[i2c] (re)introduce async.await implementation #747

ithinuel · 2024-01-01T14:03:55Z

This re-introduces the async.await i2c driver that was once remove because of requiring nightly at the time.

Now that the necessary features have been stabilized, we can all enjoy a bit of async.await-ness :)

Requires

rp2040-hal/Cargo.toml

rp2040-hal/src/async.rs

rp2040-hal/src/i2c.rs

ithinuel · 2024-01-01T18:40:26Z

This reminded me that we do not have doc nor the ability to really use IRQ based application for i2c. I shall fix that too in an upcoming PR.

thejpster · 2024-01-01T19:32:38Z

rp2040-hal/examples/i2c_async.rs

+/// if your board has a different frequency
+const XTAL_FREQ_HZ: u32 = 12_000_000u32;
+
+/// The function configures the RP2040 peripherals, reads the temperature from


It does not, in fact, do this.

rp2040-hal/examples/i2c_async_irq.rs

thejpster · 2024-01-01T19:36:31Z

rp2040-hal/examples/i2c_async_irq.rs

+            .m_rx_full()
+            .enabled()
+    });
+    critical_section::with(|cs| WAKER.borrow_ref_mut(cs).take().map(Waker::wake));


What does this do? Can we write it out as multiple lines with intermediate variable names and some comments?

I think I'd also prefer if let Some(waker) = borrowed_global_waker.take() { Waker::wake(waker) } or similar, even if clippy might say "yeah, but you can just use .map(Waker::wake)", because it's non-obvious where why you would grab a global and then map it to a different type, and then drop the value.

jannic · 2024-01-01T23:26:44Z

rp2040-hal/src/i2c/controller/eh1_0_async.rs

+    async fn write_iter_async<A, U>(&mut self, address: A, bytes: U) -> Result<(), Error>
+    where
+        U: IntoIterator<Item = u8>,
+        A: AddressMode + 'static + Into<u16>,


I wonder if we should try to keep inherent implementations free from embedded-hal defined types. If we use them in our own API, it'll be difficult to upgrade embedded-hal later.

We did that in a few places with embedded-hal 0.2, and it'll make moving everything to embedded-hal 1.0 more difficult.

Also, we do not even support 10bit address mode yet, afaik.

rather than the inherent method, should I impl the crate I made to cover the write_iter that was removed between 0.2.7 and 1.0 ?

Doesn't that complicate things further? It would mean that users of those methods would need to depend on your crate as well and import the traits, before they can call the functions.

Indeed. I wish it hadn't been removed from the embedded-hal in the first place :/

I'll make this driver impl the trait and keep this dependency feature-gated.

rp2040-hal/examples/i2c_async.rs

rp2040-hal/examples/i2c_async_irq.rs

rp2040-hal/src/i2c.rs

thejpster · 2024-01-06T12:28:30Z

I'm happy once jannic is happy

rp2040-hal/src/i2c/controller/eh1_0_async.rs

rp2040-hal/examples/i2c_async_irq.rs

thejpster · 2024-01-13T11:44:46Z

If this is draft because we have questions around the async-bits, maybe we can update the non-async stuff to be e-h 1.0 compatible in another PR and then rebase this over it? I'm keen to get e-h 1.0 support out of the door because I have other projects stuck until that happens.

ithinuel · 2024-01-13T14:17:58Z

This is in draft because I'm updating i2c implementation as I'm also adding on-board-test suites for controller/target sync/async.
Please, let's proceed with #753 first and I'll rebase this PR on main after it's merged.

ithinuel · 2024-01-31T06:11:55Z

Finally it is there !!

.github/workflows/build_and_test.yml

jannic · 2024-02-01T22:03:28Z

.github/workflows/clippy.yml

-      - run: |
-          cd on-target-tests
-          cargo clippy -- -Dwarnings
+      - run: cargo clippy --workspace --examples


Why not -Dwarnings?

It is already set in the RUSTFLAGS, but I may be confused.
Aren’t they redundant?

rp2040-hal/src/async_utils.rs

rp2040-hal/examples/i2c.rs

rp2040-hal/src/async_utils.rs

jannic · 2024-02-05T15:50:31Z

rp2040-hal/src/i2c/controller.rs

    }

-    fn read_and_clear_abort_reason(&mut self) -> Option<u32> {
+    #[inline]
+    fn read_and_clear_abort_reason(&mut self) -> Result<(), Error> {


The combination of function naming and return type causes some ambiguity: Intuitively, when a function called read_and_clear_abort_reason returns an abort reason, I'd call it a success, not an error.

Yeah, this is mostly to save some boiler plate code as there’s only Option::ok_or and no Option::err_or (Changing the Some reason into a Result::Err).

A better name could be used for this. Maybe check_abort_reason or simply check_abort ?

What about check_and_clear_abort_reason?
Making it obvious that the abort reason is cleared isn't wrong, so it can be kept in the name. And check_ doesn't imply that getting an abort reason is a successful result, in the same way as read_ does.

rp2040-hal/src/i2c/peripheral.rs

jannic · 2024-02-06T08:06:28Z

IMHO this pull request contains unrelated changes that could be discussed and merged separately.

Changes that could be separate pull request include:

unconditionally including defmt.x in the linker options
adding on-target-tests to the top-level workspace
some cleanups to the CI pipeline

Or am I missing something and these changes are inherently connected with the async i2c support?
(That's part of the problem: Because these changes are part of a bigger pull request, they are missing proper reasoning. If they were separate, there would probably be an explanation in the description of the pull request.)

They have been updated and now build fine

ithinuel · 2024-02-06T23:24:24Z

@jannic I understand. I created a PR for each commit (and address the previous reviews)

jannic · 2024-02-07T07:15:28Z

@jannic I understand. I created a PR for each commit (and address the previous reviews)

Thank you!

BTW, as many review comments I made sound so negative: That's just because I'm actively looking for things that I don't particularly like. Overall, you did a great job, and I'm sorry that it takes me so long to complete the review. It's just a lot of changes, and I'm not experienced enough with async rust that I can quickly glance over the changes to understand what's going on.

jannic · 2024-02-09T22:05:29Z

on-target-tests/tests/i2c_loopback.rs

+//! | from GPIO (pico Pin) | to GPIO (pico Pin) |
+//! | -------------------- | ------------------ |
+//! |         0 (1)        |       2 (4)        |
+//! |         1 (2)        |       3 (5)        |


Please also mention this in on-target-tests/README.md. It already describes the connection necessary for the SPI test, so the I2C connections should be there as well.

jannic

I would be lying if I said I understood every detail of this pull request.
But as far as I understand it, it looks good.

As github shows all the changes in the pull request anyway, even if some of the commits were actually already merged by some other pull requests, I did not review #763 and #764 separately. Not sure what's the best way to review multiple changes individually, short of having completely separate branches.

Also adds "-Tdefmt.x" to .cargo/config. This required for i2c async examples & on-target-tests. Note this does not affect downstream dependents.

ithinuel · 2024-02-10T01:13:02Z

If merging this PR does not auto close the two "preceding", I’ll just close them manually adding the reference to this PR in the comment.
I don’t really mind about the details as long as it’s deemed worthy to be merged and it makes it to main :)

ithinuel requested review from jannic and thejpster January 1, 2024 14:03

ithinuel force-pushed the introduce-async-i2c branch from 36a709b to 66bd3f1 Compare January 1, 2024 16:06

jannic reviewed Jan 1, 2024

View reviewed changes

rp2040-hal/Cargo.toml Show resolved Hide resolved

rp2040-hal/src/async.rs Outdated Show resolved Hide resolved

rp2040-hal/src/async.rs Outdated Show resolved Hide resolved

rp2040-hal/src/i2c.rs Outdated Show resolved Hide resolved

thejpster reviewed Jan 1, 2024

View reviewed changes

rp2040-hal/examples/i2c_async_irq.rs Outdated Show resolved Hide resolved

thejpster reviewed Jan 1, 2024

View reviewed changes

jannic reviewed Jan 1, 2024

View reviewed changes

jannic reviewed Jan 4, 2024

View reviewed changes

rp2040-hal/examples/i2c_async.rs Outdated Show resolved Hide resolved

rp2040-hal/examples/i2c_async_irq.rs Outdated Show resolved Hide resolved

rp2040-hal/examples/i2c_async_irq.rs Outdated Show resolved Hide resolved

rp2040-hal/src/i2c.rs Outdated Show resolved Hide resolved

jannic reviewed Jan 6, 2024

View reviewed changes

rp2040-hal/src/i2c/controller/eh1_0_async.rs Outdated Show resolved Hide resolved

rp2040-hal/src/i2c/controller/eh1_0_async.rs Outdated Show resolved Hide resolved

ithinuel force-pushed the introduce-async-i2c branch from b6d2c27 to 95064ff Compare January 6, 2024 19:13

jannic reviewed Jan 6, 2024

View reviewed changes

rp2040-hal/src/i2c/controller/eh1_0_async.rs Outdated Show resolved Hide resolved

