Rebase on master

76 files changed, 1894 insertions, 464 deletions

diff --git a/embassy-embedded-hal/Cargo.toml b/embassy-embedded-hal/Cargo.toml
index d0be6d195..85ee856a6 100644
--- a/embassy-embedded-hal/Cargo.toml
+++ b/embassy-embedded-hal/Cargo.toml
@@ -20,7 +20,7 @@ nightly = ["embedded-hal-async", "embedded-storage-async"]
 embassy-sync = { version = "0.1.0", path = "../embassy-sync" }
 embedded-hal-02 = { package = "embedded-hal", version = "0.2.6", features = ["unproven"] }
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "=0.1.0-alpha.2", optional = true }
+embedded-hal-async = { version = "=0.1.0-alpha.3", optional = true }
 embedded-storage = "0.3.0"
 embedded-storage-async = { version = "0.3.0", optional = true }
 nb = "1.0.0"
diff --git a/embassy-executor/src/raw/mod.rs b/embassy-executor/src/raw/mod.rs
index e1258ebb5..181dabe8e 100644
--- a/embassy-executor/src/raw/mod.rs
+++ b/embassy-executor/src/raw/mod.rs
@@ -354,46 +354,54 @@ impl Executor {
     /// somehow schedule for `poll()` to be called later, at a time you know for sure there's
     /// no `poll()` already running.
     pub unsafe fn poll(&'static self) {
-        #[cfg(feature = "integrated-timers")]
-        self.timer_queue.dequeue_expired(Instant::now(), |task| wake_task(task));
+        loop {
+            #[cfg(feature = "integrated-timers")]
+            self.timer_queue.dequeue_expired(Instant::now(), |task| wake_task(task));
 
-        self.run_queue.dequeue_all(|p| {
-            let task = p.as_ref();
+            self.run_queue.dequeue_all(|p| {
+                let task = p.as_ref();
 
-            #[cfg(feature = "integrated-timers")]
-            task.expires_at.set(Instant::MAX);
-
-            let state = task.state.fetch_and(!STATE_RUN_QUEUED, Ordering::AcqRel);
-            if state & STATE_SPAWNED == 0 {
-                // If task is not running, ignore it. This can happen in the following scenario:
-                //   - Task gets dequeued, poll starts
-                //   - While task is being polled, it gets woken. It gets placed in the queue.
-                //   - Task poll finishes, returning done=true
-                //   - RUNNING bit is cleared, but the task is already in the queue.
-                return;
-            }
+                #[cfg(feature = "integrated-timers")]
+                task.expires_at.set(Instant::MAX);
 
-            #[cfg(feature = "rtos-trace")]
-            trace::task_exec_begin(p.as_ptr() as u32);
+                let state = task.state.fetch_and(!STATE_RUN_QUEUED, Ordering::AcqRel);
+                if state & STATE_SPAWNED == 0 {
+                    // If task is not running, ignore it. This can happen in the following scenario:
+                    //   - Task gets dequeued, poll starts
+                    //   - While task is being polled, it gets woken. It gets placed in the queue.
+                    //   - Task poll finishes, returning done=true
+                    //   - RUNNING bit is cleared, but the task is already in the queue.
+                    return;
+                }
 
-            // Run the task
-            task.poll_fn.read()(p as _);
+                #[cfg(feature = "rtos-trace")]
+                trace::task_exec_begin(p.as_ptr() as u32);
 
-            #[cfg(feature = "rtos-trace")]
-            trace::task_exec_end();
+                // Run the task
+                task.poll_fn.read()(p as _);
+
+                #[cfg(feature = "rtos-trace")]
+                trace::task_exec_end();
+
+                // Enqueue or update into timer_queue
+                #[cfg(feature = "integrated-timers")]
+                self.timer_queue.update(p);
+            });
 
-            // Enqueue or update into timer_queue
             #[cfg(feature = "integrated-timers")]
-            self.timer_queue.update(p);
-        });
+            {
+                // If this is already in the past, set_alarm might return false
+                // In that case do another poll loop iteration.
+                let next_expiration = self.timer_queue.next_expiration();
+                if driver::set_alarm(self.alarm, next_expiration.as_ticks()) {
+                    break;
+                }
+            }
 
-        #[cfg(feature = "integrated-timers")]
-        {
-            // If this is already in the past, set_alarm will immediately trigger the alarm.
-            // This will cause `signal_fn` to be called, which will cause `poll()` to be called again,
-            // so we immediately do another poll loop iteration.
-            let next_expiration = self.timer_queue.next_expiration();
-            driver::set_alarm(self.alarm, next_expiration.as_ticks());
+            #[cfg(not(feature = "integrated-timers"))]
+            {
+                break;
+            }
         }
 
         #[cfg(feature = "rtos-trace")]
@@ -436,14 +444,21 @@ pub unsafe fn wake_task(task: NonNull<TaskHeader>) {
 }
 
 #[cfg(feature = "integrated-timers")]
-#[no_mangle]
-unsafe fn _embassy_time_schedule_wake(at: Instant, waker: &core::task::Waker) {
-    let task = waker::task_from_waker(waker);
-    let task = task.as_ref();
-    let expires_at = task.expires_at.get();
-    task.expires_at.set(expires_at.min(at));
+struct TimerQueue;
+
+#[cfg(feature = "integrated-timers")]
+impl embassy_time::queue::TimerQueue for TimerQueue {
+    fn schedule_wake(&'static self, at: Instant, waker: &core::task::Waker) {
+        let task = waker::task_from_waker(waker);
+        let task = unsafe { task.as_ref() };
+        let expires_at = task.expires_at.get();
+        task.expires_at.set(expires_at.min(at));
+    }
 }
 
+#[cfg(feature = "integrated-timers")]
+embassy_time::timer_queue_impl!(static TIMER_QUEUE: TimerQueue = TimerQueue);
+
 #[cfg(feature = "rtos-trace")]
 impl rtos_trace::RtosTraceOSCallbacks for Executor {
     fn task_list() {
diff --git a/embassy-lora/Cargo.toml b/embassy-lora/Cargo.toml
index ea2c3fe67..dc2004172 100644
--- a/embassy-lora/Cargo.toml
+++ b/embassy-lora/Cargo.toml
@@ -32,7 +32,7 @@ embassy-time = { version = "0.1.0", path = "../embassy-time" }
 embassy-sync = { version = "0.1.0", path = "../embassy-sync" }
 embassy-stm32 = { version = "0.1.0", path = "../embassy-stm32", default-features = false, optional = true }
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "=0.1.0-alpha.2" }
+embedded-hal-async = { version = "=0.1.0-alpha.3" }
 embassy-hal-common = { version = "0.1.0", path = "../embassy-hal-common", default-features = false }
 futures = { version = "0.3.17", default-features = false, features = [ "async-await" ] }
 embedded-hal = { version = "0.2", features = ["unproven"] }
diff --git a/embassy-net/Cargo.toml b/embassy-net/Cargo.toml
index 967ef26a7..76217075a 100644
--- a/embassy-net/Cargo.toml
+++ b/embassy-net/Cargo.toml
@@ -42,7 +42,7 @@ log = { version = "0.4.14", optional = true }
 
 embassy-time = { version = "0.1.0", path = "../embassy-time" }
 embassy-sync = { version = "0.1.0", path = "../embassy-sync" }
-embedded-io = { version = "0.3.0", optional = true }
+embedded-io = { version = "0.3.1", optional = true }
 
 managed = { version = "0.8.0", default-features = false, features = [ "map" ] }
 heapless = { version = "0.7.5", default-features = false }
diff --git a/embassy-net/src/tcp.rs b/embassy-net/src/tcp.rs
index f8fff3e2d..f3bd2361c 100644
--- a/embassy-net/src/tcp.rs
+++ b/embassy-net/src/tcp.rs
@@ -292,7 +292,7 @@ mod embedded_io_impls {
     }
 
     impl<'d> embedded_io::asynch::Read for TcpSocket<'d> {
-        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -302,7 +302,7 @@ mod embedded_io_impls {
     }
 
     impl<'d> embedded_io::asynch::Write for TcpSocket<'d> {
-        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -310,7 +310,7 @@ mod embedded_io_impls {
             self.io.write(buf)
         }
 
-        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -324,7 +324,7 @@ mod embedded_io_impls {
     }
 
     impl<'d> embedded_io::asynch::Read for TcpReader<'d> {
-        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -338,7 +338,7 @@ mod embedded_io_impls {
     }
 
     impl<'d> embedded_io::asynch::Write for TcpWriter<'d> {
-        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -346,7 +346,7 @@ mod embedded_io_impls {
             self.io.write(buf)
         }
 
-        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -445,7 +445,7 @@ pub mod client {
     impl<'d, const N: usize, const TX_SZ: usize, const RX_SZ: usize> embedded_io::asynch::Read
         for TcpConnection<'d, N, TX_SZ, RX_SZ>
     {
-        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -457,7 +457,7 @@ pub mod client {
     impl<'d, const N: usize, const TX_SZ: usize, const RX_SZ: usize> embedded_io::asynch::Write
         for TcpConnection<'d, N, TX_SZ, RX_SZ>
     {
-        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -465,7 +465,7 @@ pub mod client {
             self.socket.write(buf)
         }
 
-        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
diff --git a/embassy-nrf/Cargo.toml b/embassy-nrf/Cargo.toml
index 5459bc90c..67b6bec40 100644
--- a/embassy-nrf/Cargo.toml
+++ b/embassy-nrf/Cargo.toml
@@ -75,8 +75,8 @@ embassy-usb-driver = {version = "0.1.0", path = "../embassy-usb-driver", optiona
 
 embedded-hal-02 = { package = "embedded-hal", version = "0.2.6", features = ["unproven"] }
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9", optional = true}
-embedded-hal-async = { version = "=0.1.0-alpha.2", optional = true}
-embedded-io = { version = "0.3.0", features = ["async"], optional = true }
+embedded-hal-async = { version = "=0.1.0-alpha.3", optional = true}
+embedded-io = { version = "0.3.1", features = ["async"], optional = true }
 
 defmt = { version = "0.3", optional = true }
 log = { version = "0.4.14", optional = true }
@@ -90,13 +90,13 @@ embedded-storage = "0.3.0"
 embedded-storage-async = { version = "0.3.0", optional = true }
 cfg-if = "1.0.0"
 
-nrf52805-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf52810-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf52811-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf52820-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf52832-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf52833-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf52840-pac  = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf5340-app-pac = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf5340-net-pac = { version = "0.11.0", optional = true, features = [ "rt" ] }
-nrf9160-pac = { version = "0.11.0", optional = true, features = [ "rt" ] }
+nrf52805-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf52810-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf52811-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf52820-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf52832-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf52833-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf52840-pac  = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf5340-app-pac = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf5340-net-pac = { version = "0.12.0", optional = true, features = [ "rt" ] }
+nrf9160-pac = { version = "0.12.0", optional = true, features = [ "rt" ] }
diff --git a/embassy-nrf/src/buffered_uarte.rs b/embassy-nrf/src/buffered_uarte.rs
index 6e85a159f..9c8fe65f4 100644
--- a/embassy-nrf/src/buffered_uarte.rs
+++ b/embassy-nrf/src/buffered_uarte.rs
@@ -341,7 +341,7 @@ impl<'u, 'd, U: UarteInstance, T: TimerInstance> embedded_io::Io for BufferedUar
 }
 
 impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Read for BufferedUarte<'d, U, T> {
-    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -351,7 +351,7 @@ impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Read for Buffe
 }
 
 impl<'u, 'd: 'u, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Read for BufferedUarteRx<'u, 'd, U, T> {
-    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -361,7 +361,7 @@ impl<'u, 'd: 'u, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Read f
 }
 
 impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::BufRead for BufferedUarte<'d, U, T> {
-    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>>
+    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -375,7 +375,7 @@ impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::BufRead for Bu
 }
 
 impl<'u, 'd: 'u, U: UarteInstance, T: TimerInstance> embedded_io::asynch::BufRead for BufferedUarteRx<'u, 'd, U, T> {
-    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>>
+    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -389,7 +389,7 @@ impl<'u, 'd: 'u, U: UarteInstance, T: TimerInstance> embedded_io::asynch::BufRea
 }
 
 impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Write for BufferedUarte<'d, U, T> {
-    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -397,7 +397,7 @@ impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Write for Buff
         self.inner_write(buf)
     }
 
-    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -407,7 +407,7 @@ impl<'d, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Write for Buff
 }
 
 impl<'u, 'd: 'u, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Write for BufferedUarteTx<'u, 'd, U, T> {
-    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -415,7 +415,7 @@ impl<'u, 'd: 'u, U: UarteInstance, T: TimerInstance> embedded_io::asynch::Write
         self.inner.inner_write(buf)
     }
 
-    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
diff --git a/embassy-nrf/src/chips/nrf52805.rs b/embassy-nrf/src/chips/nrf52805.rs
index d078fa0ad..dec31a84c 100644
--- a/embassy-nrf/src/chips/nrf52805.rs
+++ b/embassy-nrf/src/chips/nrf52805.rs
@@ -193,8 +193,8 @@ impl_ppi_channel!(PPI_CH29, 29 => static);
 impl_ppi_channel!(PPI_CH30, 30 => static);
 impl_ppi_channel!(PPI_CH31, 31 => static);
 
-impl_saadc_input!(P0_04, ANALOGINPUT2);
-impl_saadc_input!(P0_05, ANALOGINPUT3);
+impl_saadc_input!(P0_04, ANALOG_INPUT2);
+impl_saadc_input!(P0_05, ANALOG_INPUT3);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf52810.rs b/embassy-nrf/src/chips/nrf52810.rs
index 3e500098c..e57a4a383 100644
--- a/embassy-nrf/src/chips/nrf52810.rs
+++ b/embassy-nrf/src/chips/nrf52810.rs
@@ -211,14 +211,14 @@ impl_ppi_channel!(PPI_CH29, 29 => static);
 impl_ppi_channel!(PPI_CH30, 30 => static);
 impl_ppi_channel!(PPI_CH31, 31 => static);
 
-impl_saadc_input!(P0_02, ANALOGINPUT0);
-impl_saadc_input!(P0_03, ANALOGINPUT1);
-impl_saadc_input!(P0_04, ANALOGINPUT2);
-impl_saadc_input!(P0_05, ANALOGINPUT3);
-impl_saadc_input!(P0_28, ANALOGINPUT4);
-impl_saadc_input!(P0_29, ANALOGINPUT5);
-impl_saadc_input!(P0_30, ANALOGINPUT6);
-impl_saadc_input!(P0_31, ANALOGINPUT7);
+impl_saadc_input!(P0_02, ANALOG_INPUT0);
+impl_saadc_input!(P0_03, ANALOG_INPUT1);
+impl_saadc_input!(P0_04, ANALOG_INPUT2);
+impl_saadc_input!(P0_05, ANALOG_INPUT3);
+impl_saadc_input!(P0_28, ANALOG_INPUT4);
+impl_saadc_input!(P0_29, ANALOG_INPUT5);
+impl_saadc_input!(P0_30, ANALOG_INPUT6);
+impl_saadc_input!(P0_31, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf52811.rs b/embassy-nrf/src/chips/nrf52811.rs
index 25c7c0d91..918404cf1 100644
--- a/embassy-nrf/src/chips/nrf52811.rs
+++ b/embassy-nrf/src/chips/nrf52811.rs
@@ -212,14 +212,14 @@ impl_ppi_channel!(PPI_CH29, 29 => static);
 impl_ppi_channel!(PPI_CH30, 30 => static);
 impl_ppi_channel!(PPI_CH31, 31 => static);
 
-impl_saadc_input!(P0_02, ANALOGINPUT0);
-impl_saadc_input!(P0_03, ANALOGINPUT1);
-impl_saadc_input!(P0_04, ANALOGINPUT2);
-impl_saadc_input!(P0_05, ANALOGINPUT3);
-impl_saadc_input!(P0_28, ANALOGINPUT4);
-impl_saadc_input!(P0_29, ANALOGINPUT5);
-impl_saadc_input!(P0_30, ANALOGINPUT6);
-impl_saadc_input!(P0_31, ANALOGINPUT7);
+impl_saadc_input!(P0_02, ANALOG_INPUT0);
+impl_saadc_input!(P0_03, ANALOG_INPUT1);
+impl_saadc_input!(P0_04, ANALOG_INPUT2);
+impl_saadc_input!(P0_05, ANALOG_INPUT3);
+impl_saadc_input!(P0_28, ANALOG_INPUT4);
+impl_saadc_input!(P0_29, ANALOG_INPUT5);
+impl_saadc_input!(P0_30, ANALOG_INPUT6);
+impl_saadc_input!(P0_31, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf52832.rs b/embassy-nrf/src/chips/nrf52832.rs
index 18b8eda67..81e66c193 100644
--- a/embassy-nrf/src/chips/nrf52832.rs
+++ b/embassy-nrf/src/chips/nrf52832.rs
@@ -225,14 +225,14 @@ impl_ppi_channel!(PPI_CH29, 29 => static);
 impl_ppi_channel!(PPI_CH30, 30 => static);
 impl_ppi_channel!(PPI_CH31, 31 => static);
 
-impl_saadc_input!(P0_02, ANALOGINPUT0);
-impl_saadc_input!(P0_03, ANALOGINPUT1);
-impl_saadc_input!(P0_04, ANALOGINPUT2);
-impl_saadc_input!(P0_05, ANALOGINPUT3);
-impl_saadc_input!(P0_28, ANALOGINPUT4);
-impl_saadc_input!(P0_29, ANALOGINPUT5);
-impl_saadc_input!(P0_30, ANALOGINPUT6);
-impl_saadc_input!(P0_31, ANALOGINPUT7);
+impl_saadc_input!(P0_02, ANALOG_INPUT0);
+impl_saadc_input!(P0_03, ANALOG_INPUT1);
+impl_saadc_input!(P0_04, ANALOG_INPUT2);
+impl_saadc_input!(P0_05, ANALOG_INPUT3);
+impl_saadc_input!(P0_28, ANALOG_INPUT4);
+impl_saadc_input!(P0_29, ANALOG_INPUT5);
+impl_saadc_input!(P0_30, ANALOG_INPUT6);
+impl_saadc_input!(P0_31, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf52833.rs b/embassy-nrf/src/chips/nrf52833.rs
index 3b33907d2..92499e3c9 100644
--- a/embassy-nrf/src/chips/nrf52833.rs
+++ b/embassy-nrf/src/chips/nrf52833.rs
@@ -271,14 +271,14 @@ impl_ppi_channel!(PPI_CH29, 29 => static);
 impl_ppi_channel!(PPI_CH30, 30 => static);
 impl_ppi_channel!(PPI_CH31, 31 => static);
 
-impl_saadc_input!(P0_02, ANALOGINPUT0);
-impl_saadc_input!(P0_03, ANALOGINPUT1);
-impl_saadc_input!(P0_04, ANALOGINPUT2);
-impl_saadc_input!(P0_05, ANALOGINPUT3);
-impl_saadc_input!(P0_28, ANALOGINPUT4);
-impl_saadc_input!(P0_29, ANALOGINPUT5);
-impl_saadc_input!(P0_30, ANALOGINPUT6);
-impl_saadc_input!(P0_31, ANALOGINPUT7);
+impl_saadc_input!(P0_02, ANALOG_INPUT0);
+impl_saadc_input!(P0_03, ANALOG_INPUT1);
+impl_saadc_input!(P0_04, ANALOG_INPUT2);
+impl_saadc_input!(P0_05, ANALOG_INPUT3);
+impl_saadc_input!(P0_28, ANALOG_INPUT4);
+impl_saadc_input!(P0_29, ANALOG_INPUT5);
+impl_saadc_input!(P0_30, ANALOG_INPUT6);
+impl_saadc_input!(P0_31, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf52840.rs b/embassy-nrf/src/chips/nrf52840.rs
index ae59f8b25..4beadfba8 100644
--- a/embassy-nrf/src/chips/nrf52840.rs
+++ b/embassy-nrf/src/chips/nrf52840.rs
@@ -276,14 +276,14 @@ impl_ppi_channel!(PPI_CH29, 29 => static);
 impl_ppi_channel!(PPI_CH30, 30 => static);
 impl_ppi_channel!(PPI_CH31, 31 => static);
 
-impl_saadc_input!(P0_02, ANALOGINPUT0);
-impl_saadc_input!(P0_03, ANALOGINPUT1);
-impl_saadc_input!(P0_04, ANALOGINPUT2);
-impl_saadc_input!(P0_05, ANALOGINPUT3);
-impl_saadc_input!(P0_28, ANALOGINPUT4);
-impl_saadc_input!(P0_29, ANALOGINPUT5);
-impl_saadc_input!(P0_30, ANALOGINPUT6);
-impl_saadc_input!(P0_31, ANALOGINPUT7);
+impl_saadc_input!(P0_02, ANALOG_INPUT0);
+impl_saadc_input!(P0_03, ANALOG_INPUT1);
+impl_saadc_input!(P0_04, ANALOG_INPUT2);
+impl_saadc_input!(P0_05, ANALOG_INPUT3);
+impl_saadc_input!(P0_28, ANALOG_INPUT4);
+impl_saadc_input!(P0_29, ANALOG_INPUT5);
+impl_saadc_input!(P0_30, ANALOG_INPUT6);
+impl_saadc_input!(P0_31, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf5340_app.rs b/embassy-nrf/src/chips/nrf5340_app.rs
index edf800ef3..7845d4a8e 100644
--- a/embassy-nrf/src/chips/nrf5340_app.rs
+++ b/embassy-nrf/src/chips/nrf5340_app.rs
@@ -458,14 +458,14 @@ impl_ppi_channel!(PPI_CH29, 29 => configurable);
 impl_ppi_channel!(PPI_CH30, 30 => configurable);
 impl_ppi_channel!(PPI_CH31, 31 => configurable);
 
