diff --git a/embassy-stm32/src/dma/mod.rs b/embassy-stm32/src/dma/mod.rs
index 371741ff..466cfa03 100644
--- a/embassy-stm32/src/dma/mod.rs
+++ b/embassy-stm32/src/dma/mod.rs
@@ -1,9 +1,11 @@
 #![macro_use]
 
+#[cfg(dma)]
 #[cfg_attr(dma_v1, path = "v1.rs")]
 #[cfg_attr(dma_v2, path = "v2.rs")]
 mod _version;
 
+#[cfg(dma)]
 #[allow(unused)]
 pub use _version::*;
 
diff --git a/embassy-stm32/src/lib.rs b/embassy-stm32/src/lib.rs
index 9ba7bbe4..76a6ecd8 100644
--- a/embassy-stm32/src/lib.rs
+++ b/embassy-stm32/src/lib.rs
@@ -26,7 +26,7 @@ pub mod adc;
 pub mod clock;
 #[cfg(dac)]
 pub mod dac;
-#[cfg(dma)]
+#[cfg(any(dma, dmamux))]
 pub mod dma;
 #[cfg(all(eth, feature = "net"))]
 pub mod eth;
diff --git a/embassy-stm32/src/usart/mod.rs b/embassy-stm32/src/usart/mod.rs
index a75793f8..2fa758ec 100644
--- a/embassy-stm32/src/usart/mod.rs
+++ b/embassy-stm32/src/usart/mod.rs
@@ -2,6 +2,7 @@
 
 #[cfg_attr(usart_v1, path = "v1.rs")]
 #[cfg_attr(usart_v2, path = "v2.rs")]
+#[cfg_attr(usart_v3, path = "v3.rs")]
 mod _version;
 use crate::peripherals;
 pub use _version::*;
@@ -10,6 +11,51 @@ use crate::gpio::Pin;
 use crate::pac::usart::Usart;
 use crate::rcc::RccPeripheral;
 
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub enum DataBits {
+    DataBits8,
+    DataBits9,
+}
+
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub enum Parity {
+    ParityNone,
+    ParityEven,
+    ParityOdd,
+}
+
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub enum StopBits {
+    #[doc = "1 stop bit"]
+    STOP1,
+    #[doc = "0.5 stop bits"]
+    STOP0P5,
+    #[doc = "2 stop bits"]
+    STOP2,
+    #[doc = "1.5 stop bits"]
+    STOP1P5,
+}
+
+#[non_exhaustive]
+#[derive(Clone, Copy, PartialEq, Eq, Debug)]
+pub struct Config {
+    pub baudrate: u32,
+    pub data_bits: DataBits,
+    pub stop_bits: StopBits,
+    pub parity: Parity,
+}
+
+impl Default for Config {
+    fn default() -> Self {
+        Self {
+            baudrate: 115200,
+            data_bits: DataBits::DataBits8,
+            stop_bits: StopBits::STOP1,
+            parity: Parity::ParityNone,
+        }
+    }
+}
+
 /// Serial error
 #[derive(Debug, Eq, PartialEq, Copy, Clone)]
 #[non_exhaustive]
@@ -27,7 +73,7 @@ pub enum Error {
 pub(crate) mod sealed {
     use super::*;
 
-    #[cfg(dma)]
+    #[cfg(any(dma, dmamux))]
     use crate::dma::WriteDma;
 
     pub trait Instance {
@@ -49,10 +95,10 @@ pub(crate) mod sealed {
         fn af_num(&self) -> u8;
     }
 
-    #[cfg(dma)]
+    #[cfg(any(dma, dmamux))]
     pub trait RxDma<T: Instance> {}
 
-    #[cfg(dma)]
+    #[cfg(any(dma, dmamux))]
     pub trait TxDma<T: Instance>: WriteDma<T> {}
 }
 
@@ -63,9 +109,10 @@ pub trait CtsPin<T: Instance>: sealed::CtsPin<T> {}
 pub trait RtsPin<T: Instance>: sealed::RtsPin<T> {}
 pub trait CkPin<T: Instance>: sealed::CkPin<T> {}
 
-#[cfg(dma)]
+#[cfg(any(dma, dmamux))]
 pub trait RxDma<T: Instance>: sealed::RxDma<T> {}
-#[cfg(dma)]
+
+#[cfg(any(dma, dmamux))]
 pub trait TxDma<T: Instance>: sealed::TxDma<T> {}
 
 crate::pac::peripherals!(
@@ -93,6 +140,9 @@ macro_rules! impl_pin {
 }
 
 crate::pac::peripheral_pins!(
+
+    // USART
+
     ($inst:ident, usart, USART, $pin:ident, TX, $af:expr) => {
         impl_pin!($inst, $pin, TxPin, $af);
     };
@@ -112,4 +162,26 @@ crate::pac::peripheral_pins!(
     ($inst:ident, usart, USART, $pin:ident, CK, $af:expr) => {
         impl_pin!($inst, $pin, CkPin, $af);
     };
+
+    // UART
+
+    ($inst:ident, uart, UART, $pin:ident, TX, $af:expr) => {
+        impl_pin!($inst, $pin, TxPin, $af);
+    };
+
+    ($inst:ident, uart, UART, $pin:ident, RX, $af:expr) => {
+        impl_pin!($inst, $pin, RxPin, $af);
+    };
+
+    ($inst:ident, uart, UART, $pin:ident, CTS, $af:expr) => {
+        impl_pin!($inst, $pin, CtsPin, $af);
+    };
+
+    ($inst:ident, uart, UART, $pin:ident, RTS, $af:expr) => {
+        impl_pin!($inst, $pin, RtsPin, $af);
+    };
+
+    ($inst:ident, uart, UART, $pin:ident, CK, $af:expr) => {
+        impl_pin!($inst, $pin, CkPin, $af);
+    };
 );
diff --git a/embassy-stm32/src/usart/v1.rs b/embassy-stm32/src/usart/v1.rs
index 6c5a1244..16993895 100644
--- a/embassy-stm32/src/usart/v1.rs
+++ b/embassy-stm32/src/usart/v1.rs
@@ -7,51 +7,6 @@ use crate::pac::usart::{regs, vals};
 
 use super::*;
 
-#[derive(Clone, Copy, PartialEq, Eq, Debug)]
-pub enum DataBits {
-    DataBits8,
-    DataBits9,
-}
-
-#[derive(Clone, Copy, PartialEq, Eq, Debug)]
-pub enum Parity {
-    ParityNone,
-    ParityEven,
-    ParityOdd,
-}
-
-#[derive(Clone, Copy, PartialEq, Eq, Debug)]
-pub enum StopBits {
-    #[doc = "1 stop bit"]
-    STOP1,
-    #[doc = "0.5 stop bits"]
-    STOP0P5,
-    #[doc = "2 stop bits"]
-    STOP2,
-    #[doc = "1.5 stop bits"]
-    STOP1P5,
-}
-
-#[non_exhaustive]
-#[derive(Clone, Copy, PartialEq, Eq, Debug)]
-pub struct Config {
-    pub baudrate: u32,
-    pub data_bits: DataBits,
-    pub stop_bits: StopBits,
-    pub parity: Parity,
-}
-
-impl Default for Config {
-    fn default() -> Self {
-        Self {
-            baudrate: 115200,
-            data_bits: DataBits::DataBits8,
-            stop_bits: StopBits::STOP1,
-            parity: Parity::ParityNone,
-        }
-    }
-}
-
 pub struct Uart<'d, T: Instance> {
     inner: T,
     phantom: PhantomData<&'d mut T>,
diff --git a/embassy-stm32/src/usart/v3.rs b/embassy-stm32/src/usart/v3.rs
new file mode 100644
index 00000000..1e905144
--- /dev/null
+++ b/embassy-stm32/src/usart/v3.rs
@@ -0,0 +1,121 @@
+use core::marker::PhantomData;
+
+use embassy::util::Unborrow;
+use embassy_extras::unborrow;
+
+use crate::pac::usart::{regs, vals};
+
+use super::*;
+
+pub struct Uart<'d, T: Instance> {
+    inner: T,
+    phantom: PhantomData<&'d mut T>,
+}
+
+impl<'d, T: Instance> Uart<'d, T> {
+    pub fn new(
+        inner: impl Unborrow<Target = T>,
+        rx: impl Unborrow<Target = impl RxPin<T>>,
+        tx: impl Unborrow<Target = impl TxPin<T>>,
+        config: Config,
+    ) -> Self {
+        unborrow!(inner, rx, tx);
+
+        // Uncomment once we find all of the H7's UART clocks.
+        T::enable();
+        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;
+
+        let r = inner.regs();
+
+        unsafe {
+            rx.set_as_af(rx.af_num());
+            tx.set_as_af(tx.af_num());
+
+            r.brr().write_value(regs::Brr(div));
+            r.cr1().write(|w| {
+                w.set_ue(true);
+                w.set_te(true);
+                w.set_re(true);
+                w.set_m(0, 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,
+                });
+            });
+            r.cr2().write(|_w| {});
+            r.cr3().write(|_w| {});
+        }
+
+        Self {
+            inner,
+            phantom: PhantomData,
+        }
+    }
+
+    #[cfg(dma)]
+    pub async fn write_dma(&mut self, ch: &mut impl TxDma<T>, buffer: &[u8]) -> Result<(), Error> {
+        unsafe {
+            self.inner.regs().cr3().modify(|reg| {
+                reg.set_dmat(true);
+            });
+        }
+        let r = self.inner.regs();
+        let dst = r.tdr().ptr() as *mut u8;
+        ch.transfer(buffer, dst).await;
+        Ok(())
+    }
+
+    pub fn read(&mut self, buffer: &mut [u8]) -> Result<(), Error> {
+        unsafe {
+            let r = self.inner.regs();
+            for b in buffer {
+                loop {
+                    let sr = r.isr().read();
+                    if sr.pe() {
+                        r.rdr().read();
+                        return Err(Error::Parity);
+                    } else if sr.fe() {
+                        r.rdr().read();
+                        return Err(Error::Framing);
+                    } else if sr.ne() {
+                        r.rdr().read();
+                        return Err(Error::Noise);
+                    } else if sr.ore() {
+                        r.rdr().read();
+                        return Err(Error::Overrun);
+                    } else if sr.rxne() {
+                        break;
+                    }
+                }
+                *b = r.rdr().read().0 as u8;
+            }
+        }
+        Ok(())
+    }
+}
+
+impl<'d, T: Instance> embedded_hal::blocking::serial::Write<u8> for Uart<'d, T> {
+    type Error = Error;
+    fn bwrite_all(&mut self, buffer: &[u8]) -> Result<(), Self::Error> {
+        unsafe {
+            let r = self.inner.regs();
+            for &b in buffer {
+                while !r.isr().read().txe() {}
+                r.tdr().write_value(regs::Tdr(b as u32))
+            }
+        }
+        Ok(())
+    }
+    fn bflush(&mut self) -> Result<(), Self::Error> {
+        unsafe {
+            let r = self.inner.regs();
+            while !r.isr().read().tc() {}
+        }
+        Ok(())
+    }
+}
diff --git a/examples/stm32h7/src/bin/usart.rs b/examples/stm32h7/src/bin/usart.rs
new file mode 100644
index 00000000..b8524f2c
--- /dev/null
+++ b/examples/stm32h7/src/bin/usart.rs
@@ -0,0 +1,99 @@
+
+#![no_std]
+#![no_main]
+#![feature(trait_alias)]
+#![feature(min_type_alias_impl_trait)]
+#![feature(impl_trait_in_bindings)]
+#![feature(type_alias_impl_trait)]
+#![allow(incomplete_features)]
+
+#[path = "../example_common.rs"]
+mod example_common;
+use cortex_m::prelude::_embedded_hal_blocking_serial_Write;
+use embassy::executor::Executor;
