12 files changed, 203 insertions, 174 deletions

diff --git a/embassy-stm32/src/spi/mod.rs b/embassy-stm32/src/spi/mod.rs
index f83ef7852..f1ea8592d 100644
--- a/embassy-stm32/src/spi/mod.rs
+++ b/embassy-stm32/src/spi/mod.rs
@@ -1,5 +1,13 @@
 #![macro_use]
 
+use core::future::Future;
+use core::marker::PhantomData;
+use core::ptr;
+use embassy::util::Unborrow;
+use embassy_hal_common::unborrow;
+use embassy_traits::spi as traits;
+
+use self::sealed::WordSize;
 use crate::dma;
 use crate::dma::NoDma;
 use crate::gpio::sealed::{AFType, Pin};
@@ -8,19 +16,14 @@ use crate::pac::spi::{regs, vals};
 use crate::peripherals;
 use crate::rcc::RccPeripheral;
 use crate::time::Hertz;
-use core::future::Future;
-use core::marker::PhantomData;
-use core::ptr;
-use embassy::util::Unborrow;
-use embassy_hal_common::unborrow;
-use embassy_traits::spi as traits;
+
+pub use embedded_hal::spi::{Mode, Phase, Polarity, MODE_0, MODE_1, MODE_2, MODE_3};
 
 #[cfg_attr(spi_v1, path = "v1.rs")]
 #[cfg_attr(spi_f1, path = "v1.rs")]
 #[cfg_attr(spi_v2, path = "v2.rs")]
 #[cfg_attr(spi_v3, path = "v3.rs")]
 mod _version;
-pub use _version::*;
 
 type Regs = &'static crate::pac::spi::Spi;
 
@@ -40,54 +43,6 @@ pub enum BitOrder {
     MsbFirst,
 }
 
-#[derive(Copy, Clone, PartialOrd, PartialEq)]
-enum WordSize {
-    EightBit,
-    SixteenBit,
-}
-
-impl WordSize {
-    #[cfg(any(spi_v1, spi_f1))]
-    fn dff(&self) -> vals::Dff {
-        match self {
-            WordSize::EightBit => vals::Dff::EIGHTBIT,
-            WordSize::SixteenBit => vals::Dff::SIXTEENBIT,
-        }
-    }
-
-    #[cfg(spi_v2)]
-    fn ds(&self) -> vals::Ds {
-        match self {
-            WordSize::EightBit => vals::Ds::EIGHTBIT,
-            WordSize::SixteenBit => vals::Ds::SIXTEENBIT,
-        }
-    }
-
-    #[cfg(spi_v2)]
-    fn frxth(&self) -> vals::Frxth {
-        match self {
-            WordSize::EightBit => vals::Frxth::QUARTER,
-            WordSize::SixteenBit => vals::Frxth::HALF,
-        }
-    }
-
-    #[cfg(spi_v3)]
-    fn dsize(&self) -> u8 {
-        match self {
-            WordSize::EightBit => 0b0111,
-            WordSize::SixteenBit => 0b1111,
-        }
-    }
-
-    #[cfg(spi_v3)]
-    fn _frxth(&self) -> vals::Fthlv {
-        match self {
-            WordSize::EightBit => vals::Fthlv::ONEFRAME,
-            WordSize::SixteenBit => vals::Fthlv::ONEFRAME,
-        }
-    }
-}
-
 #[non_exhaustive]
 #[derive(Copy, Clone)]
 pub struct Config {
@@ -379,6 +334,47 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
 
         self.current_word_size = word_size;
     }
+
+    pub async fn write(&mut self, data: &[u8]) -> Result<(), Error>
+    where
+        Tx: TxDmaChannel<T>,
+    {
+        self.write_dma_u8(data).await
+    }
+
+    pub async fn read(&mut self, data: &mut [u8]) -> Result<(), Error>
+    where
+        Tx: TxDmaChannel<T>,
+        Rx: RxDmaChannel<T>,
+    {
+        self.read_dma_u8(data).await
+    }
+
+    pub async fn transfer(&mut self, read: &mut [u8], write: &[u8]) -> Result<(), Error>
+    where
+        Tx: TxDmaChannel<T>,
+        Rx: RxDmaChannel<T>,
+    {
+        self.transfer_dma_u8(read, write).await
+    }
+
+    pub fn blocking_write<W: Word>(&mut self, words: &[W]) -> Result<(), Error> {
+        self.set_word_size(W::WORDSIZE);
+        let regs = T::regs();
+        for word in words.iter() {
+            let _ = transfer_word(regs, *word)?;
+        }
+        Ok(())
+    }
+
+    pub fn blocking_transfer_in_place<W: Word>(&mut self, words: &mut [W]) -> Result<(), Error> {
+        self.set_word_size(W::WORDSIZE);
+        let regs = T::regs();
+        for word in words.iter_mut() {
+            *word = transfer_word(regs, *word)?;
+        }
+        Ok(())
+    }
 }
 
 impl<'d, T: Instance, Tx, Rx> Drop for Spi<'d, T, Tx, Rx> {
@@ -537,17 +533,6 @@ fn finish_dma(regs: Regs) {
     }
 }
 