-impl_saadc_input!(P0_13, ANALOGINPUT0);
-impl_saadc_input!(P0_14, ANALOGINPUT1);
-impl_saadc_input!(P0_15, ANALOGINPUT2);
-impl_saadc_input!(P0_16, ANALOGINPUT3);
-impl_saadc_input!(P0_17, ANALOGINPUT4);
-impl_saadc_input!(P0_18, ANALOGINPUT5);
-impl_saadc_input!(P0_19, ANALOGINPUT6);
-impl_saadc_input!(P0_20, ANALOGINPUT7);
+impl_saadc_input!(P0_13, ANALOG_INPUT0);
+impl_saadc_input!(P0_14, ANALOG_INPUT1);
+impl_saadc_input!(P0_15, ANALOG_INPUT2);
+impl_saadc_input!(P0_16, ANALOG_INPUT3);
+impl_saadc_input!(P0_17, ANALOG_INPUT4);
+impl_saadc_input!(P0_18, ANALOG_INPUT5);
+impl_saadc_input!(P0_19, ANALOG_INPUT6);
+impl_saadc_input!(P0_20, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/chips/nrf9160.rs b/embassy-nrf/src/chips/nrf9160.rs
index f8ed11e03..b5a53ed80 100644
--- a/embassy-nrf/src/chips/nrf9160.rs
+++ b/embassy-nrf/src/chips/nrf9160.rs
@@ -339,14 +339,14 @@ impl_ppi_channel!(PPI_CH13, 13 => configurable);
 impl_ppi_channel!(PPI_CH14, 14 => configurable);
 impl_ppi_channel!(PPI_CH15, 15 => configurable);
 
-impl_saadc_input!(P0_13, ANALOGINPUT0);
-impl_saadc_input!(P0_14, ANALOGINPUT1);
-impl_saadc_input!(P0_15, ANALOGINPUT2);
-impl_saadc_input!(P0_16, ANALOGINPUT3);
-impl_saadc_input!(P0_17, ANALOGINPUT4);
-impl_saadc_input!(P0_18, ANALOGINPUT5);
-impl_saadc_input!(P0_19, ANALOGINPUT6);
-impl_saadc_input!(P0_20, ANALOGINPUT7);
+impl_saadc_input!(P0_13, ANALOG_INPUT0);
+impl_saadc_input!(P0_14, ANALOG_INPUT1);
+impl_saadc_input!(P0_15, ANALOG_INPUT2);
+impl_saadc_input!(P0_16, ANALOG_INPUT3);
+impl_saadc_input!(P0_17, ANALOG_INPUT4);
+impl_saadc_input!(P0_18, ANALOG_INPUT5);
+impl_saadc_input!(P0_19, ANALOG_INPUT6);
+impl_saadc_input!(P0_20, ANALOG_INPUT7);
 
 pub mod irqs {
     use embassy_cortex_m::interrupt::_export::declare;
diff --git a/embassy-nrf/src/time_driver.rs b/embassy-nrf/src/time_driver.rs
index c32a44637..bc2c8a3c1 100644
--- a/embassy-nrf/src/time_driver.rs
+++ b/embassy-nrf/src/time_driver.rs
@@ -243,21 +243,24 @@ impl Driver for RtcDriver {
         })
     }
 
-    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) {
+    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) -> bool {
         critical_section::with(|cs| {
             let n = alarm.id() as _;
             let alarm = self.get_alarm(cs, alarm);
             alarm.timestamp.set(timestamp);
 
-            let t = self.now();
+            let r = rtc();
 
-            // If alarm timestamp has passed, trigger it instantly.
+            let t = self.now();
             if timestamp <= t {
-                self.trigger_alarm(n, cs);
-                return;
-            }
+                // If alarm timestamp has passed the alarm will not fire.
+                // Disarm the alarm and return `false` to indicate that.
+                r.intenclr.write(|w| unsafe { w.bits(compare_n(n)) });
 
-            let r = rtc();
+                alarm.timestamp.set(u64::MAX);
+
+                return false;
+            }
 
             // If it hasn't triggered yet, setup it in the compare channel.
 
@@ -287,6 +290,8 @@ impl Driver for RtcDriver {
                 // It will be setup later by `next_period`.
                 r.intenclr.write(|w| unsafe { w.bits(compare_n(n)) });
             }
+
+            true
         })
     }
 }
diff --git a/embassy-nrf/src/usb.rs b/embassy-nrf/src/usb.rs
index 00da5c9dd..eb1472fa5 100644
--- a/embassy-nrf/src/usb.rs
+++ b/embassy-nrf/src/usb.rs
@@ -313,7 +313,7 @@ impl<'d, T: Instance, P: UsbSupply> driver::Bus for Bus<'d, T, P> {
                 }
             })
             .await;
-            regs.eventcause.write(|w| w.ready().set_bit()); // Write 1 to clear.
+            regs.eventcause.write(|w| w.ready().clear_bit_by_one());
 
             errata::post_enable();
 
@@ -367,24 +367,24 @@ impl<'d, T: Instance, P: UsbSupply> driver::Bus for Bus<'d, T, P> {
             let r = regs.eventcause.read();
 
             if r.isooutcrc().bit() {
-                regs.eventcause.write(|w| w.isooutcrc().set_bit());
+                regs.eventcause.write(|w| w.isooutcrc().detected());
                 trace!("USB event: isooutcrc");
             }
             if r.usbwuallowed().bit() {
-                regs.eventcause.write(|w| w.usbwuallowed().set_bit());
+                regs.eventcause.write(|w| w.usbwuallowed().allowed());
                 trace!("USB event: usbwuallowed");
             }
             if r.suspend().bit() {
-                regs.eventcause.write(|w| w.suspend().set_bit());
+                regs.eventcause.write(|w| w.suspend().detected());
                 regs.lowpower.write(|w| w.lowpower().low_power());
                 return Poll::Ready(Event::Suspend);
             }
             if r.resume().bit() {
-                regs.eventcause.write(|w| w.resume().set_bit());
+                regs.eventcause.write(|w| w.resume().detected());
                 return Poll::Ready(Event::Resume);
             }
             if r.ready().bit() {
-                regs.eventcause.write(|w| w.ready().set_bit());
+                regs.eventcause.write(|w| w.ready().ready());
                 trace!("USB event: ready");
             }
 
@@ -512,7 +512,7 @@ impl<'d, T: Instance, P: UsbSupply> driver::Bus for Bus<'d, T, P> {
                     } else if r.resume().bit() {
                         Poll::Ready(())
                     } else if r.usbwuallowed().bit() {
-                        regs.eventcause.write(|w| w.usbwuallowed().set_bit());
+                        regs.eventcause.write(|w| w.usbwuallowed().allowed());
 
                         regs.dpdmvalue.write(|w| w.state().resume());
                         regs.tasks_dpdmdrive.write(|w| w.tasks_dpdmdrive().set_bit());
diff --git a/embassy-rp/Cargo.toml b/embassy-rp/Cargo.toml
index 2f9c9b9af..04b0c13ce 100644
--- a/embassy-rp/Cargo.toml
+++ b/embassy-rp/Cargo.toml
@@ -53,13 +53,13 @@ cortex-m = "0.7.6"
 critical-section = "1.1"
 futures = { version = "0.3.17", default-features = false, features = ["async-await"] }
 chrono = { version = "0.4", default-features = false, optional = true }
+embedded-io = { version = "0.3.1", features = ["async"], optional = true }
 embedded-storage = { version = "0.3" }
-embedded-io = { version = "0.3.0", features = ["async"], optional = true }
 
 rp2040-pac2 = { git = "https://github.com/embassy-rs/rp2040-pac2", rev="017e3c9007b2d3b6965f0d85b5bf8ce3fa6d7364", features = ["rt"] }
 #rp2040-pac2 = { path = "../../rp2040-pac2", features = ["rt"] }
 
 embedded-hal-02 = { package = "embedded-hal", version = "0.2.6", features = ["unproven"] }
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9", optional = true}
-embedded-hal-async = { version = "=0.1.0-alpha.2", optional = true}
+embedded-hal-async = { version = "=0.1.0-alpha.3", optional = true}
 embedded-hal-nb = { version = "=1.0.0-alpha.1", optional = true}
diff --git a/embassy-rp/src/timer.rs b/embassy-rp/src/timer.rs
index 5215c0c0f..80efd779f 100644
--- a/embassy-rp/src/timer.rs
+++ b/embassy-rp/src/timer.rs
@@ -68,7 +68,7 @@ impl Driver for TimerDriver {
         })
     }
 
-    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) {
+    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) -> bool {
         let n = alarm.id() as usize;
         critical_section::with(|cs| {
             let alarm = &self.alarms.borrow(cs)[n];
@@ -81,11 +81,16 @@ impl Driver for TimerDriver {
             unsafe { pac::TIMER.alarm(n).write_value(timestamp as u32) };
 
             let now = self.now();
-
-            // If alarm timestamp has passed, trigger it instantly.
-            // This disarms it.
             if timestamp <= now {
-                self.trigger_alarm(n, cs);
+                // If alarm timestamp has passed the alarm will not fire.
+                // Disarm the alarm and return `false` to indicate that.
+                unsafe { pac::TIMER.armed().write(|w| w.set_armed(1 << n)) }
+
+                alarm.timestamp.set(u64::MAX);
+
+                false
+            } else {
+                true
             }
         })
     }
diff --git a/embassy-rp/src/uart/buffered.rs b/embassy-rp/src/uart/buffered.rs
index 87e16f0eb..4f0a55532 100644
--- a/embassy-rp/src/uart/buffered.rs
+++ b/embassy-rp/src/uart/buffered.rs
@@ -355,7 +355,7 @@ impl<'d, T: Instance> embedded_io::Io for BufferedUartTx<'d, T> {
 }
 
 impl<'d, T: Instance + 'd> embedded_io::asynch::Read for BufferedUart<'d, T> {
-    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -376,7 +376,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::Read for BufferedUart<'d, T> {
 }
 
 impl<'d, T: Instance + 'd> embedded_io::asynch::Read for BufferedUartRx<'d, T> {
-    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -397,7 +397,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::Read for BufferedUartRx<'d, T> {
 }
 
 impl<'d, T: Instance + 'd> embedded_io::asynch::BufRead for BufferedUart<'d, T> {
-    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>>
+    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -419,7 +419,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::BufRead for BufferedUart<'d, T>
 }
 
 impl<'d, T: Instance + 'd> embedded_io::asynch::BufRead for BufferedUartRx<'d, T> {
-    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>>
+    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -441,7 +441,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::BufRead for BufferedUartRx<'d, T
 }
 
 impl<'d, T: Instance + 'd> embedded_io::asynch::Write for BufferedUart<'d, T> {
-    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -455,7 +455,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::Write for BufferedUart<'d, T> {
         })
     }
 
-    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -465,7 +465,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::Write for BufferedUart<'d, T> {
 }
 
 impl<'d, T: Instance + 'd> embedded_io::asynch::Write for BufferedUartTx<'d, T> {
-    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -479,7 +479,7 @@ impl<'d, T: Instance + 'd> embedded_io::asynch::Write for BufferedUartTx<'d, T>
         })
     }
 
-    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
diff --git a/embassy-stm32/Cargo.toml b/embassy-stm32/Cargo.toml
index 9194ae788..6b00518a6 100644
--- a/embassy-stm32/Cargo.toml
+++ b/embassy-stm32/Cargo.toml
@@ -44,7 +44,7 @@ embassy-usb-driver = {version = "0.1.0", path = "../embassy-usb-driver", optiona
 
 embedded-hal-02 = { package = "embedded-hal", version = "0.2.6", features = ["unproven"] }
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9", optional = true}
-embedded-hal-async = { version = "=0.1.0-alpha.2", optional = true}
+embedded-hal-async = { version = "=0.1.0-alpha.3", optional = true}
 embedded-hal-nb = { version = "=1.0.0-alpha.1", optional = true}
 
 embedded-storage = "0.3.0"
@@ -67,7 +67,7 @@ nb = "1.0.0"
 stm32-fmc = "0.2.4"
 seq-macro = "0.3.0"
 cfg-if = "1.0.0"
-embedded-io = { version = "0.3.0", features = ["async"], optional = true }
+embedded-io = { version = "0.3.1", features = ["async"], optional = true }
 
 [build-dependencies]
 proc-macro2 = "1.0.36"
@@ -82,9 +82,12 @@ memory-x = ["stm32-metapac/memory-x"]
 subghz = []
 exti = []
 
+# Enables additional driver features that depend on embassy-time
+time = ["dep:embassy-time"]
+
 # Features starting with `_` are for internal use only. They're not intended
 # to be enabled by other crates, and are not covered by semver guarantees.
-_time-driver = ["dep:embassy-time"]
+_time-driver = ["time"]
 time-driver-any = ["_time-driver"]
 time-driver-tim2 = ["_time-driver"]
 time-driver-tim3 = ["_time-driver"]
diff --git a/embassy-stm32/src/adc/f1.rs b/embassy-stm32/src/adc/f1.rs
index 50d4f9bf9..c5b317ce9 100644
--- a/embassy-stm32/src/adc/f1.rs
+++ b/embassy-stm32/src/adc/f1.rs
@@ -86,7 +86,6 @@ pub use sample_time::SampleTime;
 
 pub struct Adc<'d, T: Instance> {
     sample_time: SampleTime,
-    calibrated_vdda: u32,
     phantom: PhantomData<&'d mut T>,
 }
 
@@ -122,7 +121,6 @@ impl<'d, T: Instance> Adc<'d, T> {
 
         Self {
             sample_time: Default::default(),
-            calibrated_vdda: VDDA_CALIB_MV,
             phantom: PhantomData,
         }
     }
@@ -162,29 +160,10 @@ impl<'d, T: Instance> Adc<'d, T> {
         Temperature {}
     }
 
-    /// Calculates the system VDDA by sampling the internal VREF channel and comparing
-    /// to the expected value. If the chip's VDDA is not stable, run this before each ADC
-    /// conversion.
-    pub fn calibrate(&mut self, vref: &mut Vref) -> u32 {
-        let old_sample_time = self.sample_time;
-        self.sample_time = SampleTime::Cycles239_5;
-
-        let vref_samp = self.read(vref);
-        self.sample_time = old_sample_time;
-
-        self.calibrated_vdda = (ADC_MAX * VREF_INT) / u32::from(vref_samp);
-        self.calibrated_vdda
-    }
-
     pub fn set_sample_time(&mut self, sample_time: SampleTime) {
         self.sample_time = sample_time;
     }
 
-    /// Convert a measurement to millivolts
-    pub fn to_millivolts(&self, sample: u16) -> u16 {
-        ((u32::from(sample) * self.calibrated_vdda) / ADC_MAX) as u16
-    }
-
     /// Perform a single conversion.
     fn convert(&mut self) -> u16 {
         unsafe {
diff --git a/embassy-stm32/src/adc/v2.rs b/embassy-stm32/src/adc/v2.rs
index 4fe4ad1f0..53419c7f2 100644
--- a/embassy-stm32/src/adc/v2.rs
+++ b/embassy-stm32/src/adc/v2.rs
@@ -80,15 +80,6 @@ impl super::sealed::InternalChannel<ADC1> for Temperature {
 }
 
 impl Temperature {
-    /// Converts temperature sensor reading in millivolts to degrees celcius
-    pub fn to_celcius(sample_mv: u16) -> f32 {
-        // From 6.3.22 Temperature sensor characteristics
-        const V25: i32 = 760; // mV
-        const AVG_SLOPE: f32 = 2.5; // mV/C
-
-        (sample_mv as i32 - V25) as f32 / AVG_SLOPE + 25.0
-    }
-
     /// Time needed for temperature sensor readings to stabilize
     pub fn start_time_us() -> u32 {
         10
@@ -172,7 +163,6 @@ impl Prescaler {
 
 pub struct Adc<'d, T: Instance> {
     sample_time: SampleTime,
-    vref_mv: u32,
     resolution: Resolution,
     phantom: PhantomData<&'d mut T>,
 }
@@ -200,7 +190,6 @@ where
         Self {
             sample_time: Default::default(),
             resolution: Resolution::default(),
-            vref_mv: VREF_DEFAULT_MV,
             phantom: PhantomData,
         }
     }
@@ -213,18 +202,6 @@ where
         self.resolution = resolution;
     }
 
-    /// Set VREF value in millivolts. This value is used for [to_millivolts()] sample conversion.
-    ///
-    /// Use this if you have a known precise VREF (VDDA) pin reference voltage.
-    pub fn set_vref_mv(&mut self, vref_mv: u32) {
-        self.vref_mv = vref_mv;
-    }
-
-    /// Convert a measurement to millivolts
-    pub fn to_millivolts(&self, sample: u16) -> u16 {
-        ((u32::from(sample) * self.vref_mv) / self.resolution.to_max_count()) as u16
-    }
-
     /// Enables internal voltage reference and returns [VrefInt], which can be used in
     /// [Adc::read_internal()] to perform conversion.
     pub fn enable_vrefint(&self) -> VrefInt {
diff --git a/embassy-stm32/src/adc/v3.rs b/embassy-stm32/src/adc/v3.rs
index 0f1090888..816feeac7 100644
--- a/embassy-stm32/src/adc/v3.rs
+++ b/embassy-stm32/src/adc/v3.rs
@@ -205,7 +205,6 @@ pub use sample_time::SampleTime;
 
 pub struct Adc<'d, T: Instance> {
     sample_time: SampleTime,
-    vref_mv: u32,
     resolution: Resolution,
     phantom: PhantomData<&'d mut T>,
 }
@@ -244,7 +243,6 @@ impl<'d, T: Instance> Adc<'d, T> {
         Self {
             sample_time: Default::default(),
             resolution: Resolution::default(),
-            vref_mv: VREF_DEFAULT_MV,
             phantom: PhantomData,
         }
     }
@@ -285,31 +283,6 @@ impl<'d, T: Instance> Adc<'d, T> {
         Vbat {}
     }
 
-    /// Calculates the system VDDA by sampling the internal VREFINT channel and comparing
-    /// the result with the value stored at the factory. If the chip's VDDA is not stable, run
-    /// this before each ADC conversion.
-    #[cfg(not(stm32g0))] // TODO is this supposed to be public?
-    #[allow(unused)] // TODO is this supposed to be public?
-    fn calibrate(&mut self, vrefint: &mut VrefInt) {
-        #[cfg(stm32l5)]
-        let vrefint_cal: u32 = todo!();
-        #[cfg(not(stm32l5))]
-        let vrefint_cal = unsafe { crate::pac::VREFINTCAL.data().read().value() };
-        let old_sample_time = self.sample_time;
-
-        // "Table 24. Embedded internal voltage reference" states that the sample time needs to be
-        // at a minimum 4 us. With 640.5 ADC cycles we have a minimum of 8 us at 80 MHz, leaving
-        // some headroom.
-        self.sample_time = SampleTime::Cycles640_5;
-
-        // This can't actually fail, it's just in a result to satisfy hal trait
-        let vrefint_samp = self.read(vrefint);
-
-        self.sample_time = old_sample_time;
-
-        self.vref_mv = (VREF_CALIB_MV * u32::from(vrefint_cal)) / u32::from(vrefint_samp);
-    }
-
     pub fn set_sample_time(&mut self, sample_time: SampleTime) {
         self.sample_time = sample_time;
     }
@@ -318,18 +291,6 @@ impl<'d, T: Instance> Adc<'d, T> {
         self.resolution = resolution;
     }
 
-    /// Set VREF value in millivolts. This value is used for [to_millivolts()] sample conversion.
-    ///
-    /// Use this if you have a known precise VREF (VDDA) pin reference voltage.
-    pub fn set_vref_mv(&mut self, vref_mv: u32) {
-        self.vref_mv = vref_mv;
-    }
-
-    /// Convert a measurement to millivolts
-    pub fn to_millivolts(&self, sample: u16) -> u16 {
-        ((u32::from(sample) * self.vref_mv) / self.resolution.to_max_count()) as u16
-    }
-
     /*
     /// Convert a raw sample from the `Temperature` to deg C
     pub fn to_degrees_centigrade(sample: u16) -> f32 {
diff --git a/embassy-stm32/src/adc/v4.rs b/embassy-stm32/src/adc/v4.rs
index eda2b2a72..2b8f10533 100644
--- a/embassy-stm32/src/adc/v4.rs
+++ b/embassy-stm32/src/adc/v4.rs
@@ -314,7 +314,6 @@ impl Prescaler {
 
 pub struct Adc<'d, T: Instance> {
     sample_time: SampleTime,
-    vref_mv: u32,
     resolution: Resolution,
     phantom: PhantomData<&'d mut T>,
 }
@@ -352,7 +351,6 @@ impl<'d, T: Instance + crate::rcc::RccPeripheral> Adc<'d, T> {
 
         let mut s = Self {
             sample_time: Default::default(),
-            vref_mv: VREF_DEFAULT_MV,
             resolution: Resolution::default(),
             phantom: PhantomData,
         };
@@ -459,18 +457,6 @@ impl<'d, T: Instance + crate::rcc::RccPeripheral> Adc<'d, T> {
         self.resolution = resolution;
     }
 
-    /// Set VREF value in millivolts. This value is used for [to_millivolts()] sample conversion.
-    ///
-    /// Use this if you have a known precise VREF (VDDA) pin reference voltage.
-    pub fn set_vref_mv(&mut self, vref_mv: u32) {
-        self.vref_mv = vref_mv;
-    }
-
-    /// Convert a measurement to millivolts
-    pub fn to_millivolts(&self, sample: u16) -> u16 {
-        ((u32::from(sample) * self.vref_mv) / self.resolution.to_max_count()) as u16
-    }
-
     /// Perform a single conversion.
     fn convert(&mut self) -> u16 {
         unsafe {
diff --git a/embassy-stm32/src/i2c/mod.rs b/embassy-stm32/src/i2c/mod.rs
index 9d314f411..f898fcc8b 100644
--- a/embassy-stm32/src/i2c/mod.rs
+++ b/embassy-stm32/src/i2c/mod.rs
@@ -7,6 +7,11 @@ use crate::interrupt::Interrupt;
 mod _version;
 pub use _version::*;
 
+#[cfg(feature = "time")]
+mod timeout;
+#[cfg(feature = "time")]
+pub use timeout::*;
+
 use crate::peripherals;
 
 #[derive(Debug)]
diff --git a/embassy-stm32/src/i2c/timeout.rs b/embassy-stm32/src/i2c/timeout.rs
new file mode 100644
index 000000000..4fca1ca2b
--- /dev/null
+++ b/embassy-stm32/src/i2c/timeout.rs
@@ -0,0 +1,142 @@
+use embassy_time::{Duration, Instant};
+
+use super::{Error, I2c, Instance};
+
+/// An I2C wrapper, which provides `embassy-time` based timeouts for all `embedded-hal` trait methods.
+///
+/// This is useful for recovering from a shorted bus or a device stuck in a clock stretching state.
+/// A regular [I2c] would freeze until condition is removed.
+pub struct TimeoutI2c<'d, T: Instance, TXDMA, RXDMA> {
+    i2c: &'d mut I2c<'d, T, TXDMA, RXDMA>,
+    timeout: Duration,
+}
+
+fn timeout_fn(timeout: Duration) -> impl Fn() -> Result<(), Error> {
+    let deadline = Instant::now() + timeout;
+    move || {
+        if Instant::now() > deadline {
+            Err(Error::Timeout)
+        } else {
+            Ok(())
+        }
+    }
+}
+
+impl<'d, T: Instance, TXDMA, RXDMA> TimeoutI2c<'d, T, TXDMA, RXDMA> {
+    pub fn new(i2c: &'d mut I2c<'d, T, TXDMA, RXDMA>, timeout: Duration) -> Self {
+        Self { i2c, timeout }
+    }
+
+    /// Blocking read with a custom timeout
+    pub fn blocking_read_timeout(&mut self, addr: u8, buffer: &mut [u8], timeout: Duration) -> Result<(), Error> {
+        self.i2c.blocking_read_timeout(addr, buffer, timeout_fn(timeout))
+    }
+
+    /// Blocking read with default timeout, provided in [`TimeoutI2c::new()`]
+    pub fn blocking_read(&mut self, addr: u8, buffer: &mut [u8]) -> Result<(), Error> {
+        self.blocking_read_timeout(addr, buffer, self.timeout)
+    }
+
+    /// Blocking write with a custom timeout
+    pub fn blocking_write_timeout(&mut self, addr: u8, bytes: &[u8], timeout: Duration) -> Result<(), Error> {
+        self.i2c.blocking_write_timeout(addr, bytes, timeout_fn(timeout))
+    }
+
+    /// Blocking write with default timeout, provided in [`TimeoutI2c::new()`]
+    pub fn blocking_write(&mut self, addr: u8, bytes: &[u8]) -> Result<(), Error> {
+        self.blocking_write_timeout(addr, bytes, self.timeout)
+    }
+
+    /// Blocking write-read with a custom timeout
+    pub fn blocking_write_read_timeout(
+        &mut self,
+        addr: u8,
+        bytes: &[u8],
+        buffer: &mut [u8],
+        timeout: Duration,
+    ) -> Result<(), Error> {
+        self.i2c
+            .blocking_write_read_timeout(addr, bytes, buffer, timeout_fn(timeout))
+    }
+
+    /// Blocking write-read with default timeout, provided in [`TimeoutI2c::new()`]
+    pub fn blocking_write_read(&mut self, addr: u8, bytes: &[u8], buffer: &mut [u8]) -> Result<(), Error> {
+        self.blocking_write_read_timeout(addr, bytes, buffer, self.timeout)
+    }
+}
+
+impl<'d, T: Instance, TXDMA, RXDMA> embedded_hal_02::blocking::i2c::Read for TimeoutI2c<'d, T, TXDMA, RXDMA> {
+    type Error = Error;
+
+    fn read(&mut self, addr: u8, buffer: &mut [u8]) -> Result<(), Self::Error> {
+        self.blocking_read(addr, buffer)
+    }
+}
+
+impl<'d, T: Instance, TXDMA, RXDMA> embedded_hal_02::blocking::i2c::Write for TimeoutI2c<'d, T, TXDMA, RXDMA> {
+    type Error = Error;
+
+    fn write(&mut self, addr: u8, bytes: &[u8]) -> Result<(), Self::Error> {
+        self.blocking_write(addr, bytes)
+    }
+}
+
+impl<'d, T: Instance, TXDMA, RXDMA> embedded_hal_02::blocking::i2c::WriteRead for TimeoutI2c<'d, T, TXDMA, RXDMA> {
+    type Error = Error;
+
+    fn write_read(&mut self, addr: u8, bytes: &[u8], buffer: &mut [u8]) -> Result<(), Self::Error> {
+        self.blocking_write_read(addr, bytes, buffer)
+    }
+}
+
+#[cfg(feature = "unstable-traits")]
+mod eh1 {
+    use super::*;
+
+    impl<'d, T: Instance, TXDMA, RXDMA> embedded_hal_1::i2c::ErrorType for TimeoutI2c<'d, T, TXDMA, RXDMA> {
+        type Error = Error;
+    }
+
+    impl<'d, T: Instance, TXDMA, RXDMA> embedded_hal_1::i2c::I2c for TimeoutI2c<'d, T, TXDMA, RXDMA> {
+        fn read(&mut self, address: u8, buffer: &mut [u8]) -> Result<(), Self::Error> {
+            self.blocking_read(address, buffer)
+        }
+
+        fn write(&mut self, address: u8, buffer: &[u8]) -> Result<(), Self::Error> {
+            self.blocking_write(address, buffer)
+        }
+
+        fn write_iter<B>(&mut self, _address: u8, _bytes: B) -> Result<(), Self::Error>
+        where
+            B: IntoIterator<Item = u8>,
+        {
+            todo!();
+        }
+
+        fn write_iter_read<B>(&mut self, _address: u8, _bytes: B, _buffer: &mut [u8]) -> Result<(), Self::Error>
+        where
+            B: IntoIterator<Item = u8>,
+        {
+            todo!();
+        }
+
+        fn write_read(&mut self, address: u8, wr_buffer: &[u8], rd_buffer: &mut [u8]) -> Result<(), Self::Error> {
+            self.blocking_write_read(address, wr_buffer, rd_buffer)
+        }
+
+        fn transaction<'a>(
+            &mut self,
+            _address: u8,
+            _operations: &mut [embedded_hal_1::i2c::Operation<'a>],
+        ) -> Result<(), Self::Error> {
+            todo!();
+        }
+
+        fn transaction_iter<'a, O>(&mut self, _address: u8, _operations: O) -> Result<(), Self::Error>
+        where
+            O: IntoIterator<Item = embedded_hal_1::i2c::Operation<'a>>,
+        {
+            todo!();
+        }
+    }
+}
diff --git a/embassy-stm32/src/i2c/v1.rs b/embassy-stm32/src/i2c/v1.rs
index f39a37df6..f140e2b0d 100644
--- a/embassy-stm32/src/i2c/v1.rs
+++ b/embassy-stm32/src/i2c/v1.rs
@@ -1,8 +1,9 @@
 use core::marker::PhantomData;
 
 use embassy_embedded_hal::SetConfig;
-use embassy_hal_common::into_ref;
+use embassy_hal_common::{into_ref, PeripheralRef};
 
+use crate::dma::NoDma;
 use crate::gpio::sealed::AFType;
 use crate::gpio::Pull;
 use crate::i2c::{Error, Instance, SclPin, SdaPin};
@@ -34,19 +35,26 @@ impl State {
     }
 }
 
-pub struct I2c<'d, T: Instance> {
+pub struct I2c<'d, T: Instance, TXDMA = NoDma, RXDMA = NoDma> {
     phantom: PhantomData<&'d mut T>,
+    #[allow(dead_code)]
+    tx_dma: PeripheralRef<'d, TXDMA>,
+    #[allow(dead_code)]
+    rx_dma: PeripheralRef<'d, RXDMA>,
 }
 
-impl<'d, T: Instance> I2c<'d, T> {
+impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
     pub fn new(
         _peri: impl Peripheral<P = T> + 'd,
         scl: impl Peripheral<P = impl SclPin<T>> + 'd,
         sda: impl Peripheral<P = impl SdaPin<T>> + 'd,
+        _irq: impl Peripheral<P = T::Interrupt> + 'd,
+        tx_dma: impl Peripheral<P = TXDMA> + 'd,
+        rx_dma: impl Peripheral<P = RXDMA> + 'd,
         freq: Hertz,
         config: Config,
     ) -> Self {
-        into_ref!(scl, sda);
+        into_ref!(scl, sda, tx_dma, rx_dma);
 
         T::enable();
         T::reset();
@@ -99,7 +107,11 @@ impl<'d, T: Instance> I2c<'d, T> {
             });
         }
 
-        Self { phantom: PhantomData }
+        Self {
+            phantom: PhantomData,
+            tx_dma,
+            rx_dma,
+        }
     }
 
     unsafe fn check_and_clear_error_flags(&self) -> Result<i2c::regs::Sr1, Error> {
@@ -141,7 +153,12 @@ impl<'d, T: Instance> I2c<'d, T> {
         Ok(sr1)
     }
 
-    unsafe fn write_bytes(&mut self, addr: u8, bytes: &[u8]) -> Result<(), Error> {
+    unsafe fn write_bytes(
+        &mut self,
+        addr: u8,
+        bytes: &[u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         // Send a START condition
 
         T::regs().cr1().modify(|reg| {
@@ -149,7 +166,9 @@ impl<'d, T: Instance> I2c<'d, T> {
         });
 
         // Wait until START condition was generated
-        while !self.check_and_clear_error_flags()?.start() {}
+        while !self.check_and_clear_error_flags()?.start() {
+            check_timeout()?;
+        }
 
         // Also wait until signalled we're master and everything is waiting for us
         while {
@@ -157,7 +176,9 @@ impl<'d, T: Instance> I2c<'d, T> {
 
             let sr2 = T::regs().sr2().read();
             !sr2.msl() && !sr2.busy()
-        } {}
+        } {
+            check_timeout()?;
+        }
 
         // Set up current address, we're trying to talk to
         T::regs().dr().write(|reg| reg.set_dr(addr << 1));
@@ -165,26 +186,30 @@ impl<'d, T: Instance> I2c<'d, T> {
         // Wait until address was sent
         // Wait for the address to be acknowledged
         // Check for any I2C errors. If a NACK occurs, the ADDR bit will never be set.
-        while !self.check_and_clear_error_flags()?.addr() {}
+        while !self.check_and_clear_error_flags()?.addr() {
+            check_timeout()?;
+        }
 
         // Clear condition by reading SR2
         let _ = T::regs().sr2().read();
 
         // Send bytes
         for c in bytes {
-            self.send_byte(*c)?;
+            self.send_byte(*c, &check_timeout)?;
         }
 
         // Fallthrough is success
         Ok(())
     }
 
-    unsafe fn send_byte(&self, byte: u8) -> Result<(), Error> {
+    unsafe fn send_byte(&self, byte: u8, check_timeout: impl Fn() -> Result<(), Error>) -> Result<(), Error> {
         // Wait until we're ready for sending
         while {
             // Check for any I2C errors. If a NACK occurs, the ADDR bit will never be set.
             !self.check_and_clear_error_flags()?.txe()
-        } {}
+        } {
+            check_timeout()?;
+        }
 
         // Push out a byte of data
         T::regs().dr().write(|reg| reg.set_dr(byte));
@@ -193,24 +218,33 @@ impl<'d, T: Instance> I2c<'d, T> {
         while {
             // Check for any potential error conditions.
             !self.check_and_clear_error_flags()?.btf()
-        } {}
+        } {
+            check_timeout()?;
+        }
 
         Ok(())
     }
 
-    unsafe fn recv_byte(&self) -> Result<u8, Error> {
+    unsafe fn recv_byte(&self, check_timeout: impl Fn() -> Result<(), Error>) -> Result<u8, Error> {
         while {
             // Check for any potential error conditions.
             self.check_and_clear_error_flags()?;
 
             !T::regs().sr1().read().rxne()
-        } {}
+        } {
+            check_timeout()?;
+        }
 
         let value = T::regs().dr().read().dr();
         Ok(value)
     }
 
-    pub fn blocking_read(&mut self, addr: u8, buffer: &mut [u8]) -> Result<(), Error> {
+    pub fn blocking_read_timeout(
+        &mut self,
+        addr: u8,
+        buffer: &mut [u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         if let Some((last, buffer)) = buffer.split_last_mut() {
             // Send a START condition and set ACK bit
             unsafe {
@@ -221,27 +255,33 @@ impl<'d, T: Instance> I2c<'d, T> {
             }
 
             // Wait until START condition was generated
-            while unsafe { !T::regs().sr1().read().start() } {}
+            while unsafe { !self.check_and_clear_error_flags()?.start() } {
+                check_timeout()?;
+            }
 
             // Also wait until signalled we're master and everything is waiting for us
             while {
                 let sr2 = unsafe { T::regs().sr2().read() };
                 !sr2.msl() && !sr2.busy()
-            } {}
+            } {
+                check_timeout()?;
+            }
 
             // Set up current address, we're trying to talk to
             unsafe { T::regs().dr().write(|reg| reg.set_dr((addr << 1) + 1)) }
 
             // Wait until address was sent
             // Wait for the address to be acknowledged
-            while unsafe { !self.check_and_clear_error_flags()?.addr() } {}
+            while unsafe { !self.check_and_clear_error_flags()?.addr() } {
+                check_timeout()?;
+            }
 
             // Clear condition by reading SR2
             let _ = unsafe { T::regs().sr2().read() };
 
             // Receive bytes into buffer
             for c in buffer {
-                *c = unsafe { self.recv_byte()? };
+                *c = unsafe { self.recv_byte(&check_timeout)? };
             }
 
             // Prepare to send NACK then STOP after next byte
@@ -253,10 +293,12 @@ impl<'d, T: Instance> I2c<'d, T> {
             }
 
             // Receive last byte
-            *last = unsafe { self.recv_byte()? };
+            *last = unsafe { self.recv_byte(&check_timeout)? };
 
             // Wait for the STOP to be sent.
-            while unsafe { T::regs().cr1().read().stop() } {}
+            while unsafe { T::regs().cr1().read().stop() } {
+                check_timeout()?;
+            }
 
             // Fallthrough is success
             Ok(())
@@ -265,25 +307,50 @@ impl<'d, T: Instance> I2c<'d, T> {
         }
     }
 
-    pub fn blocking_write(&mut self, addr: u8, bytes: &[u8]) -> Result<(), Error> {
+    pub fn blocking_read(&mut self, addr: u8, buffer: &mut [u8]) -> Result<(), Error> {
+        self.blocking_read_timeout(addr, buffer, || Ok(()))
+    }
+
+    pub fn blocking_write_timeout(
+        &mut self,
+        addr: u8,
+        bytes: &[u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         unsafe {
-            self.write_bytes(addr, bytes)?;
+            self.write_bytes(addr, bytes, &check_timeout)?;
             // Send a STOP condition
             T::regs().cr1().modify(|reg| reg.set_stop(true));
             // Wait for STOP condition to transmit.
-            while T::regs().cr1().read().stop() {}
+            while T::regs().cr1().read().stop() {
+                check_timeout()?;
+            }
         };
 
         // Fallthrough is success
         Ok(())
     }
 
-    pub fn blocking_write_read(&mut self, addr: u8, bytes: &[u8], buffer: &mut [u8]) -> Result<(), Error> {
-        unsafe { self.write_bytes(addr, bytes)? };
-        self.blocking_read(addr, buffer)?;
+    pub fn blocking_write(&mut self, addr: u8, bytes: &[u8]) -> Result<(), Error> {
+        self.blocking_write_timeout(addr, bytes, || Ok(()))
+    }
+
+    pub fn blocking_write_read_timeout(
+        &mut self,
+        addr: u8,
+        bytes: &[u8],
+        buffer: &mut [u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
+        unsafe { self.write_bytes(addr, bytes, &check_timeout)? };
+        self.blocking_read_timeout(addr, buffer, &check_timeout)?;
 
         Ok(())
     }
+
+    pub fn blocking_write_read(&mut self, addr: u8, bytes: &[u8], buffer: &mut [u8]) -> Result<(), Error> {
+        self.blocking_write_read_timeout(addr, bytes, buffer, || Ok(()))
+    }
 }
 
 impl<'d, T: Instance> embedded_hal_02::blocking::i2c::Read for I2c<'d, T> {
diff --git a/embassy-stm32/src/i2c/v2.rs b/embassy-stm32/src/i2c/v2.rs
index 89b52da98..aa4e6bb08 100644
--- a/embassy-stm32/src/i2c/v2.rs
+++ b/embassy-stm32/src/i2c/v2.rs
@@ -147,14 +147,23 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         }
     }
 
-    unsafe fn master_read(address: u8, length: usize, stop: Stop, reload: bool, restart: bool) {
+    unsafe fn master_read(
+        address: u8,
+        length: usize,
+        stop: Stop,
+        reload: bool,
+        restart: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         assert!(length < 256);
 
         if !restart {
             // Wait for any previous address sequence to end
             // automatically. This could be up to 50% of a bus
             // cycle (ie. up to 0.5/freq)
-            while T::regs().cr2().read().start() {}
+            while T::regs().cr2().read().start() {
+                check_timeout()?;
+            }
         }
 
         // Set START and prepare to receive bytes into
@@ -176,15 +185,25 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             w.set_autoend(stop.autoend());
             w.set_reload(reload);
         });
+
+        Ok(())
     }
 
-    unsafe fn master_write(address: u8, length: usize, stop: Stop, reload: bool) {
+    unsafe fn master_write(
+        address: u8,
+        length: usize,
+        stop: Stop,
+        reload: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         assert!(length < 256);
 
         // Wait for any previous address sequence to end
         // automatically. This could be up to 50% of a bus
         // cycle (ie. up to 0.5/freq)
-        while T::regs().cr2().read().start() {}
+        while T::regs().cr2().read().start() {
+            check_timeout()?;
+        }
 
         let reload = if reload {
             i2c::vals::Reload::NOTCOMPLETED
@@ -204,12 +223,20 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             w.set_autoend(stop.autoend());
             w.set_reload(reload);
         });
+
+        Ok(())
     }
 
-    unsafe fn master_continue(length: usize, reload: bool) {
+    unsafe fn master_continue(
+        length: usize,
+        reload: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         assert!(length < 256 && length > 0);
 
-        while !T::regs().isr().read().tcr() {}
+        while !T::regs().isr().read().tcr() {
+            check_timeout()?;
+        }
 
         let reload = if reload {
             i2c::vals::Reload::NOTCOMPLETED
@@ -221,6 +248,8 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             w.set_nbytes(length as u8);
             w.set_reload(reload);
         });
+
+        Ok(())
     }
 
     fn flush_txdr(&self) {
@@ -243,7 +272,7 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         //}
     }
 
-    fn wait_txe(&self) -> Result<(), Error> {
+    fn wait_txe(&self, check_timeout: impl Fn() -> Result<(), Error>) -> Result<(), Error> {
         loop {
             unsafe {
                 let isr = T::regs().isr().read();
@@ -261,10 +290,12 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                     return Err(Error::Nack);
                 }
             }
+
+            check_timeout()?;
         }
     }
 
-    fn wait_rxne(&self) -> Result<(), Error> {
+    fn wait_rxne(&self, check_timeout: impl Fn() -> Result<(), Error>) -> Result<(), Error> {
         loop {
             unsafe {
                 let isr = T::regs().isr().read();
@@ -282,10 +313,12 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                     return Err(Error::Nack);
                 }
             }
+
+            check_timeout()?;
         }
     }
 
-    fn wait_tc(&self) -> Result<(), Error> {
+    fn wait_tc(&self, check_timeout: impl Fn() -> Result<(), Error>) -> Result<(), Error> {
         loop {
             unsafe {
                 let isr = T::regs().isr().read();
@@ -303,10 +336,18 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                     return Err(Error::Nack);
                 }
             }
+
+            check_timeout()?;
         }
     }
 
-    fn read_internal(&mut self, address: u8, buffer: &mut [u8], restart: bool) -> Result<(), Error> {
+    fn read_internal(
+        &mut self,
+        address: u8,
+        buffer: &mut [u8],
+        restart: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         let completed_chunks = buffer.len() / 255;
         let total_chunks = if completed_chunks * 255 == buffer.len() {
             completed_chunks
@@ -322,20 +363,21 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                 Stop::Automatic,
                 last_chunk_idx != 0,
                 restart,
-            );
+                &check_timeout,
+            )?;
         }
 
         for (number, chunk) in buffer.chunks_mut(255).enumerate() {
             if number != 0 {
                 // NOTE(unsafe) We have &mut self
                 unsafe {
-                    Self::master_continue(chunk.len(), number != last_chunk_idx);
+                    Self::master_continue(chunk.len(), number != last_chunk_idx, &check_timeout)?;
                 }
             }
 
             for byte in chunk {
                 // Wait until we have received something
-                self.wait_rxne()?;
+                self.wait_rxne(&check_timeout)?;
 
                 unsafe {
                     *byte = T::regs().rxdr().read().rxdata();
@@ -345,7 +387,13 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         Ok(())
     }
 
-    fn write_internal(&mut self, address: u8, bytes: &[u8], send_stop: bool) -> Result<(), Error> {
+    fn write_internal(
+        &mut self,
+        address: u8,
+        bytes: &[u8],
+        send_stop: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         let completed_chunks = bytes.len() / 255;
         let total_chunks = if completed_chunks * 255 == bytes.len() {
             completed_chunks
@@ -359,14 +407,20 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         // ST SAD+W
         // NOTE(unsafe) We have &mut self
         unsafe {
-            Self::master_write(address, bytes.len().min(255), Stop::Software, last_chunk_idx != 0);
+            Self::master_write(
+                address,
+                bytes.len().min(255),
+                Stop::Software,
+                last_chunk_idx != 0,
+                &check_timeout,
+            )?;
         }
 
         for (number, chunk) in bytes.chunks(255).enumerate() {
             if number != 0 {
                 // NOTE(unsafe) We have &mut self
                 unsafe {
-                    Self::master_continue(chunk.len(), number != last_chunk_idx);
+                    Self::master_continue(chunk.len(), number != last_chunk_idx, &check_timeout)?;
                 }
             }
 
@@ -374,7 +428,7 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                 // Wait until we are allowed to send data
                 // (START has been ACKed or last byte when
                 // through)
-                self.wait_txe()?;
+                self.wait_txe(&check_timeout)?;
 
                 unsafe {
                     T::regs().txdr().write(|w| w.set_txdata(*byte));
@@ -382,7 +436,7 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             }
         }
         // Wait until the write finishes
-        self.wait_tc()?;
+        self.wait_tc(&check_timeout)?;
 
         if send_stop {
             self.master_stop();
@@ -396,6 +450,7 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         bytes: &[u8],
         first_slice: bool,
         last_slice: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
     ) -> Result<(), Error>
     where
         TXDMA: crate::i2c::TxDma<T>,
@@ -447,11 +502,12 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                     total_len.min(255),
                     Stop::Software,
                     (total_chunks != 1) || !last_slice,
-                );
+                    &check_timeout,
+                )?;
             }
         } else {
             unsafe {
-                Self::master_continue(total_len.min(255), (total_chunks != 1) || !last_slice);
+                Self::master_continue(total_len.min(255), (total_chunks != 1) || !last_slice, &check_timeout)?;
                 T::regs().cr1().modify(|w| w.set_tcie(true));
             }
         }
@@ -461,32 +517,40 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             let chunks_transferred = state.chunks_transferred.load(Ordering::Relaxed);
 
             if chunks_transferred == total_chunks {
-                return Poll::Ready(());
+                return Poll::Ready(Ok(()));
             } else if chunks_transferred != 0 {
                 remaining_len = remaining_len.saturating_sub(255);
                 let last_piece = (chunks_transferred + 1 == total_chunks) && last_slice;
 
                 // NOTE(unsafe) self.tx_dma does not fiddle with the i2c registers
                 unsafe {
-                    Self::master_continue(remaining_len.min(255), !last_piece);
+                    if let Err(e) = Self::master_continue(remaining_len.min(255), !last_piece, &check_timeout) {
+                        return Poll::Ready(Err(e));
+                    }
                     T::regs().cr1().modify(|w| w.set_tcie(true));
                 }
             }
             Poll::Pending
         })
-        .await;
+        .await?;
 
         dma_transfer.await;
 
         if last_slice {
             // This should be done already
-            self.wait_tc()?;
+            self.wait_tc(&check_timeout)?;
             self.master_stop();
         }
         Ok(())
     }
 
-    async fn read_dma_internal(&mut self, address: u8, buffer: &mut [u8], restart: bool) -> Result<(), Error>
+    async fn read_dma_internal(
+        &mut self,
+        address: u8,
+        buffer: &mut [u8],
+        restart: bool,
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error>
     where
         RXDMA: crate::i2c::RxDma<T>,
     {
@@ -527,7 +591,14 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
 
         // NOTE(unsafe) self.rx_dma does not fiddle with the i2c registers
         unsafe {
-            Self::master_read(address, total_len.min(255), Stop::Software, total_chunks != 1, restart);
+            Self::master_read(
+                address,
+                total_len.min(255),
+                Stop::Software,
+                total_chunks != 1,
+                restart,
+                &check_timeout,
+            )?;
         }
 
         poll_fn(|cx| {
@@ -535,25 +606,27 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             let chunks_transferred = state.chunks_transferred.load(Ordering::Relaxed);
 
             if chunks_transferred == total_chunks {
-                return Poll::Ready(());
+                return Poll::Ready(Ok(()));
             } else if chunks_transferred != 0 {
                 remaining_len = remaining_len.saturating_sub(255);
                 let last_piece = chunks_transferred + 1 == total_chunks;
 
                 // NOTE(unsafe) self.rx_dma does not fiddle with the i2c registers
                 unsafe {
-                    Self::master_continue(remaining_len.min(255), !last_piece);
+                    if let Err(e) = Self::master_continue(remaining_len.min(255), !last_piece, &check_timeout) {
+                        return Poll::Ready(Err(e));
+                    }
                     T::regs().cr1().modify(|w| w.set_tcie(true));
                 }
             }
             Poll::Pending
         })
-        .await;
+        .await?;
 
         dma_transfer.await;
 
         // This should be done already
-        self.wait_tc()?;
+        self.wait_tc(&check_timeout)?;
         self.master_stop();
         Ok(())
     }
@@ -566,9 +639,9 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         TXDMA: crate::i2c::TxDma<T>,
     {
         if bytes.is_empty() {
-            self.write_internal(address, bytes, true)
+            self.write_internal(address, bytes, true, || Ok(()))
         } else {
-            self.write_dma_internal(address, bytes, true, true).await
+            self.write_dma_internal(address, bytes, true, true, || Ok(())).await
         }
     }
 
@@ -587,7 +660,7 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             let next = iter.next();
             let is_last = next.is_none();
 
-            self.write_dma_internal(address, c, first, is_last).await?;
+            self.write_dma_internal(address, c, first, is_last, || Ok(())).await?;
             first = false;
             current = next;
         }
@@ -599,9 +672,9 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         RXDMA: crate::i2c::RxDma<T>,
     {
         if buffer.is_empty() {
-            self.read_internal(address, buffer, false)
+            self.read_internal(address, buffer, false, || Ok(()))
         } else {
-            self.read_dma_internal(address, buffer, false).await
+            self.read_dma_internal(address, buffer, false, || Ok(())).await
         }
     }
 
@@ -611,15 +684,15 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
         RXDMA: super::RxDma<T>,
     {
         if bytes.is_empty() {
-            self.write_internal(address, bytes, false)?;
+            self.write_internal(address, bytes, false, || Ok(()))?;
         } else {
-            self.write_dma_internal(address, bytes, true, true).await?;
+            self.write_dma_internal(address, bytes, true, true, || Ok(())).await?;
         }
 
         if buffer.is_empty() {
-            self.read_internal(address, buffer, true)?;
+            self.read_internal(address, buffer, true, || Ok(()))?;
         } else {
-            self.read_dma_internal(address, buffer, true).await?;
+            self.read_dma_internal(address, buffer, true, || Ok(())).await?;
         }
 
         Ok(())
@@ -628,22 +701,55 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
     // =========================
     //  Blocking public API
 
-    pub fn blocking_read(&mut self, address: u8, buffer: &mut [u8]) -> Result<(), Error> {
-        self.read_internal(address, buffer, false)
+    pub fn blocking_read_timeout(
+        &mut self,
+        address: u8,
+        buffer: &mut [u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
+        self.read_internal(address, buffer, false, &check_timeout)
         // Automatic Stop
     }
 
+    pub fn blocking_read(&mut self, address: u8, buffer: &mut [u8]) -> Result<(), Error> {
+        self.blocking_read_timeout(address, buffer, || Ok(()))
+    }
+
+    pub fn blocking_write_timeout(
+        &mut self,
+        address: u8,
+        bytes: &[u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
+        self.write_internal(address, bytes, true, &check_timeout)
+    }
+
     pub fn blocking_write(&mut self, address: u8, bytes: &[u8]) -> Result<(), Error> {
-        self.write_internal(address, bytes, true)
+        self.blocking_write_timeout(address, bytes, || Ok(()))
     }
 
-    pub fn blocking_write_read(&mut self, address: u8, bytes: &[u8], buffer: &mut [u8]) -> Result<(), Error> {
-        self.write_internal(address, bytes, false)?;
-        self.read_internal(address, buffer, true)
+    pub fn blocking_write_read_timeout(
+        &mut self,
+        address: u8,
+        bytes: &[u8],
+        buffer: &mut [u8],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
+        self.write_internal(address, bytes, false, &check_timeout)?;
+        self.read_internal(address, buffer, true, &check_timeout)
         // Automatic Stop
     }
 
-    pub fn blocking_write_vectored(&mut self, address: u8, bytes: &[&[u8]]) -> Result<(), Error> {
+    pub fn blocking_write_read(&mut self, address: u8, bytes: &[u8], buffer: &mut [u8]) -> Result<(), Error> {
+        self.blocking_write_read_timeout(address, bytes, buffer, || Ok(()))
+    }
+
+    pub fn blocking_write_vectored_timeout(
+        &mut self,
+        address: u8,
+        bytes: &[&[u8]],
+        check_timeout: impl Fn() -> Result<(), Error>,
+    ) -> Result<(), Error> {
         if bytes.is_empty() {
             return Err(Error::ZeroLengthTransfer);
         }
@@ -657,7 +763,8 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                 first_length.min(255),
                 Stop::Software,
                 (first_length > 255) || (last_slice_index != 0),
-            );
+                &check_timeout,
+            )?;
         }
 
         for (idx, slice) in bytes.iter().enumerate() {
@@ -673,7 +780,11 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             if idx != 0 {
                 // NOTE(unsafe) We have &mut self
                 unsafe {
-                    Self::master_continue(slice_len.min(255), (idx != last_slice_index) || (slice_len > 255));
+                    Self::master_continue(
+                        slice_len.min(255),
+                        (idx != last_slice_index) || (slice_len > 255),
+                        &check_timeout,
+                    )?;
                 }
             }
 
@@ -681,7 +792,11 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                 if number != 0 {
                     // NOTE(unsafe) We have &mut self
                     unsafe {
-                        Self::master_continue(chunk.len(), (number != last_chunk_idx) || (idx != last_slice_index));
+                        Self::master_continue(
+                            chunk.len(),
+                            (number != last_chunk_idx) || (idx != last_slice_index),
+                            &check_timeout,
+                        )?;
                     }
                 }
 
@@ -689,7 +804,7 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
                     // Wait until we are allowed to send data
                     // (START has been ACKed or last byte when
                     // through)
-                    self.wait_txe()?;
+                    self.wait_txe(&check_timeout)?;
 
                     // Put byte on the wire
                     //self.i2c.txdr.write(|w| w.txdata().bits(*byte));
@@ -700,11 +815,15 @@ impl<'d, T: Instance, TXDMA, RXDMA> I2c<'d, T, TXDMA, RXDMA> {
             }
         }
         // Wait until the write finishes
-        self.wait_tc()?;
+        self.wait_tc(&check_timeout)?;
         self.master_stop();
 
         Ok(())
     }
+
+    pub fn blocking_write_vectored(&mut self, address: u8, bytes: &[&[u8]]) -> Result<(), Error> {
+        self.blocking_write_vectored_timeout(address, bytes, || Ok(()))
+    }
 }
 
 mod eh02 {
diff --git a/embassy-stm32/src/lib.rs b/embassy-stm32/src/lib.rs
index 0392e8086..bcf2feee8 100644
--- a/embassy-stm32/src/lib.rs
+++ b/embassy-stm32/src/lib.rs
@@ -52,7 +52,7 @@ pub mod sdmmc;
 pub mod spi;
 #[cfg(usart)]
 pub mod usart;
-#[cfg(usb)]
+#[cfg(all(usb, feature = "time"))]
 pub mod usb;
 #[cfg(any(otgfs, otghs))]
 pub mod usb_otg;
diff --git a/embassy-stm32/src/sdmmc/mod.rs b/embassy-stm32/src/sdmmc/mod.rs
index a8bc6385f..c91f3c8bf 100644
--- a/embassy-stm32/src/sdmmc/mod.rs
+++ b/embassy-stm32/src/sdmmc/mod.rs
@@ -1534,14 +1534,14 @@ mod sdmmc_rs {
 
     impl<'d, T: Instance, P: Pins<T>> BlockDevice for Sdmmc<'d, T, P> {
         type Error = Error;
-        type ReadFuture<'a>
+
+        type ReadFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
         where
-            Self: 'a,
-        = impl Future<Output = Result<(), Self::Error>> + 'a;
-        type WriteFuture<'a>
+            Self: 'a;
+
+        type WriteFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
         where
-            Self: 'a,
-        = impl Future<Output = Result<(), Self::Error>> + 'a;
+            Self: 'a;
 
         fn read<'a>(
             &'a mut self,
diff --git a/embassy-stm32/src/subghz/timeout.rs b/embassy-stm32/src/subghz/timeout.rs
index 28b3b0c21..0ae49dd90 100644
--- a/embassy-stm32/src/subghz/timeout.rs
+++ b/embassy-stm32/src/subghz/timeout.rs
@@ -439,6 +439,7 @@ impl From<Timeout> for [u8; 3] {
     }
 }
 
+#[cfg(feature = "time")]
 impl From<Timeout> for embassy_time::Duration {
     fn from(to: Timeout) -> Self {
         embassy_time::Duration::from_micros(to.as_micros().into())
diff --git a/embassy-stm32/src/subghz/tx_params.rs b/embassy-stm32/src/subghz/tx_params.rs
index cede6f2c1..03bdb1ea8 100644
--- a/embassy-stm32/src/subghz/tx_params.rs
+++ b/embassy-stm32/src/subghz/tx_params.rs
@@ -44,6 +44,7 @@ impl From<RampTime> for core::time::Duration {
     }
 }
 
+#[cfg(feature = "time")]
 impl From<RampTime> for embassy_time::Duration {
     fn from(rt: RampTime) -> Self {
         match rt {
diff --git a/embassy-stm32/src/time_driver.rs b/embassy-stm32/src/time_driver.rs
index ed3225c51..8e84570a4 100644
--- a/embassy-stm32/src/time_driver.rs
+++ b/embassy-stm32/src/time_driver.rs
@@ -292,19 +292,23 @@ impl Driver for RtcDriver {
         })
     }
 
-    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) {
+    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) -> bool {
         critical_section::with(|cs| {
             let r = T::regs_gp16();
 
-            let n = alarm.id() as _;
+            let n = alarm.id() as usize;
             let alarm = self.get_alarm(cs, alarm);
             alarm.timestamp.set(timestamp);
 
             let t = self.now();
             if timestamp <= t {
+                // If alarm timestamp has passed the alarm will not fire.
+                // Disarm the alarm and return `false` to indicate that.
                 unsafe { r.dier().modify(|w| w.set_ccie(n + 1, false)) };
-                self.trigger_alarm(n, cs);
-                return;
+
+                alarm.timestamp.set(u64::MAX);
+
+                return false;
             }
 
             let safe_timestamp = timestamp.max(t + 3);
@@ -317,6 +321,8 @@ impl Driver for RtcDriver {
             let diff = timestamp - t;
             // NOTE(unsafe) We're in a critical section
             unsafe { r.dier().modify(|w| w.set_ccie(n + 1, diff < 0xc000)) };
+
+            true
         })
     }
 }
diff --git a/embassy-stm32/src/usart/buffered.rs b/embassy-stm32/src/usart/buffered.rs
index 2a711bc06..3be0677bd 100644
--- a/embassy-stm32/src/usart/buffered.rs
+++ b/embassy-stm32/src/usart/buffered.rs
@@ -46,16 +46,44 @@ impl<'d, T: BasicInstance> Unpin for BufferedUart<'d, T> {}
 impl<'d, T: BasicInstance> BufferedUart<'d, T> {
     pub fn new(
         state: &'d mut State<'d, T>,
-        _uart: Uart<'d, T, NoDma, NoDma>,
+        _peri: impl Peripheral<P = T> + 'd,
+        rx: impl Peripheral<P = impl RxPin<T>> + 'd,
+        tx: impl Peripheral<P = impl TxPin<T>> + 'd,
         irq: impl Peripheral<P = T::Interrupt> + 'd,
         tx_buffer: &'d mut [u8],
         rx_buffer: &'d mut [u8],
+        config: Config,
     ) -> BufferedUart<'d, T> {
-        into_ref!(irq);
+        into_ref!(_peri, rx, tx, irq);
+
+        T::enable();
+        T::reset();
 
         let r = T::regs();
+
+        configure(r, &config, T::frequency(), T::MULTIPLIER);
+
         unsafe {
-            r.cr1().modify(|w| {
+            rx.set_as_af(rx.af_num(), AFType::Input);
+            tx.set_as_af(tx.af_num(), AFType::OutputPushPull);
+
+            r.cr2().write(|_w| {});
+            r.cr3().write(|_w| {});
+            r.cr1().write(|w| {
+                w.set_ue(true);
+                w.set_te(true);
+                w.set_re(true);
+                w.set_m0(if config.parity != Parity::ParityNone {
+                    vals::M0::BIT9
+                } else {
+                    vals::M0::BIT8
+                });
+                w.set_pce(config.parity != Parity::ParityNone);
+                w.set_ps(match config.parity {
+                    Parity::ParityOdd => vals::Ps::ODD,
+                    Parity::ParityEven => vals::Ps::EVEN,
+                    _ => vals::Ps::EVEN,
+                });
                 w.set_rxneie(true);
                 w.set_idleie(true);
             });
@@ -283,7 +311,7 @@ impl<'u, 'd, T: BasicInstance> embedded_io::Io for BufferedUartTx<'u, 'd, T> {
 }
 
 impl<'d, T: BasicInstance> embedded_io::asynch::Read for BufferedUart<'d, T> {
-    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -293,7 +321,7 @@ impl<'d, T: BasicInstance> embedded_io::asynch::Read for BufferedUart<'d, T> {
 }
 
 impl<'u, 'd, T: BasicInstance> embedded_io::asynch::Read for BufferedUartRx<'u, 'd, T> {
-    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -303,7 +331,7 @@ impl<'u, 'd, T: BasicInstance> embedded_io::asynch::Read for BufferedUartRx<'u,
 }
 
 impl<'d, T: BasicInstance> embedded_io::asynch::BufRead for BufferedUart<'d, T> {
-    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>>
+    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -317,7 +345,7 @@ impl<'d, T: BasicInstance> embedded_io::asynch::BufRead for BufferedUart<'d, T>
 }
 
 impl<'u, 'd, T: BasicInstance> embedded_io::asynch::BufRead for BufferedUartRx<'u, 'd, T> {
-    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>>
+    type FillBufFuture<'a> = impl Future<Output = Result<&'a [u8], Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -331,7 +359,7 @@ impl<'u, 'd, T: BasicInstance> embedded_io::asynch::BufRead for BufferedUartRx<'
 }
 
 impl<'d, T: BasicInstance> embedded_io::asynch::Write for BufferedUart<'d, T> {
-    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -339,7 +367,7 @@ impl<'d, T: BasicInstance> embedded_io::asynch::Write for BufferedUart<'d, T> {
         self.inner_write(buf)
     }
 
-    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -349,7 +377,7 @@ impl<'d, T: BasicInstance> embedded_io::asynch::Write for BufferedUart<'d, T> {
 }
 
 impl<'u, 'd, T: BasicInstance> embedded_io::asynch::Write for BufferedUartTx<'u, 'd, T> {
-    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+    type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
     where
         Self: 'a;
 
@@ -357,7 +385,7 @@ impl<'u, 'd, T: BasicInstance> embedded_io::asynch::Write for BufferedUartTx<'u,
         self.inner.inner_write(buf)
     }
 
