for any crimes against these!

Andrew

Copy link
Member

@ithinuel ithinuel left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you!

I had started doing things in that direction but never got to complete it 🤦.
Because the Dormant mode doesn't really need to be represented (the device does not run while in dormant mode).

I think Initialized and Dormant can be merged into a Unstable state.

You can find my WIP here: https://github.com/ithinuel/rp-hal/tree/clock_dormant_recovery

@andrewh42
Copy link
Contributor Author

andrewh42 commented Oct 20, 2023

Thanks @ithinuel, that's a good suggestion. I've replaced the XOSC Initialized and Dormant states with an Unstable state. And ROSC no longer changes state on after waking.

I couldn’t see the benefit of adding StableOscillatorTokenAlreadyIssuedError however, given that await_stabilization() appears to be idempotent, and the token can only be redeemed once via get_stable(), regardless of how many times we request a token. Would you be able to help me understand what I’m missing here please?

@andrewh42 andrewh42 requested a review from ithinuel October 20, 2023 07:43
@andrewh42
Copy link
Contributor Author

Hi @ithinuel, is there anything I can do to help progress this PR?

@jannic
Copy link
Member

jannic commented Nov 14, 2023

I never used the dormant state and didn't think about it enough to decide if these changes are sufficient to make dormant mode usable. From reading the diffs I get the impression that the changes make sense, and I don't see anything obviously wrong. @ithinuel, as you already worked on your own implementation of dormant support, you are probably more qualified to review this merge request. But if you don't have time to do a review, I could plug together some test setup on a breadboard to try out the example, and approve the PR if everything works as expected.

@andrewh42, you wrote that you avoided breaking changes to make merging easier. As the next release probably will include breaking changes anyway, are there any improvements you'd want to add?

@andrewh42
Copy link
Contributor Author

@andrewh42, you wrote that you avoided breaking changes to make merging easier. As the next release probably will include breaking changes anyway, are there any improvements you'd want to add?

Thanks for asking @jannic. I was thinking it could be useful to add an addiitional pll::State to represent a powered down PLL. At present the PLL's disable() reconstructs the Disabled state by reading the PLL's registers, so that initialize() can be used to restart the disabled PLL. The alternative I have in mind is to replace disable() with, say, stop() and return a new Stopped state that then requires less work to restart. initialize() might then return a Stopped PLL, in order to keep the code DRY. Giving us the following state transitions when starting up:

Disabled -> Stopped -> Locking -> Locked

And these transitions when stopping and then restarting (e.g. as part of going dormant):

Locked -> Stopped -> Locking -> Locked

Does that sound reasonable? Either way, I'd prefer to do this in a separate PR so this one doesn't stick around for too long and gather more complexity. Does that sound OK?

@andrewh42 andrewh42 requested a review from jannic November 15, 2023 10:53
rp2040-hal/src/xosc.rs Show resolved Hide resolved
rp2040-hal/src/rosc.rs Outdated Show resolved Hide resolved
@andrewh42
Copy link
Contributor Author

Hi @ithinuel and @jannic, is there anything I can do to help complete this PR?

@jannic
Copy link
Member

jannic commented Mar 1, 2024

Hi @ithinuel and @jannic, is there anything I can do to help complete this PR?

Sorry for not answering sooner. Currently, my focus is on getting 0.10.0 out the door, so I neglected some older pull requests.

Currently, there are some merge conflicts. Could you try to resolve them?

@andrewh42
Copy link
Contributor Author

andrewh42 commented Mar 2, 2024

Hi @ithinuel and @jannic, is there anything I can do to help complete this PR?

Sorry for not answering sooner. Currently, my focus is on getting 0.10.0 out the door, so I neglected some older pull requests.

Currently, there are some merge conflicts. Could you try to resolve them?

No worries @jannic, done.

Copy link
Member

@jannic jannic left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks, Andrew! I think the pull request can be merged now. @ithinuel, any objections?

@jannic
Copy link
Member

jannic commented Mar 2, 2024

BTW I also verified that the dormant mode actually makes a difference. Even with a cheap USB power meter, I see 0.02A in wfi mode, while in dormant mode, it shows 0.00A. The absolute numbers might be very inaccurate, but it's obvious that the dormant mode is different from the wfi sleep.

@ithinuel
Copy link
Member

ithinuel commented Mar 2, 2024

I'm happy with this. We can always come back later to this if we need to make improvements.
I have a ppk2 at home so I could take some samples and post screenshots here.

@jannic jannic merged commit 1000a10 into rp-rs:main Mar 2, 2024
6 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants