diff --git a/book/en/src/by-example/delay.md b/book/en/src/by-example/delay.md index 8d05d7ce01..f2863633ec 100644 --- a/book/en/src/by-example/delay.md +++ b/book/en/src/by-example/delay.md @@ -6,13 +6,14 @@ This can be achieved by instantiating a monotonic timer: ``` rust ... -rtic_monotonics::make_systick_timer_queue!(TIMER); +rtic_monotonics::make_systick_handler!(); #[init] -fn init(cx: init::Context) -> (Shared, Local) { - let systick = Systick::start(cx.core.SYST, 12_000_000); - TIMER.initialize(systick); - ... +fn init(cx: init::Context) -> (Shared, Local) { + hprintln!("init"); + + Systick::start(cx.core.SYST, 12_000_000); + ... ``` A *software* task can `await` the delay to expire: @@ -21,8 +22,10 @@ A *software* task can `await` the delay to expire: #[task] async fn foo(_cx: foo::Context) { ... - TIMER.delay(100.millis()).await; + Systick::delay(100.millis()).await; ... +} + ``` Technically, the timer queue is implemented as a list based priority queue, where list-nodes are statically allocated as part of the underlying task `Future`. Thus, the timer queue is infallible at run-time (its size and allocation is determined at compile time). @@ -51,21 +54,21 @@ A common use case is transactions with associated timeout. In the examples shown Using the `select_biased` macro from the `futures` crate it may look like this: -```rust +``` rust // Call hal with short relative timeout using `select_biased` select_biased! { - v = hal_get(&TIMER, 1).fuse() => hprintln!("hal returned {}", v), - _ = TIMER.delay(200.millis()).fuse() => hprintln!("timeout", ), // this will finish first + v = hal_get(1).fuse() => hprintln!("hal returned {}", v), + _ = Systick::delay(200.millis()).fuse() => hprintln!("timeout", ), // this will finish first } ``` Assuming the `hal_get` will take 450ms to finish, a short timeout of 200ms will expire. -```rust +``` rust // Call hal with long relative timeout using `select_biased` select_biased! { - v = hal_get(&TIMER, 1).fuse() => hprintln!("hal returned {}", v), // hal finish first - _ = TIMER.delay(1000.millis()).fuse() => hprintln!("timeout", ), + v = hal_get(1).fuse() => hprintln!("hal returned {}", v), // hal finish first + _ = Systick::delay(1000.millis()).fuse() => hprintln!("timeout", ), } ``` @@ -73,9 +76,9 @@ By extending the timeout to 1000ms, the `hal_get` will finish first. Using `select_biased` any number of futures can be combined, so its very powerful. However, as the timeout pattern is frequently used, it is directly supported by the RTIC [rtc-monotonics] and [rtic-time] crates. The second example from above using `timeout_after`: -```rust +``` rust // Call hal with long relative timeout using monotonic `timeout_after` -match TIMER.timeout_after(1000.millis(), hal_get(&TIMER, 1)).await { +match Systick::timeout_after(1000.millis(), hal_get(1)).await { Ok(v) => hprintln!("hal returned {}", v), _ => hprintln!("timeout"), } @@ -85,28 +88,28 @@ In cases you want exact control over time without drift. For this purpose we can [fugit]: https://crates.io/crates/fugit -```rust +``` rust // get the current time instance -let mut instant = TIMER.now(); +let mut instant = Systick::now(); // do this 3 times for n in 0..3 { - // exact point in time without drift + // absolute point in time without drift instant += 1000.millis(); - TIMER.delay_until(instant).await; + Systick::delay_until(instant).await; - // exact point it time for timeout + // absolute point it time for timeout let timeout = instant + 500.millis(); - hprintln!("now is {:?}, timeout at {:?}", TIMER.now(), timeout); + hprintln!("now is {:?}, timeout at {:?}", Systick::now(), timeout); - match TIMER.timeout_at(timeout, hal_get(&TIMER, n)).await { - Ok(v) => hprintln!("hal returned {} at time {:?}", v, TIMER.now()), + match Systick::timeout_at(timeout, hal_get(n)).await { + Ok(v) => hprintln!("hal returned {} at time {:?}", v, Systick::now()), _ => hprintln!("timeout"), } } ``` -`instant = TIMER.now()` gives the baseline (i.e., the exact current point in time). We want to call `hal_get` after 1000ms relative to this exact point in time. This can be accomplished by `TIMER.delay_until(instant).await;`. We define the absolute point in time for the `timeout`, and call `TIMER.timeout_at(timeout, hal_get(&TIMER, n)).await`. For the first loop iteration `n == 0`, and the `hal_get` will take 350ms (and finishes before the timeout). For the second iteration `n == 1`, and `hal_get` will take 450ms (and again succeeds to finish before the timeout). For the third iteration `n == 2` (`hal_get` will take 5500ms to finish). In this case we will run into a timeout. +`instant = Systick::now()` gives the baseline (i.e., the absolute current point in time). We want to call `hal_get` after 1000ms relative to this absolute point in time. This can be accomplished by `Systick::delay_until(instant).await;`. We define the absolute point in time for the `timeout`, and call `Systick::timeout_at(timeout, hal_get(n)).await`. For the first loop iteration `n == 0`, and the `hal_get` will take 350ms (and finishes before the timeout). For the second iteration `n == 1`, and `hal_get` will take 450ms (and again succeeds to finish before the timeout). For the third iteration `n == 2` (`hal_get` will take 5500ms to finish). In this case we will run into a timeout. The complete example: diff --git a/rtic/examples/async-timeout.rs b/rtic/examples/async-timeout.rs index fda385aae5..cdccb3a428 100644 --- a/rtic/examples/async-timeout.rs +++ b/rtic/examples/async-timeout.rs @@ -61,11 +61,11 @@ mod app { // do this 3 times for n in 0..3 { - // exact point in time without drift + // absolute point in time without drift instant += 1000.millis(); Systick::delay_until(instant).await; - // exact point it time for timeout + // absolute point it time for timeout let timeout = instant + 500.millis(); hprintln!("now is {:?}, timeout at {:?}", Systick::now(), timeout);