-    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+    type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
     where
         Self: 'a;
 
diff --git a/embassy-stm32/src/usart/mod.rs b/embassy-stm32/src/usart/mod.rs
index a152a0c15..dee466b21 100644
--- a/embassy-stm32/src/usart/mod.rs
+++ b/embassy-stm32/src/usart/mod.rs
@@ -1,7 +1,11 @@
 #![macro_use]
 
+use core::future::poll_fn;
 use core::marker::PhantomData;
+use core::task::Poll;
 
+use atomic_polyfill::{compiler_fence, Ordering};
+use embassy_cortex_m::interrupt::InterruptExt;
 use embassy_hal_common::{into_ref, PeripheralRef};
 
 use crate::dma::NoDma;
@@ -10,6 +14,7 @@ use crate::gpio::sealed::AFType;
 use crate::pac::lpuart::{regs, vals, Lpuart as Regs};
 #[cfg(not(any(lpuart_v1, lpuart_v2)))]
 use crate::pac::usart::{regs, vals, Usart as Regs};
+use crate::time::Hertz;
 use crate::{peripherals, Peripheral};
 
 #[derive(Clone, Copy, PartialEq, Eq, Debug)]
@@ -44,6 +49,10 @@ pub struct Config {
     pub data_bits: DataBits,
     pub stop_bits: StopBits,
     pub parity: Parity,
+    /// if true, on read-like method, if there is a latent error pending,
+    /// read will abort, the error reported and cleared
+    /// if false, the error is ignored and cleared
+    pub detect_previous_overrun: bool,
 }
 
 impl Default for Config {
@@ -53,6 +62,8 @@ impl Default for Config {
             data_bits: DataBits::DataBits8,
             stop_bits: StopBits::STOP1,
             parity: Parity::ParityNone,
+            // historical behavior
+            detect_previous_overrun: false,
         }
     }
 }
@@ -70,10 +81,11 @@ pub enum Error {
     Overrun,
     /// Parity check error
     Parity,
+    /// Buffer too large for DMA
+    BufferTooLong,
 }
 
 pub struct Uart<'d, T: BasicInstance, TxDma = NoDma, RxDma = NoDma> {
-    phantom: PhantomData<&'d mut T>,
     tx: UartTx<'d, T, TxDma>,
     rx: UartRx<'d, T, RxDma>,
 }
@@ -84,8 +96,9 @@ pub struct UartTx<'d, T: BasicInstance, TxDma = NoDma> {
 }
 
 pub struct UartRx<'d, T: BasicInstance, RxDma = NoDma> {
-    phantom: PhantomData<&'d mut T>,
+    _peri: PeripheralRef<'d, T>,
     rx_dma: PeripheralRef<'d, RxDma>,
+    detect_previous_overrun: bool,
 }
 
 impl<'d, T: BasicInstance, TxDma> UartTx<'d, T, TxDma> {
@@ -135,10 +148,112 @@ impl<'d, T: BasicInstance, TxDma> UartTx<'d, T, TxDma> {
 }
 
 impl<'d, T: BasicInstance, RxDma> UartRx<'d, T, RxDma> {
-    fn new(rx_dma: PeripheralRef<'d, RxDma>) -> Self {
+    /// usefull if you only want Uart Rx. It saves 1 pin and consumes a little less power
+    pub fn new(
+        peri: impl Peripheral<P = T> + 'd,
+        irq: impl Peripheral<P = T::Interrupt> + 'd,
+        rx: impl Peripheral<P = impl RxPin<T>> + 'd,
+        rx_dma: impl Peripheral<P = RxDma> + 'd,
+        config: Config,
+    ) -> Self {
+        into_ref!(peri, irq, rx, rx_dma);
+
+        T::enable();
+        T::reset();
+
+        let r = T::regs();
+
+        configure(r, &config, T::frequency(), T::MULTIPLIER);
+
+        unsafe {
+            rx.set_as_af(rx.af_num(), AFType::Input);
+
+            r.cr2().write(|_w| {});
+            r.cr3().write(|w| {
+                // enable Error Interrupt: (Frame error, Noise error, Overrun error)
+                w.set_eie(true);
+            });
+            r.cr1().write(|w| {
+                // enable uart
+                w.set_ue(true);
+                // enable receiver
+                w.set_re(true);
+                // configure word size
+                w.set_m0(if config.parity != Parity::ParityNone {
+                    vals::M0::BIT9
+                } else {
+                    vals::M0::BIT8
+                });
+                // configure parity
+                w.set_pce(config.parity != Parity::ParityNone);
+                w.set_ps(match config.parity {
+                    Parity::ParityOdd => vals::Ps::ODD,
+                    Parity::ParityEven => vals::Ps::EVEN,
+                    _ => vals::Ps::EVEN,
+                });
+            });
+        }
+
+        irq.set_handler(Self::on_interrupt);
+        irq.unpend();
+        irq.enable();
+
+        // create state once!
+        let _s = T::state();
+
         Self {
+            _peri: peri,
             rx_dma,
-            phantom: PhantomData,
+            detect_previous_overrun: config.detect_previous_overrun,
+        }
+    }
+
+    fn on_interrupt(_: *mut ()) {
+        let r = T::regs();
+        let s = T::state();
+
+        let (sr, cr1, cr3) = unsafe { (sr(r).read(), r.cr1().read(), r.cr3().read()) };
+
+        let has_errors = (sr.pe() && cr1.peie()) || ((sr.fe() || sr.ne() || sr.ore()) && cr3.eie());
+
+        if has_errors {
+            // clear all interrupts and DMA Rx Request
+            unsafe {
+                r.cr1().modify(|w| {
+                    // disable RXNE interrupt
+                    w.set_rxneie(false);
+                    // disable parity interrupt
+                    w.set_peie(false);
+                    // disable idle line interrupt
+                    w.set_idleie(false);
+                });
+                r.cr3().modify(|w| {
+                    // disable Error Interrupt: (Frame error, Noise error, Overrun error)
+                    w.set_eie(false);
+                    // disable DMA Rx Request
+                    w.set_dmar(false);
+                });
+            }
+
+            compiler_fence(Ordering::SeqCst);
+
+            s.rx_waker.wake();
+        } else if cr1.idleie() && sr.idle() {
+            // IDLE detected: no more data will come
+            unsafe {
+                r.cr1().modify(|w| {
+                    // disable idle line detection
+                    w.set_idleie(false);
+                });
+
+                r.cr3().modify(|w| {
+                    // disable DMA Rx Request
+                    w.set_dmar(false);
+                });
+            }
+            compiler_fence(Ordering::SeqCst);
+
+            s.rx_waker.wake();
         }
     }
 
@@ -146,17 +261,8 @@ impl<'d, T: BasicInstance, RxDma> UartRx<'d, T, RxDma> {
     where
         RxDma: crate::usart::RxDma<T>,
     {
-        let ch = &mut self.rx_dma;
-        let request = ch.request();
-        unsafe {
-            T::regs().cr3().modify(|reg| {
-                reg.set_dmar(true);
-            });
-        }
-        // If we don't assign future to a variable, the data register pointer
-        // is held across an await and makes the future non-Send.
-        let transfer = crate::dma::read(ch, request, rdr(T::regs()), buffer);
-        transfer.await;
+        self.inner_read(buffer, false).await?;
+
         Ok(())
     }
 
@@ -211,35 +317,259 @@ impl<'d, T: BasicInstance, RxDma> UartRx<'d, T, RxDma> {
         }
         Ok(())
     }
+
+    pub async fn read_until_idle(&mut self, buffer: &mut [u8]) -> Result<usize, Error>
+    where
+        RxDma: crate::usart::RxDma<T>,
+    {
+        self.inner_read(buffer, true).await
+    }
+
+    async fn inner_read(&mut self, buffer: &mut [u8], enable_idle_line_detection: bool) -> Result<usize, Error>
+    where
+        RxDma: crate::usart::RxDma<T>,
+    {
+        if buffer.is_empty() {
+            return Ok(0);
+        } else if buffer.len() > 0xFFFF {
+            return Err(Error::BufferTooLong);
+        }
+
+        let r = T::regs();
+
+        let buffer_len = buffer.len();
+
+        let ch = &mut self.rx_dma;
+        let request = ch.request();
+
+        // SAFETY: The only way we might have a problem is using split rx and tx
+        // here we only modify or read Rx related flags, interrupts and DMA channel
+        unsafe {
+            // Start USART DMA
+            // will not do anything yet because DMAR is not yet set
+            ch.start_read(request, rdr(r), buffer, Default::default());
+
+            // clear ORE flag just before enabling DMA Rx Request: can be mandatory for the second transfer
+            if !self.detect_previous_overrun {
+                let sr = sr(r).read();
+                // This read also clears the error and idle interrupt flags on v1.
+                rdr(r).read_volatile();
+                clear_interrupt_flags(r, sr);
+            }
+
+            r.cr1().modify(|w| {
+                // disable RXNE interrupt
+                w.set_rxneie(false);
+                // enable parity interrupt if not ParityNone
+                w.set_peie(w.pce());
+            });
+
+            r.cr3().modify(|w| {
+                // enable Error Interrupt: (Frame error, Noise error, Overrun error)
+                w.set_eie(true);
+                // enable DMA Rx Request
+                w.set_dmar(true);
+            });
+
+            compiler_fence(Ordering::SeqCst);
+
+            // In case of errors already pending when reception started, interrupts may have already been raised
+            // and lead to reception abortion (Overrun error for instance). In such a case, all interrupts
+            // have been disabled in interrupt handler and DMA Rx Request has been disabled.
+
+            let cr3 = r.cr3().read();
+
+            if !cr3.dmar() {
+                // something went wrong
+                // because the only way to get this flag cleared is to have an interrupt
+
+                // abort DMA transfer
+                ch.request_stop();
+                while ch.is_running() {}
+
+                let sr = sr(r).read();
+                // This read also clears the error and idle interrupt flags on v1.
+                rdr(r).read_volatile();
+                clear_interrupt_flags(r, sr);
+
+                if sr.pe() {
+                    return Err(Error::Parity);
+                }
+                if sr.fe() {
+                    return Err(Error::Framing);
+                }
+                if sr.ne() {
+                    return Err(Error::Noise);
+                }
+                if sr.ore() {
+                    return Err(Error::Overrun);
+                }
+
+                unreachable!();
+            }
+
+            // clear idle flag
+            if enable_idle_line_detection {
+                let sr = sr(r).read();
+                // This read also clears the error and idle interrupt flags on v1.
+                rdr(r).read_volatile();
+                clear_interrupt_flags(r, sr);
+
+                // enable idle interrupt
+                r.cr1().modify(|w| {
+                    w.set_idleie(true);
+                });
+            }
+        }
+
+        compiler_fence(Ordering::SeqCst);
+
+        let res = poll_fn(move |cx| {
+            let s = T::state();
+
+            ch.set_waker(cx.waker());
+            s.rx_waker.register(cx.waker());
+
+            // SAFETY: read only and we only use Rx related flags
+            let sr = unsafe { sr(r).read() };
+
+            // SAFETY: only clears Rx related flags
+            unsafe {
+                // This read also clears the error and idle interrupt flags on v1.
+                rdr(r).read_volatile();
+                clear_interrupt_flags(r, sr);
+            }
+
+            compiler_fence(Ordering::SeqCst);
+
+            let has_errors = sr.pe() || sr.fe() || sr.ne() || sr.ore();
+
+            if has_errors {
+                // all Rx interrupts and Rx DMA Request have already been cleared in interrupt handler
+
+                // stop dma transfer
+                ch.request_stop();
+                while ch.is_running() {}
+
+                if sr.pe() {
+                    return Poll::Ready(Err(Error::Parity));
+                }
+                if sr.fe() {
+                    return Poll::Ready(Err(Error::Framing));
+                }
+                if sr.ne() {
+                    return Poll::Ready(Err(Error::Noise));
+                }
+                if sr.ore() {
+                    return Poll::Ready(Err(Error::Overrun));
+                }
+            }
+
+            if sr.idle() {
+                // Idle line
+
+                // stop dma transfer
+                ch.request_stop();
+                while ch.is_running() {}
+
+                let n = buffer_len - (ch.remaining_transfers() as usize);
+
+                return Poll::Ready(Ok(n));
+            } else if !ch.is_running() {
+                // DMA complete
+                return Poll::Ready(Ok(buffer_len));
+            }
+
+            Poll::Pending
+        })
+        .await;
+
+        // clear all interrupts and DMA Rx Request
+        // SAFETY: only clears Rx related flags
+        unsafe {
+            r.cr1().modify(|w| {
+                // disable RXNE interrupt
+                w.set_rxneie(false);
+                // disable parity interrupt
+                w.set_peie(false);
+                // disable idle line interrupt
+                w.set_idleie(false);
+            });
+            r.cr3().modify(|w| {
+                // disable Error Interrupt: (Frame error, Noise error, Overrun error)
+                w.set_eie(false);
+                // disable DMA Rx Request
+                w.set_dmar(false);
+            });
+        }
+
+        res
+    }
 }
 
 impl<'d, T: BasicInstance, TxDma, RxDma> Uart<'d, T, TxDma, RxDma> {
     pub fn new(
-        _inner: impl Peripheral<P = T> + 'd,
+        peri: impl Peripheral<P = T> + 'd,
         rx: impl Peripheral<P = impl RxPin<T>> + 'd,
         tx: impl Peripheral<P = impl TxPin<T>> + 'd,
+        irq: impl Peripheral<P = T::Interrupt> + 'd,
         tx_dma: impl Peripheral<P = TxDma> + 'd,
         rx_dma: impl Peripheral<P = RxDma> + 'd,
         config: Config,
     ) -> Self {
-        into_ref!(_inner, rx, tx, tx_dma, rx_dma);
+        T::enable();
+        T::reset();
+
+        Self::new_inner(peri, rx, tx, irq, tx_dma, rx_dma, config)
+    }
+
+    pub fn new_with_rtscts(
+        peri: impl Peripheral<P = T> + 'd,
+        rx: impl Peripheral<P = impl RxPin<T>> + 'd,
+        tx: impl Peripheral<P = impl TxPin<T>> + 'd,
+        irq: impl Peripheral<P = T::Interrupt> + 'd,
+        rts: impl Peripheral<P = impl RtsPin<T>> + 'd,
+        cts: impl Peripheral<P = impl CtsPin<T>> + 'd,
+        tx_dma: impl Peripheral<P = TxDma> + 'd,
+        rx_dma: impl Peripheral<P = RxDma> + 'd,
+        config: Config,
+    ) -> Self {
+        into_ref!(cts, rts);
 
         T::enable();
         T::reset();
-        let pclk_freq = T::frequency();
 
-        // TODO: better calculation, including error checking and OVER8 if possible.
-        let div = (pclk_freq.0 + (config.baudrate / 2)) / config.baudrate * T::MULTIPLIER;
+        unsafe {
+            rts.set_as_af(rts.af_num(), AFType::OutputPushPull);
+            cts.set_as_af(cts.af_num(), AFType::Input);
+            T::regs().cr3().write(|w| {
+                w.set_rtse(true);
+                w.set_ctse(true);
+            });
+        }
+        Self::new_inner(peri, rx, tx, irq, tx_dma, rx_dma, config)
+    }
+
+    fn new_inner(
+        peri: impl Peripheral<P = T> + 'd,
+        rx: impl Peripheral<P = impl RxPin<T>> + 'd,
+        tx: impl Peripheral<P = impl TxPin<T>> + 'd,
+        irq: impl Peripheral<P = T::Interrupt> + 'd,
+        tx_dma: impl Peripheral<P = TxDma> + 'd,
+        rx_dma: impl Peripheral<P = RxDma> + 'd,
+        config: Config,
+    ) -> Self {
+        into_ref!(peri, rx, tx, irq, tx_dma, rx_dma);
 
         let r = T::regs();
 
+        configure(r, &config, T::frequency(), T::MULTIPLIER);
+
         unsafe {
             rx.set_as_af(rx.af_num(), AFType::Input);
             tx.set_as_af(tx.af_num(), AFType::OutputPushPull);
 
             r.cr2().write(|_w| {});
-            r.cr3().write(|_w| {});
-            r.brr().write_value(regs::Brr(div));
             r.cr1().write(|w| {
                 w.set_ue(true);
                 w.set_te(true);
@@ -258,10 +588,20 @@ impl<'d, T: BasicInstance, TxDma, RxDma> Uart<'d, T, TxDma, RxDma> {
             });
         }
 
+        irq.set_handler(UartRx::<T, RxDma>::on_interrupt);
+        irq.unpend();
+        irq.enable();
+
+        // create state once!
+        let _s = T::state();
+
         Self {
             tx: UartTx::new(tx_dma),
-            rx: UartRx::new(rx_dma),
-            phantom: PhantomData {},
+            rx: UartRx {
+                _peri: peri,
+                rx_dma,
+                detect_previous_overrun: config.detect_previous_overrun,
+            },
         }
     }
 
@@ -295,6 +635,13 @@ impl<'d, T: BasicInstance, TxDma, RxDma> Uart<'d, T, TxDma, RxDma> {
         self.rx.blocking_read(buffer)
     }
 
+    pub async fn read_until_idle(&mut self, buffer: &mut [u8]) -> Result<usize, Error>
+    where
+        RxDma: crate::usart::RxDma<T>,
+    {
+        self.rx.read_until_idle(buffer).await
+    }
+
     /// Split the Uart into a transmitter and receiver, which is
     /// particuarly useful when having two tasks correlating to
     /// transmitting and receiving.
@@ -303,6 +650,15 @@ impl<'d, T: BasicInstance, TxDma, RxDma> Uart<'d, T, TxDma, RxDma> {
     }
 }
 
+fn configure(r: Regs, config: &Config, pclk_freq: Hertz, multiplier: u32) {
+    // TODO: better calculation, including error checking and OVER8 if possible.
+    let div = (pclk_freq.0 + (config.baudrate / 2)) / config.baudrate * multiplier;
+
+    unsafe {
+        r.brr().write_value(regs::Brr(div));
+    }
+}
+
 mod eh02 {
     use super::*;
 
@@ -352,6 +708,7 @@ mod eh1 {
                 Self::Noise => embedded_hal_1::serial::ErrorKind::Noise,
                 Self::Overrun => embedded_hal_1::serial::ErrorKind::Overrun,
                 Self::Parity => embedded_hal_1::serial::ErrorKind::Parity,
+                Self::BufferTooLong => embedded_hal_1::serial::ErrorKind::Other,
             }
         }
     }
@@ -536,13 +893,30 @@ unsafe fn clear_interrupt_flags(r: Regs, sr: regs::Isr) {
 }
 
 pub(crate) mod sealed {
+    use embassy_sync::waitqueue::AtomicWaker;
+
     use super::*;
 
+    pub struct State {
+        pub rx_waker: AtomicWaker,
+        pub tx_waker: AtomicWaker,
+    }
+
+    impl State {
+        pub const fn new() -> Self {
+            Self {
+                rx_waker: AtomicWaker::new(),
+                tx_waker: AtomicWaker::new(),
+            }
+        }
+    }
+
     pub trait BasicInstance: crate::rcc::RccPeripheral {
         const MULTIPLIER: u32;
         type Interrupt: crate::interrupt::Interrupt;
 
         fn regs() -> Regs;
+        fn state() -> &'static State;
     }
 
     pub trait FullInstance: BasicInstance {
@@ -550,7 +924,7 @@ pub(crate) mod sealed {
     }
 }
 
-pub trait BasicInstance: sealed::BasicInstance {}
+pub trait BasicInstance: Peripheral<P = Self> + sealed::BasicInstance + 'static + Send {}
 
 pub trait FullInstance: sealed::FullInstance {}
 
@@ -572,6 +946,11 @@ macro_rules! impl_lpuart {
             fn regs() -> Regs {
                 Regs(crate::pac::$inst.0)
             }
+
+            fn state() -> &'static crate::usart::sealed::State {
+                static STATE: crate::usart::sealed::State = crate::usart::sealed::State::new();
+                &STATE
+            }
         }
 
         impl BasicInstance for peripherals::$inst {}
diff --git a/embassy-sync/Cargo.toml b/embassy-sync/Cargo.toml
index 584d5ba9f..b7fe1643c 100644
--- a/embassy-sync/Cargo.toml
+++ b/embassy-sync/Cargo.toml
@@ -19,6 +19,9 @@ src_base_git = "https://github.com/embassy-rs/embassy/blob/$COMMIT/embassy-sync/
 features = ["nightly"]
 target = "thumbv7em-none-eabi"
 
+[package.metadata.docs.rs]
+features = ["nightly"]
+
 [features]
 nightly = ["embedded-io/async"]
 std = []
@@ -32,7 +35,7 @@ atomic-polyfill = "1.0.1"
 critical-section = "1.1"
 heapless = "0.7.5"
 cfg-if = "1.0.0"
-embedded-io = "0.3.0"
+embedded-io = "0.3.1"
 
 [dev-dependencies]
 futures-executor = { version = "0.3.17", features = [ "thread-pool" ] }
diff --git a/embassy-sync/src/pipe.rs b/embassy-sync/src/pipe.rs
index 7d64b648e..cd577f34f 100644
--- a/embassy-sync/src/pipe.rs
+++ b/embassy-sync/src/pipe.rs
@@ -361,7 +361,7 @@ mod io_impls {
     }
 
     impl<M: RawMutex, const N: usize> embedded_io::asynch::Read for Pipe<M, N> {
-        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -371,7 +371,7 @@ mod io_impls {
     }
 
     impl<M: RawMutex, const N: usize> embedded_io::asynch::Write for Pipe<M, N> {
-        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -379,7 +379,7 @@ mod io_impls {
             Pipe::write(self, buf).map(Ok)
         }
 