jannic reviewed Jan 7, 2024

View reviewed changes

rp2040-hal/examples/i2c_async_irq.rs Outdated Show resolved Hide resolved

ithinuel marked this pull request as draft January 9, 2024 06:53

ithinuel force-pushed the introduce-async-i2c branch from 643d281 to 38b110e Compare January 9, 2024 06:54

jannic mentioned this pull request Jan 10, 2024

Make embedded-hal 1.0.0 non-optional #753

Merged

ithinuel force-pushed the introduce-async-i2c branch from 38b110e to 408c434 Compare January 13, 2024 01:06

ithinuel force-pushed the introduce-async-i2c branch from 408c434 to 38bf821 Compare January 18, 2024 17:17

jannic mentioned this pull request Jan 19, 2024

Clean up the I2C impls. #758

Open

ithinuel force-pushed the introduce-async-i2c branch 7 times, most recently from 2c4d76b to 1edeee8 Compare January 29, 2024 23:23

ithinuel force-pushed the introduce-async-i2c branch 7 times, most recently from e6d3616 to ef6a686 Compare January 31, 2024 06:03

ithinuel marked this pull request as ready for review January 31, 2024 06:10

ithinuel requested review from thejpster and jannic January 31, 2024 06:11

jannic reviewed Feb 5, 2024

View reviewed changes

Bump cargo-hack & cargo-udeps versions & nightly for cargo-udeps

2c451ac

They have been updated and now build fine

ithinuel force-pushed the introduce-async-i2c branch from ef6a686 to 2003212 Compare February 6, 2024 22:56

ithinuel mentioned this pull request Feb 6, 2024

[i2c] Implement i2c-write-iter traits #765

Merged

ithinuel force-pushed the introduce-async-i2c branch 2 times, most recently from 40e95d5 to be9b0ab Compare February 6, 2024 23:17

ithinuel changed the title ~~(re)introduce async i2c.~~ [i2c] (re)introduce async.await implementation Feb 6, 2024

ithinuel force-pushed the introduce-async-i2c branch from be9b0ab to abc7fb8 Compare February 8, 2024 23:04

jannic reviewed Feb 9, 2024

View reviewed changes

jannic approved these changes Feb 9, 2024

View reviewed changes

ithinuel added 2 commits February 9, 2024 23:43

i2c: add on-target-tests & fix discovered issues

7d1bd63

i2c: implement embedded-hal-async with tests

7a47574

Also adds "-Tdefmt.x" to .cargo/config. This required for i2c async examples & on-target-tests. Note this does not affect downstream dependents.

ithinuel force-pushed the introduce-async-i2c branch from abc7fb8 to 7a47574 Compare February 9, 2024 23:44

jannic merged commit 07ba575 into rp-rs:main Feb 10, 2024
6 checks passed

ithinuel deleted the introduce-async-i2c branch February 10, 2024 13:41

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[i2c] (re)introduce async.await implementation #747

[i2c] (re)introduce async.await implementation #747

ithinuel commented Jan 1, 2024 •

edited

Loading

ithinuel commented Jan 1, 2024

thejpster Jan 1, 2024

thejpster Jan 1, 2024

thejpster Jan 1, 2024 •

edited

Loading

jannic Jan 1, 2024

ithinuel Jan 6, 2024

jannic Jan 6, 2024

ithinuel Jan 7, 2024

thejpster commented Jan 6, 2024

thejpster commented Jan 13, 2024

ithinuel commented Jan 13, 2024 •

edited

Loading

ithinuel commented Jan 31, 2024

jannic Feb 1, 2024

ithinuel Feb 5, 2024

jannic Feb 5, 2024

ithinuel Feb 5, 2024

jannic Feb 9, 2024

jannic commented Feb 6, 2024

ithinuel commented Feb 6, 2024

jannic commented Feb 7, 2024

jannic Feb 9, 2024

jannic left a comment

ithinuel commented Feb 10, 2024

[i2c] (re)introduce async.await implementation #747

[i2c] (re)introduce async.await implementation #747

Conversation

ithinuel commented Jan 1, 2024 • edited Loading

Requires

ithinuel commented Jan 1, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

thejpster Jan 1, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

thejpster commented Jan 6, 2024

thejpster commented Jan 13, 2024

ithinuel commented Jan 13, 2024 • edited Loading

ithinuel commented Jan 31, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jannic commented Feb 6, 2024

ithinuel commented Feb 6, 2024

jannic commented Feb 7, 2024

Choose a reason for hiding this comment

jannic left a comment

Choose a reason for hiding this comment

ithinuel commented Feb 10, 2024

ithinuel commented Jan 1, 2024 •

edited

Loading

thejpster Jan 1, 2024 •

edited

Loading

ithinuel commented Jan 13, 2024 •

edited

Loading