+use embassy::time::Clock;
+use embassy::util::Forever;
+use embassy_stm32::usart::{Config, Uart};
+use example_common::*;
+
+use stm32h7xx_hal as hal;
+use hal::prelude::*;
+
+use cortex_m_rt::entry;
+use stm32h7::stm32h743 as pac;
+
+#[embassy::task]
+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, config);
+
+    usart.bwrite_all(b"Hello Embassy World!\r\n").unwrap();
+    info!("wrote Hello, starting echo");
+
+    let mut buf = [0u8; 1];
+    loop {
+        usart.read(&mut buf).unwrap();
+        usart.bwrite_all(&buf).unwrap();
+    }
+}
+
+struct ZeroClock;
+
+impl Clock for ZeroClock {
+    fn now(&self) -> u64 {
+        0
+    }
+}
+
+static EXECUTOR: Forever<Executor> = Forever::new();
+
+#[entry]
+fn main() -> ! {
+    info!("Hello World!");
+
+    let pp = pac::Peripherals::take().unwrap();
+
+    let pwrcfg = pp.PWR.constrain().freeze();
+
+    let rcc = pp.RCC.constrain();
+
+    let ccdr = rcc
+        .sys_ck(96.mhz())
+        .pclk1(48.mhz())
+        .pclk2(48.mhz())
+        .pclk3(48.mhz())
+        .pclk4(48.mhz())
+        .pll1_q_ck(48.mhz())
+        .freeze(pwrcfg, &pp.SYSCFG);
+
+    let pp = unsafe { pac::Peripherals::steal() };
+
+    pp.DBGMCU.cr.modify(|_, w| {
+        w.dbgsleep_d1().set_bit();
+        w.dbgstby_d1().set_bit();
+        w.dbgstop_d1().set_bit();
+        w.d1dbgcken().set_bit();
+        w
+    });
+
+    pp.RCC.ahb4enr.modify(|_, w| {
+        w.gpioaen().set_bit();
+        w.gpioben().set_bit();
+        w.gpiocen().set_bit();
+        w.gpioden().set_bit();
+        w.gpioeen().set_bit();
+        w.gpiofen().set_bit();
+        w
+    });
+
+
+    unsafe { embassy::time::set_clock(&ZeroClock) };
+
+    let executor = EXECUTOR.put(Executor::new());
+
+    executor.run(|spawner| {
+        unwrap!(spawner.spawn(main_task()));
+    })
+}
\ No newline at end of file
diff --git a/examples/stm32l4/src/bin/usart.rs b/examples/stm32l4/src/bin/usart.rs
new file mode 100644
index 00000000..5b6d9eaa
--- /dev/null
+++ b/examples/stm32l4/src/bin/usart.rs
@@ -0,0 +1,97 @@
+
+#![no_std]
+#![no_main]
+#![feature(trait_alias)]
+#![feature(min_type_alias_impl_trait)]
+#![feature(impl_trait_in_bindings)]
+#![feature(type_alias_impl_trait)]
+#![allow(incomplete_features)]
+
+#[path = "../example_common.rs"]
+mod example_common;
+use cortex_m::prelude::_embedded_hal_blocking_serial_Write;
+use embassy::executor::Executor;
+use embassy::time::Clock;
+use embassy::util::Forever;
+use embassy_stm32::usart::{Config, Uart};
+use example_common::*;
+
+use cortex_m_rt::entry;
+use stm32l4::stm32l4x5 as pac;
+
+#[embassy::task]
+async fn main_task() {
+    let p = embassy_stm32::init(Default::default());
+
+    let config = Config::default();
+    let mut usart = Uart::new(p.UART4, p.PA1, p.PA0, config);
+
+    usart.bwrite_all(b"Hello Embassy World!\r\n").unwrap();
+    info!("wrote Hello, starting echo");
+
+    let mut buf = [0u8; 1];
+    loop {
+        usart.read(&mut buf).unwrap();
+        usart.bwrite_all(&buf).unwrap();
+    }
+}
+
+struct ZeroClock;
+
+impl Clock for ZeroClock {
+    fn now(&self) -> u64 {
+        0
+    }
+}
+
+static EXECUTOR: Forever<Executor> = Forever::new();
+
+#[entry]
+fn main() -> ! {
+    info!("Hello World!");
+
+    let pp = pac::Peripherals::take().unwrap();
+
+    pp.DBGMCU.cr.modify(|_, w| {
+        w.dbg_sleep().set_bit();
+        w.dbg_standby().set_bit();
+        w.dbg_stop().set_bit()
+    });
+
+    pp.RCC.ahb1enr.modify(|_, w| {
+        w.dma1en().set_bit();
+        w
+    });
+
+    pp.RCC.ahb2enr.modify(|_, w| {
+        w.gpioaen().set_bit();
+        w.gpioben().set_bit();
+        w.gpiocen().set_bit();
+        w.gpioden().set_bit();
+        w.gpioeen().set_bit();
+        w.gpiofen().set_bit();
+        w
+    });
+
+    pp.RCC.apb1enr1.modify(|_, w| {
+        w.uart4en().set_bit();
+        w
+    });
+
+    pp.RCC.apb2enr.modify(|_, w| {
+        w.syscfgen().set_bit();
+        w
+    });
+    //pp.RCC.apb1enr.modify(|_, w| {
+    //w.usart3en().enabled();
+    //w
+    //});
+
+    unsafe { embassy::time::set_clock(&ZeroClock) };
+
+    let executor = EXECUTOR.put(Executor::new());
+
+    executor.run(|spawner| {
+        unwrap!(spawner.spawn(main_task()));
+    })
+}
\ No newline at end of file
diff --git a/stm32-data b/stm32-data
index f0a6585b..0877c27c 160000
--- a/stm32-data
+++ b/stm32-data
@@ -1 +1 @@
-Subproject commit f0a6585b4806b1f7c6836126d063eaaf970cc5a4
+Subproject commit 0877c27cb1332237e65d74700b7bfb768996ca66
diff --git a/stm32-metapac/gen/src/lib.rs b/stm32-metapac/gen/src/lib.rs
index a552c8ce..8eae4911 100644
--- a/stm32-metapac/gen/src/lib.rs
+++ b/stm32-metapac/gen/src/lib.rs
@@ -311,6 +311,7 @@ pub fn gen(options: Options) {
 
                 for dma_request in &p.dma_requests {
                     let mut row = Vec::new();
+                    row.push(bi.module.clone());
                     row.push(name.clone());
                     row.push(dma_request.0.clone());
                     row.push(dma_request.1.to_string());