-        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -393,7 +393,7 @@ mod io_impls {
     }
 
     impl<M: RawMutex, const N: usize> embedded_io::asynch::Read for &Pipe<M, N> {
-        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -403,7 +403,7 @@ mod io_impls {
     }
 
     impl<M: RawMutex, const N: usize> embedded_io::asynch::Write for &Pipe<M, N> {
-        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -411,7 +411,7 @@ mod io_impls {
             Pipe::write(self, buf).map(Ok)
         }
 
-        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -425,7 +425,7 @@ mod io_impls {
     }
 
     impl<M: RawMutex, const N: usize> embedded_io::asynch::Read for Reader<'_, M, N> {
-        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type ReadFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -439,7 +439,7 @@ mod io_impls {
     }
 
     impl<M: RawMutex, const N: usize> embedded_io::asynch::Write for Writer<'_, M, N> {
-        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>>
+        type WriteFuture<'a> = impl Future<Output = Result<usize, Self::Error>> + 'a
         where
             Self: 'a;
 
@@ -447,7 +447,7 @@ mod io_impls {
             Writer::write(self, buf).map(Ok)
         }
 
-        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>>
+        type FlushFuture<'a> = impl Future<Output = Result<(), Self::Error>> + 'a
         where
             Self: 'a;
 
diff --git a/embassy-time/Cargo.toml b/embassy-time/Cargo.toml
index c51a71d01..9487003cc 100644
--- a/embassy-time/Cargo.toml
+++ b/embassy-time/Cargo.toml
@@ -2,8 +2,16 @@
 name = "embassy-time"
 version = "0.1.0"
 edition = "2021"
+description = "Instant and Duration for embedded no-std systems, with async timer support"
+repository = "https://github.com/embassy-rs/embassy"
+readme = "README.md"
 license = "MIT OR Apache-2.0"
-
+categories = [
+    "embedded",
+    "no-std",
+    "concurrency",
+    "asynchronous",
+]
 
 [package.metadata.embassy_docs]
 src_base = "https://github.com/embassy-rs/embassy/blob/embassy-time-v$VERSION/embassy-time/src/"
@@ -11,6 +19,9 @@ src_base_git = "https://github.com/embassy-rs/embassy/blob/$COMMIT/embassy-time/
 features = ["nightly", "defmt", "unstable-traits", "std"]
 target = "x86_64-unknown-linux-gnu"
 
+[package.metadata.docs.rs]
+features = ["nightly", "defmt", "unstable-traits", "std"]
+
 [features]
 std = ["tick-hz-1_000_000"]
 wasm = ["dep:wasm-bindgen", "dep:js-sys", "dep:wasm-timer", "tick-hz-1_000_000"]
@@ -26,6 +37,22 @@ unstable-traits = ["embedded-hal-1"]
 # To use this you must have a time driver provided.
 defmt-timestamp-uptime = ["defmt"]
 
+# Create a global, generic queue that can be used with any executor
+# To use this you must have a time driver provided.
+generic-queue = []
+
+# Set the number of timers for the generic queue.
+#
+# At most 1 `generic-queue-*` feature can be enabled. If none is enabled, a default of 64 timers is used.
+# 
+# When using embassy-time from libraries, you should *not* enable any `generic-queue-*` feature, to allow the
+# end user to pick.
+generic-queue-8 = ["generic-queue"]
+generic-queue-16 = ["generic-queue"]
+generic-queue-32 = ["generic-queue"]
+generic-queue-64 = ["generic-queue"]
+generic-queue-128 = ["generic-queue"]
+
 # Set the `embassy_time` tick rate.
 #
 # At most 1 `tick-*` feature can be enabled. If none is enabled, a default of 1MHz is used.
@@ -107,15 +134,21 @@ log = { version = "0.4.14", optional = true }
 
 embedded-hal-02 = { package = "embedded-hal", version = "0.2.6" }
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9", optional = true}
-embedded-hal-async = { version = "=0.1.0-alpha.2", optional = true}
+embedded-hal-async = { version = "=0.1.0-alpha.3", optional = true}
 
 futures-util = { version = "0.3.17", default-features = false }
-embassy-macros  = { version = "0.1.0", path = "../embassy-macros"}
+embassy-sync = { version = "0.1", path = "../embassy-sync" }
 atomic-polyfill = "1.0.1"
 critical-section = "1.1"
 cfg-if = "1.0.0"
+heapless = "0.7"
 
 # WASM dependencies
 wasm-bindgen = { version = "0.2.81", optional = true }
 js-sys = { version = "0.3", optional = true }
 wasm-timer = { version = "0.2.5", optional = true }
+
+[dev-dependencies]
+serial_test = "0.9"
+critical-section = { version = "1.1", features = ["std"] }
+
diff --git a/embassy-time/src/driver.rs b/embassy-time/src/driver.rs
index 79ae14b91..5c2ad3b23 100644
--- a/embassy-time/src/driver.rs
+++ b/embassy-time/src/driver.rs
@@ -105,20 +105,21 @@ pub trait Driver: Send + Sync + 'static {
     /// Sets an alarm at the given timestamp. When the current timestamp reaches the alarm
     /// timestamp, the provided callback function will be called.
     ///
-    /// If `timestamp` is already in the past, the alarm callback must be immediately fired.
-    /// In this case, it is allowed (but not mandatory) to call the alarm callback synchronously from `set_alarm`.
+    /// The `Driver` implementation should guarantee that the alarm callback is never called synchronously from `set_alarm`.
+    /// Rather - if `timestamp` is already in the past - `false` should be returned and alarm should not be set,
+    /// or alternatively, the driver should return `true` and arrange to call the alarm callback as soon as possible, but not synchronously.
     ///
     /// When callback is called, it is guaranteed that now() will return a value greater or equal than timestamp.
     ///
     /// Only one alarm can be active at a time for each AlarmHandle. This overwrites any previously-set alarm if any.
-    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64);
+    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) -> bool;
 }
 
 extern "Rust" {
     fn _embassy_time_now() -> u64;
     fn _embassy_time_allocate_alarm() -> Option<AlarmHandle>;
     fn _embassy_time_set_alarm_callback(alarm: AlarmHandle, callback: fn(*mut ()), ctx: *mut ());
-    fn _embassy_time_set_alarm(alarm: AlarmHandle, timestamp: u64);
+    fn _embassy_time_set_alarm(alarm: AlarmHandle, timestamp: u64) -> bool;
 }
 
 /// See [`Driver::now`]
@@ -139,7 +140,7 @@ pub fn set_alarm_callback(alarm: AlarmHandle, callback: fn(*mut ()), ctx: *mut (
 }
 
 /// See [`Driver::set_alarm`]
-pub fn set_alarm(alarm: AlarmHandle, timestamp: u64) {
+pub fn set_alarm(alarm: AlarmHandle, timestamp: u64) -> bool {
     unsafe { _embassy_time_set_alarm(alarm, timestamp) }
 }
 
@@ -167,7 +168,7 @@ macro_rules! time_driver_impl {
         }
 
         #[no_mangle]
-        fn _embassy_time_set_alarm(alarm: $crate::driver::AlarmHandle, timestamp: u64) {
+        fn _embassy_time_set_alarm(alarm: $crate::driver::AlarmHandle, timestamp: u64) -> bool {
             <$t as $crate::driver::Driver>::set_alarm(&$name, alarm, timestamp)
         }
     };
diff --git a/embassy-time/src/driver_std.rs b/embassy-time/src/driver_std.rs
index 2ddb2e604..fc7fd1979 100644
--- a/embassy-time/src/driver_std.rs
+++ b/embassy-time/src/driver_std.rs
@@ -127,12 +127,14 @@ impl Driver for TimeDriver {
         alarm.ctx = ctx;
     }
 
-    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) {
+    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) -> bool {
         self.init();
         let mut alarms = unsafe { self.alarms.as_ref() }.lock().unwrap();
         let alarm = &mut alarms[alarm.id() as usize];
         alarm.timestamp = timestamp;
         unsafe { self.signaler.as_ref() }.signal();
+
+        true
     }
 }
 
diff --git a/embassy-time/src/driver_wasm.rs b/embassy-time/src/driver_wasm.rs
index e4497e6a2..63d049897 100644
--- a/embassy-time/src/driver_wasm.rs
+++ b/embassy-time/src/driver_wasm.rs
@@ -90,15 +90,23 @@ impl Driver for TimeDriver {
         }));
     }
 
-    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) {
+    fn set_alarm(&self, alarm: AlarmHandle, timestamp: u64) -> bool {
         self.init();
         let mut alarms = unsafe { self.alarms.as_ref() }.lock().unwrap();
         let alarm = &mut alarms[alarm.id() as usize];
-        let timeout = (timestamp - self.now()) as u32;
         if let Some(token) = alarm.token {
             clearTimeout(token);
         }
-        alarm.token = Some(setTimeout(alarm.closure.as_ref().unwrap(), timeout / 1000));
+
+        let now = self.now();
+        if timestamp <= now {
+            false
+        } else {
+            let timeout = (timestamp - now) as u32;
+            alarm.token = Some(setTimeout(alarm.closure.as_ref().unwrap(), timeout / 1000));
+
+            true
+        }
     }
 }
 
diff --git a/embassy-time/src/lib.rs b/embassy-time/src/lib.rs
index 4edc883fe..586aa28de 100644
--- a/embassy-time/src/lib.rs
+++ b/embassy-time/src/lib.rs
@@ -1,4 +1,4 @@
-#![cfg_attr(not(any(feature = "std", feature = "wasm")), no_std)]
+#![cfg_attr(not(any(feature = "std", feature = "wasm", test)), no_std)]
 #![cfg_attr(feature = "nightly", feature(type_alias_impl_trait))]
 #![doc = include_str!("../README.md")]
 #![allow(clippy::new_without_default)]
@@ -11,6 +11,7 @@ mod delay;
 pub mod driver;
 mod duration;
 mod instant;
+pub mod queue;
 mod tick;
 mod timer;
 
@@ -18,6 +19,8 @@ mod timer;
 mod driver_std;
 #[cfg(feature = "wasm")]
 mod driver_wasm;
+#[cfg(feature = "generic-queue")]
+mod queue_generic;
 
 pub use delay::{block_for, Delay};
 pub use duration::Duration;
diff --git a/embassy-time/src/queue.rs b/embassy-time/src/queue.rs
new file mode 100644
index 000000000..c6f8b440a
--- /dev/null
+++ b/embassy-time/src/queue.rs
@@ -0,0 +1,58 @@
+//! Timer queue implementation
+//!
+//! This module defines the interface a timer queue needs to implement to power the `embassy_time` module.
+//!
+//! # Implementing a timer queue
+//!
+//! - Define a struct `MyTimerQueue`
+//! - Implement [`TimerQueue`] for it
+//! - Register it as the global timer queue with [`timer_queue_impl`](crate::timer_queue_impl).
+//!
+//! # Linkage details
+//!
+//! Check the documentation of the [`driver`](crate::driver) module for more information.
+//!
+//! Similarly to driver, if there is none or multiple timer queues in the crate tree, linking will fail.
+//!
+//! # Example
+//!
+//! ```
+//! use core::task::Waker;
+//!
+//! use embassy_time::Instant;
+//! use embassy_time::queue::{TimerQueue};
+//!
+//! struct MyTimerQueue{}; // not public!
+//! embassy_time::timer_queue_impl!(static QUEUE: MyTimerQueue = MyTimerQueue{});
+//!
+//! impl TimerQueue for MyTimerQueue {
+//!     fn schedule_wake(&'static self, at: Instant, waker: &Waker) {
+//!         todo!()
+//!     }
+//! }
+//! ```
+use core::task::Waker;
+
+use crate::Instant;
+
+/// Timer queue
+pub trait TimerQueue {
+    /// Schedules a waker in the queue to be awoken at moment `at`.
+    /// If this moment is in the past, the waker might be awoken immediately.
+    fn schedule_wake(&'static self, at: Instant, waker: &Waker);
+}
+
+/// Set the TimerQueue implementation.
+///
+/// See the module documentation for an example.
+#[macro_export]
+macro_rules! timer_queue_impl {
+    (static $name:ident: $t: ty = $val:expr) => {
+        static $name: $t = $val;
+
+        #[no_mangle]
+        fn _embassy_time_schedule_wake(at: $crate::Instant, waker: &core::task::Waker) {
+            <$t as $crate::queue::TimerQueue>::schedule_wake(&$name, at, waker);
+        }
+    };
+}
diff --git a/embassy-time/src/queue_generic.rs b/embassy-time/src/queue_generic.rs
new file mode 100644
index 000000000..20ae7e6cc
--- /dev/null
+++ b/embassy-time/src/queue_generic.rs
@@ -0,0 +1,449 @@
+use core::cell::RefCell;
+use core::cmp::{min, Ordering};
+use core::task::Waker;
+
+use embassy_sync::blocking_mutex::raw::CriticalSectionRawMutex;
+use embassy_sync::blocking_mutex::Mutex;
+use heapless::Vec;
+
+use crate::driver::{allocate_alarm, set_alarm, set_alarm_callback, AlarmHandle};
+use crate::queue::TimerQueue;
+use crate::Instant;
+
+#[cfg(feature = "generic-queue-8")]
+const QUEUE_SIZE: usize = 8;
+#[cfg(feature = "generic-queue-16")]
+const QUEUE_SIZE: usize = 16;
+#[cfg(feature = "generic-queue-32")]
+const QUEUE_SIZE: usize = 32;
+#[cfg(feature = "generic-queue-64")]
+const QUEUE_SIZE: usize = 32;
+#[cfg(feature = "generic-queue-128")]
+const QUEUE_SIZE: usize = 128;
+#[cfg(not(any(
+    feature = "generic-queue-8",
+    feature = "generic-queue-16",
+    feature = "generic-queue-32",
+    feature = "generic-queue-64",
+    feature = "generic-queue-128"
+)))]
+const QUEUE_SIZE: usize = 64;
+
+#[derive(Debug)]
+struct Timer {
+    at: Instant,
+    waker: Waker,
+}
+
+impl PartialEq for Timer {
+    fn eq(&self, other: &Self) -> bool {
+        self.at == other.at
+    }
+}
+
+impl Eq for Timer {}
+
+impl PartialOrd for Timer {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        self.at.partial_cmp(&other.at)
+    }
+}
+
+impl Ord for Timer {
+    fn cmp(&self, other: &Self) -> Ordering {
+        self.at.cmp(&other.at)
+    }
+}
+
+struct InnerQueue {
+    queue: Vec<Timer, QUEUE_SIZE>,
+    alarm: AlarmHandle,
+}
+
+impl InnerQueue {
+    fn schedule_wake(&mut self, at: Instant, waker: &Waker) {
+        self.queue
+            .iter_mut()
+            .find(|timer| timer.waker.will_wake(waker))
+            .map(|mut timer| {
+                timer.at = min(timer.at, at);
+            })
+            .unwrap_or_else(|| {
+                let mut timer = Timer {
+                    waker: waker.clone(),
+                    at,
+                };
+
+                loop {
+                    match self.queue.push(timer) {
+                        Ok(()) => break,
+                        Err(e) => timer = e,
+                    }
+
+                    self.queue.pop().unwrap().waker.wake();
+                }
+            });
+
+        // Don't wait for the alarm callback to trigger and directly
+        // dispatch all timers that are already due
+        //
+        // Then update the alarm if necessary
+        self.dispatch();
+    }
+
+    fn dispatch(&mut self) {
+        loop {
+            let now = Instant::now();
+
+            let mut next_alarm = Instant::MAX;
+
+            let mut i = 0;
+            while i < self.queue.len() {
+                let timer = &self.queue[i];
+                if timer.at <= now {
+                    let timer = self.queue.swap_remove(i);
+                    timer.waker.wake();
+                } else {
+                    next_alarm = min(next_alarm, timer.at);
+                    i += 1;
+                }
+            }
+
+            if self.update_alarm(next_alarm) {
+                break;
+            }
+        }
+    }
+
+    fn update_alarm(&mut self, next_alarm: Instant) -> bool {
+        if next_alarm == Instant::MAX {
+            true
+        } else {
+            set_alarm(self.alarm, next_alarm.as_ticks())
+        }
+    }
+
+    fn handle_alarm(&mut self) {
+        self.dispatch();
+    }
+}
+
+struct Queue {
+    inner: Mutex<CriticalSectionRawMutex, RefCell<Option<InnerQueue>>>,
+}
+
+impl Queue {
+    const fn new() -> Self {
+        Self {
+            inner: Mutex::new(RefCell::new(None)),
+        }
+    }
+
+    fn schedule_wake(&'static self, at: Instant, waker: &Waker) {
+        self.inner.lock(|inner| {
+            let mut inner = inner.borrow_mut();
+
+            if inner.is_none() {}
+
+            inner
+                .get_or_insert_with(|| {
+                    let handle = unsafe { allocate_alarm() }.unwrap();
+                    set_alarm_callback(handle, Self::handle_alarm_callback, self as *const _ as _);
+                    InnerQueue {
+                        queue: Vec::new(),
+                        alarm: handle,
+                    }
+                })
+                .schedule_wake(at, waker)
+        });
+    }
+
+    fn handle_alarm(&self) {
+        self.inner
+            .lock(|inner| inner.borrow_mut().as_mut().unwrap().handle_alarm());
+    }
+
+    fn handle_alarm_callback(ctx: *mut ()) {
+        unsafe { (ctx as *const Self).as_ref().unwrap() }.handle_alarm();
+    }
+}
+
+impl TimerQueue for Queue {
+    fn schedule_wake(&'static self, at: Instant, waker: &Waker) {
+        Queue::schedule_wake(self, at, waker);
+    }
+}
+
+crate::timer_queue_impl!(static QUEUE: Queue = Queue::new());
+
+#[cfg(test)]
+mod tests {
+    use core::cell::Cell;
+    use core::task::{RawWaker, RawWakerVTable, Waker};
+    use std::rc::Rc;
+    use std::sync::Mutex;
+
+    use serial_test::serial;
+
+    use super::InnerQueue;
+    use crate::driver::{AlarmHandle, Driver};
+    use crate::queue_generic::QUEUE;
+    use crate::Instant;
+
+    struct InnerTestDriver {
+        now: u64,
+        alarm: u64,
+        callback: fn(*mut ()),
+        ctx: *mut (),
+    }
+
+    impl InnerTestDriver {
+        const fn new() -> Self {
+            Self {
+                now: 0,
+                alarm: u64::MAX,
+                callback: Self::noop,
+                ctx: core::ptr::null_mut(),
+            }
+        }
+
+        fn noop(_ctx: *mut ()) {}
+    }
+
+    unsafe impl Send for InnerTestDriver {}
+
+    struct TestDriver(Mutex<InnerTestDriver>);
+
+    impl TestDriver {
+        const fn new() -> Self {
+            Self(Mutex::new(InnerTestDriver::new()))
+        }
+
+        fn reset(&self) {
+            *self.0.lock().unwrap() = InnerTestDriver::new();
+        }
+
+        fn set_now(&self, now: u64) {
+            let notify = {
+                let mut inner = self.0.lock().unwrap();
+
+                if inner.now < now {
+                    inner.now = now;
+
+                    if inner.alarm <= now {
+                        inner.alarm = u64::MAX;
+
+                        Some((inner.callback, inner.ctx))
+                    } else {
+                        None
+                    }
+                } else {
+                    panic!("Going back in time?");
+                }
+            };
+
+            if let Some((callback, ctx)) = notify {
+                (callback)(ctx);
+            }
+        }
+    }
+
+    impl Driver for TestDriver {
+        fn now(&self) -> u64 {
+            self.0.lock().unwrap().now
+        }
+
+        unsafe fn allocate_alarm(&self) -> Option<AlarmHandle> {
+            Some(AlarmHandle::new(0))
+        }
+
+        fn set_alarm_callback(&self, _alarm: AlarmHandle, callback: fn(*mut ()), ctx: *mut ()) {
+            let mut inner = self.0.lock().unwrap();
+
+            inner.callback = callback;
+            inner.ctx = ctx;
+        }
+
+        fn set_alarm(&self, _alarm: AlarmHandle, timestamp: u64) -> bool {
+            let mut inner = self.0.lock().unwrap();
+
+            if timestamp <= inner.now {
+                false
+            } else {
+                inner.alarm = timestamp;
+                true
+            }
+        }
+    }
+
+    struct TestWaker {
+        pub awoken: Rc<Cell<bool>>,
+        pub waker: Waker,
+    }
+
+    impl TestWaker {
+        fn new() -> Self {
+            let flag = Rc::new(Cell::new(false));
+
+            const VTABLE: RawWakerVTable = RawWakerVTable::new(
+                |data: *const ()| {
+                    unsafe {
+                        Rc::increment_strong_count(data as *const Cell<bool>);
+                    }
+
+                    RawWaker::new(data as _, &VTABLE)
+                },
+                |data: *const ()| unsafe {
+                    let data = data as *const Cell<bool>;
+                    data.as_ref().unwrap().set(true);
+                    Rc::decrement_strong_count(data);
+                },
+                |data: *const ()| unsafe {
+                    (data as *const Cell<bool>).as_ref().unwrap().set(true);
+                },
+                |data: *const ()| unsafe {
+                    Rc::decrement_strong_count(data);
+                },
+            );
+
+            let raw = RawWaker::new(Rc::into_raw(flag.clone()) as _, &VTABLE);
+
+            Self {
+                awoken: flag.clone(),
+                waker: unsafe { Waker::from_raw(raw) },
+            }
+        }
+    }
+
+    crate::time_driver_impl!(static DRIVER: TestDriver = TestDriver::new());
+
+    fn setup() {
+        DRIVER.reset();
+
+        QUEUE.inner.lock(|inner| {
+            *inner.borrow_mut() = InnerQueue::new();
+        });
+    }
+
+    fn queue_len() -> usize {
+        QUEUE.inner.lock(|inner| inner.borrow().queue.iter().count())
+    }
+
+    #[test]
+    #[serial]
+    fn test_schedule() {
+        setup();
+
+        assert_eq!(queue_len(), 0);
+
+        let waker = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(1), &waker.waker);
+
+        assert!(!waker.awoken.get());
+        assert_eq!(queue_len(), 1);
+    }
+
+    #[test]
+    #[serial]
+    fn test_schedule_same() {
+        setup();
+
+        let waker = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(1), &waker.waker);
+
+        assert_eq!(queue_len(), 1);
+
+        QUEUE.schedule_wake(Instant::from_secs(1), &waker.waker);
+
+        assert_eq!(queue_len(), 1);
+
+        QUEUE.schedule_wake(Instant::from_secs(100), &waker.waker);
+
+        assert_eq!(queue_len(), 1);
+
+        let waker2 = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(100), &waker2.waker);
+
+        assert_eq!(queue_len(), 2);
+    }
+
+    #[test]
+    #[serial]
+    fn test_trigger() {
+        setup();
+
+        let waker = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(100), &waker.waker);
+
+        assert!(!waker.awoken.get());
+
+        DRIVER.set_now(Instant::from_secs(99).as_ticks());
+
+        assert!(!waker.awoken.get());
+
+        assert_eq!(queue_len(), 1);
+
+        DRIVER.set_now(Instant::from_secs(100).as_ticks());
+
+        assert!(waker.awoken.get());
+
+        assert_eq!(queue_len(), 0);
+    }
+
+    #[test]
+    #[serial]
+    fn test_immediate_trigger() {
+        setup();
+
+        let waker = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(100), &waker.waker);
+
+        DRIVER.set_now(Instant::from_secs(50).as_ticks());
+
+        let waker2 = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(40), &waker2.waker);
+
+        assert!(!waker.awoken.get());
+        assert!(waker2.awoken.get());
+        assert_eq!(queue_len(), 1);
+    }
+
+    #[test]
+    #[serial]
+    fn test_queue_overflow() {
+        setup();
+
+        for i in 1..super::QUEUE_SIZE {
+            let waker = TestWaker::new();
+
+            QUEUE.schedule_wake(Instant::from_secs(310), &waker.waker);
+
+            assert_eq!(queue_len(), i);
+            assert!(!waker.awoken.get());
+        }
+
+        let first_waker = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(300), &first_waker.waker);
+
+        assert_eq!(queue_len(), super::QUEUE_SIZE);
+        assert!(!first_waker.awoken.get());
+
+        let second_waker = TestWaker::new();
+
+        QUEUE.schedule_wake(Instant::from_secs(305), &second_waker.waker);
+
+        assert_eq!(queue_len(), super::QUEUE_SIZE);
+        assert!(first_waker.awoken.get());
+
+        QUEUE.schedule_wake(Instant::from_secs(320), &TestWaker::new().waker);
+        assert_eq!(queue_len(), super::QUEUE_SIZE);
+        assert!(second_waker.awoken.get());
+    }
+}
diff --git a/examples/nrf/Cargo.toml b/examples/nrf/Cargo.toml
index 6949042e2..c633f82f5 100644
--- a/examples/nrf/Cargo.toml
+++ b/examples/nrf/Cargo.toml
@@ -17,7 +17,7 @@ embassy-time = { version = "0.1.0", path = "../../embassy-time", features = ["de
 embassy-nrf = { version = "0.1.0", path = "../../embassy-nrf", features = ["defmt", "nrf52840", "time-driver-rtc1", "gpiote", "unstable-pac"] }
 embassy-net = { version = "0.1.0", path = "../../embassy-net", features = ["defmt", "tcp", "dhcpv4", "medium-ethernet", "pool-16"], optional = true }
 embassy-usb = { version = "0.1.0", path = "../../embassy-usb", features = ["defmt"], optional = true }
-embedded-io = "0.3.0"
+embedded-io = "0.3.1"
 embassy-lora = { version = "0.1.0", path = "../../embassy-lora", features = ["sx126x", "time", "defmt"], optional = true }
 
 lorawan-device = { version = "0.8.0", default-features = false, features = ["async"], optional = true }
diff --git a/examples/rp/Cargo.toml b/examples/rp/Cargo.toml
index 38355bbf8..31f688305 100644
--- a/examples/rp/Cargo.toml
+++ b/examples/rp/Cargo.toml
@@ -28,8 +28,8 @@ display-interface = "0.4.1"
 byte-slice-cast = { version = "1.2.0", default-features = false }
 
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "0.1.0-alpha.1" }
-embedded-io = { version = "0.3.0", features = ["async", "defmt"] }
+embedded-hal-async = { version = "0.1.0-alpha.3" }
+embedded-io = { version = "0.3.1", features = ["async", "defmt"] }
 embedded-storage = { version = "0.3" }
 static_cell = "1.0.0"
 
diff --git a/examples/std/Cargo.toml b/examples/std/Cargo.toml
index b9bd1e718..790258382 100644
--- a/examples/std/Cargo.toml
+++ b/examples/std/Cargo.toml
@@ -9,7 +9,7 @@ embassy-sync = { version = "0.1.0", path = "../../embassy-sync", features = ["lo
 embassy-executor = { version = "0.1.0", path = "../../embassy-executor", features = ["log", "std", "nightly", "integrated-timers"] }
 embassy-time = { version = "0.1.0", path = "../../embassy-time", features = ["log", "std", "nightly"] }
 embassy-net = { version = "0.1.0", path = "../../embassy-net", features=[ "std", "nightly", "log", "medium-ethernet", "tcp", "udp", "dhcpv4", "pool-16"] }
-embedded-io = { version = "0.3.0", features = ["async", "std", "futures"] }
+embedded-io = { version = "0.3.1", features = ["async", "std", "futures"] }
 critical-section = { version = "1.1", features = ["std"] }
 
 async-io = "1.6.0"
diff --git a/examples/stm32f1/src/bin/adc.rs b/examples/stm32f1/src/bin/adc.rs
index 2d6b4a0e9..ed59e2799 100644
--- a/examples/stm32f1/src/bin/adc.rs
+++ b/examples/stm32f1/src/bin/adc.rs
@@ -16,11 +16,19 @@ async fn main(_spawner: Spawner) {
     let mut adc = Adc::new(p.ADC1, &mut Delay);
     let mut pin = p.PB1;
 
-    let mut vref = adc.enable_vref(&mut Delay);
-    adc.calibrate(&mut vref);
+    let mut vrefint = adc.enable_vref(&mut Delay);
+    let vrefint_sample = adc.read(&mut vrefint);
+    let convert_to_millivolts = |sample| {
+        // From http://www.st.com/resource/en/datasheet/CD00161566.pdf
+        // 5.3.4 Embedded reference voltage
+        const VREFINT_MV: u32 = 1200; // mV
+
+        (u32::from(sample) * VREFINT_MV / u32::from(vrefint_sample)) as u16
+    };
+
     loop {
         let v = adc.read(&mut pin);
-        info!("--> {} - {} mV", v, adc.to_millivolts(v));
+        info!("--> {} - {} mV", v, convert_to_millivolts(v));
         Timer::after(Duration::from_millis(100)).await;
     }
 }
diff --git a/examples/stm32f3/src/bin/usart_dma.rs b/examples/stm32f3/src/bin/usart_dma.rs
index 3bc5a287f..47121acf1 100644
--- a/examples/stm32f3/src/bin/usart_dma.rs
+++ b/examples/stm32f3/src/bin/usart_dma.rs
@@ -7,6 +7,7 @@ use core::fmt::Write;
 use defmt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use heapless::String;
 use {defmt_rtt as _, panic_probe as _};
@@ -17,7 +18,8 @@ async fn main(_spawner: Spawner) {
     info!("Hello World!");
 
     let config = Config::default();
-    let mut usart = Uart::new(p.USART1, p.PE1, p.PE0, p.DMA1_CH4, NoDma, config);
+    let irq = interrupt::take!(USART1);
+    let mut usart = Uart::new(p.USART1, p.PE1, p.PE0, irq, p.DMA1_CH4, NoDma, config);
 
     for n in 0u32.. {
         let mut s: String<128> = String::new();
diff --git a/examples/stm32f4/Cargo.toml b/examples/stm32f4/Cargo.toml
index 6d4f09fba..b05457eaa 100644
--- a/examples/stm32f4/Cargo.toml
+++ b/examples/stm32f4/Cargo.toml
@@ -17,7 +17,7 @@ defmt-rtt = "0.3"
 cortex-m = { version = "0.7.6", features = ["critical-section-single-core"] }
 cortex-m-rt = "0.7.0"
 embedded-hal = "0.2.6"
-embedded-io = "0.3.0"
+embedded-io = "0.3.1"
 panic-probe = { version = "0.3", features = ["print-defmt"] }
 futures = { version = "0.3.17", default-features = false, features = ["async-await"] }
 heapless = { version = "0.7.5", default-features = false }
diff --git a/examples/stm32f4/src/bin/adc.rs b/examples/stm32f4/src/bin/adc.rs
index 1d030f7dc..1c9a0b35d 100644
--- a/examples/stm32f4/src/bin/adc.rs
+++ b/examples/stm32f4/src/bin/adc.rs
@@ -24,19 +24,44 @@ async fn main(_spawner: Spawner) {
     // Startup delay can be combined to the maximum of either
     delay.delay_us(Temperature::start_time_us().max(VrefInt::start_time_us()));
 
+    let vrefint_sample = adc.read_internal(&mut vrefint);
+
+    let convert_to_millivolts = |sample| {
+        // From http://www.st.com/resource/en/datasheet/DM00071990.pdf
+        // 6.3.24 Reference voltage
+        const VREFINT_MV: u32 = 1210; // mV
+
+        (u32::from(sample) * VREFINT_MV / u32::from(vrefint_sample)) as u16
+    };
+
+    let convert_to_celcius = |sample| {
+        // From http://www.st.com/resource/en/datasheet/DM00071990.pdf
+        // 6.3.22 Temperature sensor characteristics
+        const V25: i32 = 760; // mV
+        const AVG_SLOPE: f32 = 2.5; // mV/C
+
+        let sample_mv = convert_to_millivolts(sample) as i32;
+
+        (sample_mv - V25) as f32 / AVG_SLOPE + 25.0
+    };
+
+    info!("VrefInt: {}", vrefint_sample);
+    const MAX_ADC_SAMPLE: u16 = (1 << 12) - 1;
+    info!("VCCA: {} mV", convert_to_millivolts(MAX_ADC_SAMPLE));
+
     loop {
         // Read pin
         let v = adc.read(&mut pin);
-        info!("PC1: {} ({} mV)", v, adc.to_millivolts(v));
+        info!("PC1: {} ({} mV)", v, convert_to_millivolts(v));
 
         // Read internal temperature
         let v = adc.read_internal(&mut temp);
-        let celcius = Temperature::to_celcius(adc.to_millivolts(v));
+        let celcius = convert_to_celcius(v);
         info!("Internal temp: {} ({} C)", v, celcius);
 
         // Read internal voltage reference
         let v = adc.read_internal(&mut vrefint);
-        info!("VrefInt: {} ({} mV)", v, adc.to_millivolts(v));
+        info!("VrefInt: {}", v);
 
         Timer::after(Duration::from_millis(100)).await;
     }
diff --git a/examples/stm32f4/src/bin/i2c.rs b/examples/stm32f4/src/bin/i2c.rs
new file mode 100644
index 000000000..6e51c211d
--- /dev/null
+++ b/examples/stm32f4/src/bin/i2c.rs
@@ -0,0 +1,45 @@
+#![no_std]
+#![no_main]
+#![feature(type_alias_impl_trait)]
+
+use defmt::*;
+use embassy_executor::Spawner;
+use embassy_stm32::dma::NoDma;
+use embassy_stm32::i2c::{Error, I2c, TimeoutI2c};
+use embassy_stm32::interrupt;
+use embassy_stm32::time::Hertz;
+use embassy_time::Duration;
+use {defmt_rtt as _, panic_probe as _};
+
+const ADDRESS: u8 = 0x5F;
+const WHOAMI: u8 = 0x0F;
+
+#[embassy_executor::main]
+async fn main(_spawner: Spawner) -> ! {
+    info!("Hello world!");
+    let p = embassy_stm32::init(Default::default());
+
+    let irq = interrupt::take!(I2C2_EV);
+    let mut i2c = I2c::new(
+        p.I2C2,
+        p.PB10,
+        p.PB11,
+        irq,
+        NoDma,
+        NoDma,
+        Hertz(100_000),
+        Default::default(),
+    );
+
+    // I2C bus can freeze if SCL line is shorted or due to a broken device that clock stretches for too long.
+    // TimeoutI2c allows recovering from such errors by throwing `Error::Timeout` after a given delay.
+    let mut timeout_i2c = TimeoutI2c::new(&mut i2c, Duration::from_millis(1000));
+
+    let mut data = [0u8; 1];
+
+    match timeout_i2c.blocking_write_read(ADDRESS, &[WHOAMI], &mut data) {
+        Ok(()) => info!("Whoami: {}", data[0]),
+        Err(Error::Timeout) => error!("Operation timed out"),
+        Err(e) => error!("I2c Error: {:?}", e),
+    }
+}
diff --git a/examples/stm32f4/src/bin/usart.rs b/examples/stm32f4/src/bin/usart.rs
index 90ad882b8..8f41bb6c4 100644
--- a/examples/stm32f4/src/bin/usart.rs
+++ b/examples/stm32f4/src/bin/usart.rs
@@ -5,6 +5,7 @@
 use cortex_m_rt::entry;
 use defmt::*;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use {defmt_rtt as _, panic_probe as _};
 
@@ -15,7 +16,8 @@ fn main() -> ! {
     let p = embassy_stm32::init(Default::default());
 
     let config = Config::default();
-    let mut usart = Uart::new(p.USART3, p.PD9, p.PD8, NoDma, NoDma, config);
+    let irq = interrupt::take!(USART3);
+    let mut usart = Uart::new(p.USART3, p.PD9, p.PD8, irq, NoDma, NoDma, config);
 
     unwrap!(usart.blocking_write(b"Hello Embassy World!\r\n"));
     info!("wrote Hello, starting echo");
diff --git a/examples/stm32f4/src/bin/usart_buffered.rs b/examples/stm32f4/src/bin/usart_buffered.rs
index 7bcecbd26..dd171fe13 100644
--- a/examples/stm32f4/src/bin/usart_buffered.rs
+++ b/examples/stm32f4/src/bin/usart_buffered.rs
@@ -4,9 +4,8 @@
 
 use defmt::*;
 use embassy_executor::Spawner;
-use embassy_stm32::dma::NoDma;
 use embassy_stm32::interrupt;
-use embassy_stm32::usart::{BufferedUart, Config, State, Uart};
+use embassy_stm32::usart::{BufferedUart, Config, State};
 use embedded_io::asynch::BufRead;
 use {defmt_rtt as _, panic_probe as _};
 
@@ -16,13 +15,21 @@ async fn main(_spawner: Spawner) {
     info!("Hello World!");
 
     let config = Config::default();
-    let usart = Uart::new(p.USART3, p.PD9, p.PD8, NoDma, NoDma, config);
 
     let mut state = State::new();
     let irq = interrupt::take!(USART3);
     let mut tx_buf = [0u8; 32];
     let mut rx_buf = [0u8; 32];
-    let mut buf_usart = BufferedUart::new(&mut state, usart, irq, &mut tx_buf, &mut rx_buf);
+    let mut buf_usart = BufferedUart::new(
+        &mut state,
+        p.USART3,
+        p.PD9,
+        p.PD8,
+        irq,
+        &mut tx_buf,
+        &mut rx_buf,
+        config,
+    );
 
     loop {
         let buf = buf_usart.fill_buf().await.unwrap();
diff --git a/examples/stm32f4/src/bin/usart_dma.rs b/examples/stm32f4/src/bin/usart_dma.rs
index bb41b8b4f..78baeaa0d 100644
--- a/examples/stm32f4/src/bin/usart_dma.rs
+++ b/examples/stm32f4/src/bin/usart_dma.rs
@@ -7,6 +7,7 @@ use core::fmt::Write;
 use defmt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use heapless::String;
 use {defmt_rtt as _, panic_probe as _};
@@ -17,7 +18,8 @@ async fn main(_spawner: Spawner) {
     info!("Hello World!");
 
     let config = Config::default();
-    let mut usart = Uart::new(p.USART3, p.PD9, p.PD8, p.DMA1_CH3, NoDma, config);
+    let irq = interrupt::take!(USART3);
+    let mut usart = Uart::new(p.USART3, p.PD9, p.PD8, irq, p.DMA1_CH3, NoDma, config);
 
     for n in 0u32.. {
         let mut s: String<128> = String::new();
diff --git a/examples/stm32f7/Cargo.toml b/examples/stm32f7/Cargo.toml
index dad92c0fc..b14afd2fe 100644
--- a/examples/stm32f7/Cargo.toml
+++ b/examples/stm32f7/Cargo.toml
@@ -10,7 +10,7 @@ embassy-executor = { version = "0.1.0", path = "../../embassy-executor", feature
 embassy-time = { version = "0.1.0", path = "../../embassy-time", features = ["defmt", "defmt-timestamp-uptime", "tick-hz-32_768"] }
 embassy-stm32 = { version = "0.1.0", path = "../../embassy-stm32", features = ["nightly", "defmt", "net", "stm32f767zi", "unstable-pac", "time-driver-any", "exti"]  }
 embassy-net = { path = "../../embassy-net", features = ["defmt", "nightly", "tcp", "dhcpv4", "medium-ethernet", "pool-16"] }
-embedded-io = { version = "0.3.0", features = ["async"] }
+embedded-io = { version = "0.3.1", features = ["async"] }
 
 defmt = "0.3"
 defmt-rtt = "0.3"
diff --git a/examples/stm32f7/src/bin/adc.rs b/examples/stm32f7/src/bin/adc.rs
index 80fad8c41..70b3b2a75 100644
--- a/examples/stm32f7/src/bin/adc.rs
+++ b/examples/stm32f7/src/bin/adc.rs
@@ -16,9 +16,19 @@ async fn main(_spawner: Spawner) {
     let mut adc = Adc::new(p.ADC1, &mut Delay);
     let mut pin = p.PA3;
 
+    let mut vrefint = adc.enable_vrefint();
+    let vrefint_sample = adc.read_internal(&mut vrefint);
+    let convert_to_millivolts = |sample| {
+        // From http://www.st.com/resource/en/datasheet/DM00273119.pdf
+        // 6.3.27 Reference voltage
+        const VREFINT_MV: u32 = 1210; // mV
+
+        (u32::from(sample) * VREFINT_MV / u32::from(vrefint_sample)) as u16
+    };
+
     loop {
         let v = adc.read(&mut pin);
-        info!("--> {} - {} mV", v, adc.to_millivolts(v));
+        info!("--> {} - {} mV", v, convert_to_millivolts(v));
         Timer::after(Duration::from_millis(100)).await;
     }
 }
diff --git a/examples/stm32f7/src/bin/usart_dma.rs b/examples/stm32f7/src/bin/usart_dma.rs
index 07270479c..4827c52ae 100644
--- a/examples/stm32f7/src/bin/usart_dma.rs
+++ b/examples/stm32f7/src/bin/usart_dma.rs
@@ -7,6 +7,7 @@ use core::fmt::Write;
 use defmt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use heapless::String;
 use {defmt_rtt as _, panic_probe as _};
@@ -15,7 +16,8 @@ use {defmt_rtt as _, panic_probe as _};
 async fn main(_spawner: Spawner) {
     let p = embassy_stm32::init(Default::default());
     let config = Config::default();
-    let mut usart = Uart::new(p.UART7, p.PA8, p.PA15, p.DMA1_CH1, NoDma, config);
+    let irq = interrupt::take!(UART7);
+    let mut usart = Uart::new(p.UART7, p.PA8, p.PA15, irq, p.DMA1_CH1, NoDma, config);
 
     for n in 0u32.. {
         let mut s: String<128> = String::new();
diff --git a/examples/stm32h7/Cargo.toml b/examples/stm32h7/Cargo.toml
index 1a05b9ecb..0dccff6e8 100644
--- a/examples/stm32h7/Cargo.toml
+++ b/examples/stm32h7/Cargo.toml
@@ -10,7 +10,7 @@ embassy-executor = { version = "0.1.0", path = "../../embassy-executor", feature
 embassy-time = { version = "0.1.0", path = "../../embassy-time", features = ["defmt", "defmt-timestamp-uptime", "unstable-traits", "tick-hz-32_768"] }
 embassy-stm32 = { version = "0.1.0", path = "../../embassy-stm32", features = ["nightly", "defmt", "stm32h743bi", "net", "time-driver-any", "exti", "unstable-pac", "unstable-traits"] }
 embassy-net = { path = "../../embassy-net", features = ["defmt", "nightly", "tcp", "dhcpv4", "medium-ethernet", "pool-16", "unstable-traits"] }
-embedded-io = { version = "0.3.0", features = ["async"] }
+embedded-io = { version = "0.3.1", features = ["async"] }
 
 defmt = "0.3"
 defmt-rtt = "0.3"
@@ -19,7 +19,7 @@ cortex-m = { version = "0.7.6", features = ["critical-section-single-core"] }
 cortex-m-rt = "0.7.0"
 embedded-hal = "0.2.6"
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "=0.1.0-alpha.2" }
+embedded-hal-async = { version = "=0.1.0-alpha.3" }
 embedded-nal-async = "0.2.0"
 panic-probe = { version = "0.3", features = ["print-defmt"] }
 futures = { version = "0.3.17", default-features = false, features = ["async-await"] }
diff --git a/examples/stm32h7/src/bin/i2c.rs b/examples/stm32h7/src/bin/i2c.rs
new file mode 100644
index 000000000..d44319ae6
--- /dev/null
+++ b/examples/stm32h7/src/bin/i2c.rs
@@ -0,0 +1,44 @@
+#![no_std]
+#![no_main]
+#![feature(type_alias_impl_trait)]
+
+use defmt::*;
+use embassy_executor::Spawner;
+use embassy_stm32::i2c::{Error, I2c, TimeoutI2c};
+use embassy_stm32::interrupt;
+use embassy_stm32::time::Hertz;
+use embassy_time::Duration;
+use {defmt_rtt as _, panic_probe as _};
+
+const ADDRESS: u8 = 0x5F;
+const WHOAMI: u8 = 0x0F;
+
+#[embassy_executor::main]
+async fn main(_spawner: Spawner) -> ! {
+    info!("Hello world!");
+    let p = embassy_stm32::init(Default::default());
+
+    let irq = interrupt::take!(I2C2_EV);
+    let mut i2c = I2c::new(
+        p.I2C2,
+        p.PB10,
+        p.PB11,
+        irq,
+        p.DMA1_CH4,
+        p.DMA1_CH5,
+        Hertz(100_000),
+        Default::default(),
+    );
+
+    // I2C bus can freeze if SCL line is shorted or due to a broken device that clock stretches for too long.
+    // TimeoutI2c allows recovering from such errors by throwing `Error::Timeout` after a given delay.
+    let mut timeout_i2c = TimeoutI2c::new(&mut i2c, Duration::from_millis(1000));
+
+    let mut data = [0u8; 1];
+
+    match timeout_i2c.blocking_write_read(ADDRESS, &[WHOAMI], &mut data) {
+        Ok(()) => info!("Whoami: {}", data[0]),
+        Err(Error::Timeout) => error!("Operation timed out"),
+        Err(e) => error!("I2c Error: {:?}", e),
+    }
+}
diff --git a/examples/stm32h7/src/bin/usart.rs b/examples/stm32h7/src/bin/usart.rs
index 87c2b1253..405f18ec7 100644
--- a/examples/stm32h7/src/bin/usart.rs
+++ b/examples/stm32h7/src/bin/usart.rs
@@ -6,6 +6,7 @@ use cortex_m_rt::entry;
 use defmt::*;
 use embassy_executor::Executor;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use static_cell::StaticCell;
 use {defmt_rtt as _, panic_probe as _};
@@ -15,7 +16,8 @@ async fn main_task() {
     let p = embassy_stm32::init(Default::default());
 
     let config = Config::default();
-    let mut usart = Uart::new(p.UART7, p.PF6, p.PF7, NoDma, NoDma, config);
+    let irq = interrupt::take!(UART7);
+    let mut usart = Uart::new(p.UART7, p.PF6, p.PF7, irq, NoDma, NoDma, config);
 
     unwrap!(usart.blocking_write(b"Hello Embassy World!\r\n"));
     info!("wrote Hello, starting echo");
diff --git a/examples/stm32h7/src/bin/usart_dma.rs b/examples/stm32h7/src/bin/usart_dma.rs
index 3adffcbeb..6e3491e55 100644
--- a/examples/stm32h7/src/bin/usart_dma.rs
+++ b/examples/stm32h7/src/bin/usart_dma.rs
@@ -8,6 +8,7 @@ use cortex_m_rt::entry;
 use defmt::*;
 use embassy_executor::Executor;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use heapless::String;
 use static_cell::StaticCell;
@@ -18,7 +19,8 @@ async fn main_task() {
     let p = embassy_stm32::init(Default::default());
 
     let config = Config::default();
-    let mut usart = Uart::new(p.UART7, p.PF6, p.PF7, p.DMA1_CH0, NoDma, config);
+    let irq = interrupt::take!(UART7);
+    let mut usart = Uart::new(p.UART7, p.PF6, p.PF7, irq, p.DMA1_CH0, NoDma, config);
 
     for n in 0u32.. {
         let mut s: String<128> = String::new();
diff --git a/examples/stm32h7/src/bin/usart_split.rs b/examples/stm32h7/src/bin/usart_split.rs
index df2b600f8..f97176ecb 100644
--- a/examples/stm32h7/src/bin/usart_split.rs
+++ b/examples/stm32h7/src/bin/usart_split.rs
@@ -5,6 +5,7 @@
 use defmt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::peripherals::{DMA1_CH1, UART7};
 use embassy_stm32::usart::{Config, Uart, UartRx};
 use embassy_sync::blocking_mutex::raw::ThreadModeRawMutex;
@@ -31,7 +32,8 @@ async fn main(spawner: Spawner) -> ! {
     info!("Hello World!");
 
     let config = Config::default();
-    let mut usart = Uart::new(p.UART7, p.PF6, p.PF7, p.DMA1_CH0, p.DMA1_CH1, config);
+    let irq = interrupt::take!(UART7);
+    let mut usart = Uart::new(p.UART7, p.PF6, p.PF7, irq, p.DMA1_CH0, p.DMA1_CH1, config);
     unwrap!(usart.blocking_write(b"Type 8 chars to echo!\r\n"));
 
     let (mut tx, rx) = usart.split();
diff --git a/examples/stm32l0/Cargo.toml b/examples/stm32l0/Cargo.toml
index 7e1120f48..8b00773be 100644
--- a/examples/stm32l0/Cargo.toml
+++ b/examples/stm32l0/Cargo.toml
@@ -22,7 +22,7 @@ defmt = "0.3"
 defmt-rtt = "0.3"
 
 embedded-storage = "0.3.0"
-embedded-io = "0.3.0"
+embedded-io = "0.3.1"
 
 cortex-m = { version = "0.7.6", features = ["critical-section-single-core"] }
 cortex-m-rt = "0.7.0"
diff --git a/examples/stm32l0/src/bin/usart_dma.rs b/examples/stm32l0/src/bin/usart_dma.rs
index 66657d0f0..c307f857a 100644
--- a/examples/stm32l0/src/bin/usart_dma.rs
+++ b/examples/stm32l0/src/bin/usart_dma.rs
@@ -4,13 +4,15 @@
 
 use defmt::*;
 use embassy_executor::Spawner;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use {defmt_rtt as _, panic_probe as _};
 
 #[embassy_executor::main]
 async fn main(_spawner: Spawner) {
     let p = embassy_stm32::init(Default::default());
-    let mut usart = Uart::new(p.USART1, p.PB7, p.PB6, p.DMA1_CH2, p.DMA1_CH3, Config::default());
+    let irq = interrupt::take!(USART1);
+    let mut usart = Uart::new(p.USART1, p.PB7, p.PB6, irq, p.DMA1_CH2, p.DMA1_CH3, Config::default());
 
     usart.write(b"Hello Embassy World!\r\n").await.unwrap();
     info!("wrote Hello, starting echo");
diff --git a/examples/stm32l0/src/bin/usart_irq.rs b/examples/stm32l0/src/bin/usart_irq.rs
index 0e2237388..8e84cd092 100644
--- a/examples/stm32l0/src/bin/usart_irq.rs
+++ b/examples/stm32l0/src/bin/usart_irq.rs
@@ -4,9 +4,8 @@
 
 use defmt::*;
 use embassy_executor::Spawner;
-use embassy_stm32::dma::NoDma;
 use embassy_stm32::interrupt;
-use embassy_stm32::usart::{BufferedUart, Config, State, Uart};
+use embassy_stm32::usart::{BufferedUart, Config, State};
 use embedded_io::asynch::{Read, Write};
 use {defmt_rtt as _, panic_probe as _};
 
@@ -21,15 +20,18 @@ async fn main(_spawner: Spawner) {
     let mut config = Config::default();
     config.baudrate = 9600;
 
-    let usart = Uart::new(p.USART2, p.PA3, p.PA2, NoDma, NoDma, config);
     let mut state = State::new();
+    let irq = interrupt::take!(USART2);
     let mut usart = unsafe {
         BufferedUart::new(
             &mut state,
-            usart,
-            interrupt::take!(USART2),
+            p.USART2,
+            p.PA3,
+            p.PA2,
+            irq,
             &mut TX_BUFFER,
             &mut RX_BUFFER,
+            config,
         )
     };
 
diff --git a/examples/stm32l4/Cargo.toml b/examples/stm32l4/Cargo.toml
index 657605ebe..83d456b26 100644
--- a/examples/stm32l4/Cargo.toml
+++ b/examples/stm32l4/Cargo.toml
@@ -20,7 +20,7 @@ cortex-m = { version = "0.7.6", features = ["critical-section-single-core"] }
 cortex-m-rt = "0.7.0"
 embedded-hal = "0.2.6"
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "=0.1.0-alpha.2" }
+embedded-hal-async = { version = "=0.1.0-alpha.3" }
 panic-probe = { version = "0.3", features = ["print-defmt"] }
 futures = { version = "0.3.17", default-features = false, features = ["async-await"] }
 heapless = { version = "0.7.5", default-features = false }
diff --git a/examples/stm32l4/src/bin/usart.rs b/examples/stm32l4/src/bin/usart.rs
index 4a4b46c53..7d874d9d7 100644
--- a/examples/stm32l4/src/bin/usart.rs
+++ b/examples/stm32l4/src/bin/usart.rs
@@ -4,6 +4,7 @@
 
 use defmt::*;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use {defmt_rtt as _, panic_probe as _};
 
@@ -14,7 +15,8 @@ fn main() -> ! {
     let p = embassy_stm32::init(Default::default());
 
     let config = Config::default();
-    let mut usart = Uart::new(p.UART4, p.PA1, p.PA0, NoDma, NoDma, config);
+    let irq = interrupt::take!(UART4);
+    let mut usart = Uart::new(p.UART4, p.PA1, p.PA0, irq, NoDma, NoDma, config);
 
     unwrap!(usart.blocking_write(b"Hello Embassy World!\r\n"));
     info!("wrote Hello, starting echo");
diff --git a/examples/stm32l4/src/bin/usart_dma.rs b/examples/stm32l4/src/bin/usart_dma.rs
index 728906897..452bede30 100644
--- a/examples/stm32l4/src/bin/usart_dma.rs
+++ b/examples/stm32l4/src/bin/usart_dma.rs
@@ -7,6 +7,7 @@ use core::fmt::Write;
 use defmt::*;
 use embassy_executor::Spawner;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use heapless::String;
 use {defmt_rtt as _, panic_probe as _};
@@ -17,7 +18,8 @@ async fn main(_spawner: Spawner) {
     info!("Hello World!");
 
     let config = Config::default();
-    let mut usart = Uart::new(p.UART4, p.PA1, p.PA0, p.DMA1_CH3, NoDma, config);
+    let irq = interrupt::take!(UART4);
+    let mut usart = Uart::new(p.UART4, p.PA1, p.PA0, irq, p.DMA1_CH3, NoDma, config);
 
     for n in 0u32.. {
         let mut s: String<128> = String::new();
diff --git a/examples/stm32l5/Cargo.toml b/examples/stm32l5/Cargo.toml
index 63eac3ed2..848723f8b 100644
--- a/examples/stm32l5/Cargo.toml
+++ b/examples/stm32l5/Cargo.toml
@@ -26,5 +26,5 @@ embedded-hal = "0.2.6"
 futures = { version = "0.3.17", default-features = false, features = ["async-await"] }
 heapless = { version = "0.7.5", default-features = false }
 rand_core = { version = "0.6.3", default-features = false }
-embedded-io = { version = "0.3.0", features = ["async"] }
+embedded-io = { version = "0.3.1", features = ["async"] }
 static_cell = "1.0"
diff --git a/rust-toolchain.toml b/rust-toolchain.toml
index 1ec19e58b..c05aac3fc 100644
--- a/rust-toolchain.toml
+++ b/rust-toolchain.toml
@@ -1,7 +1,7 @@
 # Before upgrading check that everything is available on all tier1 targets here:
 # https://rust-lang.github.io/rustup-components-history
 [toolchain]
-channel = "nightly-2022-09-22"
+channel = "nightly-2022-10-25"
 components = [ "rust-src", "rustfmt" ]
 targets = [
     "thumbv7em-none-eabi",
diff --git a/tests/rp/Cargo.toml b/tests/rp/Cargo.toml
index 48c88e85b..069b7fb8c 100644
--- a/tests/rp/Cargo.toml
+++ b/tests/rp/Cargo.toml
@@ -18,10 +18,10 @@ cortex-m = { version = "0.7.6", features = ["critical-section-single-core"] }
 cortex-m-rt = "0.7.0"
 embedded-hal = "0.2.6"
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "=0.1.0-alpha.2" }
+embedded-hal-async = { version = "=0.1.0-alpha.3" }
 panic-probe = { version = "0.3.0", features = ["print-defmt"] }
 futures = { version = "0.3.17", default-features = false, features = ["async-await"] }
-embedded-io = { version = "0.3.0", features = ["async"] }
+embedded-io = { version = "0.3.1", features = ["async"] }
 embedded-storage = { version = "0.3" }
 
 [profile.dev]
diff --git a/tests/stm32/Cargo.toml b/tests/stm32/Cargo.toml
index bebbf557e..602c1fb57 100644
--- a/tests/stm32/Cargo.toml
+++ b/tests/stm32/Cargo.toml
@@ -26,7 +26,7 @@ cortex-m = { version = "0.7.6", features = ["critical-section-single-core"] }
 cortex-m-rt = "0.7.0"
 embedded-hal = "0.2.6"
 embedded-hal-1 = { package = "embedded-hal", version = "=1.0.0-alpha.9" }
-embedded-hal-async = { version = "=0.1.0-alpha.2" }
+embedded-hal-async = { version = "=0.1.0-alpha.3" }
 panic-probe = { version = "0.3.0", features = ["print-defmt"] }
 
 [profile.dev]
diff --git a/tests/stm32/src/bin/usart.rs b/tests/stm32/src/bin/usart.rs
index 7673bfe6d..af55867f2 100644
--- a/tests/stm32/src/bin/usart.rs
+++ b/tests/stm32/src/bin/usart.rs
@@ -7,6 +7,7 @@ mod example_common;
 use defmt::assert_eq;
 use embassy_executor::Spawner;
 use embassy_stm32::dma::NoDma;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use example_common::*;
 
@@ -18,22 +19,22 @@ async fn main(_spawner: Spawner) {
     // Arduino pins D0 and D1
     // They're connected together with a 1K resistor.
     #[cfg(feature = "stm32f103c8")]
-    let (tx, rx, usart) = (p.PA9, p.PA10, p.USART1);
+    let (tx, rx, usart, irq) = (p.PA9, p.PA10, p.USART1, interrupt::take!(USART1));
     #[cfg(feature = "stm32g491re")]
-    let (tx, rx, usart) = (p.PC4, p.PC5, p.USART1);
+    let (tx, rx, usart, irq) = (p.PC4, p.PC5, p.USART1, interrupt::take!(USART1));
     #[cfg(feature = "stm32g071rb")]
-    let (tx, rx, usart) = (p.PC4, p.PC5, p.USART1);
+    let (tx, rx, usart, irq) = (p.PC4, p.PC5, p.USART1, interrupt::take!(USART1));
     #[cfg(feature = "stm32f429zi")]
-    let (tx, rx, usart) = (p.PG14, p.PG9, p.USART6);
+    let (tx, rx, usart, irq) = (p.PG14, p.PG9, p.USART6, interrupt::take!(USART6));
     #[cfg(feature = "stm32wb55rg")]
-    let (tx, rx, usart) = (p.PA2, p.PA3, p.LPUART1);
+    let (tx, rx, usart, irq) = (p.PA2, p.PA3, p.LPUART1, interrupt::take!(LPUART1));
     #[cfg(feature = "stm32h755zi")]
-    let (tx, rx, usart) = (p.PB6, p.PB7, p.USART1);
+    let (tx, rx, usart, irq) = (p.PB6, p.PB7, p.USART1, interrupt::take!(USART1));
     #[cfg(feature = "stm32u585ai")]
-    let (tx, rx, usart) = (p.PD8, p.PD9, p.USART3);
+    let (tx, rx, usart, irq) = (p.PD8, p.PD9, p.USART3, interrupt::take!(USART3));
 
     let config = Config::default();
-    let mut usart = Uart::new(usart, rx, tx, NoDma, NoDma, config);
+    let mut usart = Uart::new(usart, rx, tx, irq, NoDma, NoDma, config);
 
     // We can't send too many bytes, they have to fit in the FIFO.
     // This is because we aren't sending+receiving at the same time.
diff --git a/tests/stm32/src/bin/usart_dma.rs b/tests/stm32/src/bin/usart_dma.rs
index e0389446f..d12605a9a 100644
--- a/tests/stm32/src/bin/usart_dma.rs
+++ b/tests/stm32/src/bin/usart_dma.rs
@@ -6,6 +6,7 @@
 mod example_common;
 use defmt::assert_eq;
 use embassy_executor::Spawner;
+use embassy_stm32::interrupt;
 use embassy_stm32::usart::{Config, Uart};
 use example_common::*;
 
@@ -17,22 +18,53 @@ async fn main(_spawner: Spawner) {
     // Arduino pins D0 and D1
     // They're connected together with a 1K resistor.
     #[cfg(feature = "stm32f103c8")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PA9, p.PA10, p.USART1, p.DMA1_CH4, p.DMA1_CH5);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) = (
+        p.PA9,
+        p.PA10,
+        p.USART1,
+        interrupt::take!(USART1),
+        p.DMA1_CH4,
+        p.DMA1_CH5,
+    );
     #[cfg(feature = "stm32g491re")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PC4, p.PC5, p.USART1, p.DMA1_CH1, p.DMA1_CH2);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) =
+        (p.PC4, p.PC5, p.USART1, interrupt::take!(USART1), p.DMA1_CH1, p.DMA1_CH2);
     #[cfg(feature = "stm32g071rb")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PC4, p.PC5, p.USART1, p.DMA1_CH1, p.DMA1_CH2);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) =
+        (p.PC4, p.PC5, p.USART1, interrupt::take!(USART1), p.DMA1_CH1, p.DMA1_CH2);
     #[cfg(feature = "stm32f429zi")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PG14, p.PG9, p.USART6, p.DMA2_CH6, p.DMA2_CH1);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) = (
+        p.PG14,
+        p.PG9,
+        p.USART6,
+        interrupt::take!(USART6),
+        p.DMA2_CH6,
+        p.DMA2_CH1,
+    );
     #[cfg(feature = "stm32wb55rg")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PA2, p.PA3, p.LPUART1, p.DMA1_CH1, p.DMA1_CH2);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) = (
+        p.PA2,
+        p.PA3,
+        p.LPUART1,
+        interrupt::take!(LPUART1),
+        p.DMA1_CH1,
+        p.DMA1_CH2,
+    );
     #[cfg(feature = "stm32h755zi")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PB6, p.PB7, p.USART1, p.DMA1_CH0, p.DMA1_CH1);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) =
+        (p.PB6, p.PB7, p.USART1, interrupt::take!(USART1), p.DMA1_CH0, p.DMA1_CH1);
     #[cfg(feature = "stm32u585ai")]
-    let (tx, rx, usart, tx_dma, rx_dma) = (p.PD8, p.PD9, p.USART3, p.GPDMA1_CH0, p.GPDMA1_CH1);
+    let (tx, rx, usart, irq, tx_dma, rx_dma) = (
+        p.PD8,
+        p.PD9,
+        p.USART3,
+        interrupt::take!(USART3),
+        p.GPDMA1_CH0,
+        p.GPDMA1_CH1,
+    );
 
     let config = Config::default();
-    let mut usart = Uart::new(usart, rx, tx, tx_dma, rx_dma, config);
+    let mut usart = Uart::new(usart, rx, tx, irq, tx_dma, rx_dma, config);
 
     // We can't send too many bytes, they have to fit in the FIFO.
     // This is because we aren't sending+receiving at the same time.