-trait Word {
-    const WORDSIZE: WordSize;
-}
-
-impl Word for u8 {
-    const WORDSIZE: WordSize = WordSize::EightBit;
-}
-impl Word for u16 {
-    const WORDSIZE: WordSize = WordSize::SixteenBit;
-}
-
 fn transfer_word<W: Word>(regs: Regs, tx_word: W) -> Result<W, Error> {
     spin_until_tx_ready(regs)?;
 
@@ -572,14 +557,7 @@ macro_rules! impl_blocking {
             type Error = Error;
 
             fn write(&mut self, words: &[$w]) -> Result<(), Self::Error> {
-                self.set_word_size($w::WORDSIZE);
-                let regs = T::regs();
-
-                for word in words.iter() {
-                    let _ = transfer_word(regs, *word)?;
-                }
-
-                Ok(())
+                self.blocking_write(words)
             }
         }
 
@@ -589,13 +567,7 @@ macro_rules! impl_blocking {
             type Error = Error;
 
             fn transfer<'w>(&mut self, words: &'w mut [$w]) -> Result<&'w [$w], Self::Error> {
-                self.set_word_size($w::WORDSIZE);
-                let regs = T::regs();
-
-                for word in words.iter_mut() {
-                    *word = transfer_word(regs, *word)?;
-                }
-
+                self.blocking_transfer_in_place(words)?;
                 Ok(words)
             }
         }
@@ -616,7 +588,7 @@ impl<'d, T: Instance, Tx: TxDmaChannel<T>, Rx> traits::Write<u8> for Spi<'d, T,
     = impl Future<Output = Result<(), Self::Error>> + 'a;
 
     fn write<'a>(&'a mut self, data: &'a [u8]) -> Self::WriteFuture<'a> {
-        self.write_dma_u8(data)
+        self.write(data)
     }
 }
 
@@ -629,7 +601,7 @@ impl<'d, T: Instance, Tx: TxDmaChannel<T>, Rx: RxDmaChannel<T>> traits::Read<u8>
     = impl Future<Output = Result<(), Self::Error>> + 'a;
 
     fn read<'a>(&'a mut self, data: &'a mut [u8]) -> Self::ReadFuture<'a> {
-        self.read_dma_u8(data)
+        self.read(data)
     }
 }
 
@@ -646,7 +618,7 @@ impl<'d, T: Instance, Tx: TxDmaChannel<T>, Rx: RxDmaChannel<T>> traits::FullDupl
         read: &'a mut [u8],
         write: &'a [u8],
     ) -> Self::WriteReadFuture<'a> {
-        self.read_write_dma_u8(read, write)
+        self.transfer(read, write)
     }
 }
 
@@ -676,8 +648,72 @@ pub(crate) mod sealed {
     pub trait RxDmaChannel<T: Instance> {
         fn request(&self) -> dma::Request;
     }
+
+    pub trait Word: Copy + 'static {
+        const WORDSIZE: WordSize;
+    }
+
+    impl Word for u8 {
+        const WORDSIZE: WordSize = WordSize::EightBit;
+    }
+    impl Word for u16 {
+        const WORDSIZE: WordSize = WordSize::SixteenBit;
+    }
+
+    #[derive(Copy, Clone, PartialOrd, PartialEq)]
+    pub enum WordSize {
+        EightBit,
+        SixteenBit,
+    }
+
+    impl WordSize {
+        #[cfg(any(spi_v1, spi_f1))]
+        pub fn dff(&self) -> vals::Dff {
+            match self {
+                WordSize::EightBit => vals::Dff::EIGHTBIT,
+                WordSize::SixteenBit => vals::Dff::SIXTEENBIT,
+            }
+        }
+
+        #[cfg(spi_v2)]
+        pub fn ds(&self) -> vals::Ds {
+            match self {
+                WordSize::EightBit => vals::Ds::EIGHTBIT,
+                WordSize::SixteenBit => vals::Ds::SIXTEENBIT,
+            }
+        }
+
+        #[cfg(spi_v2)]
+        pub fn frxth(&self) -> vals::Frxth {
+            match self {
+                WordSize::EightBit => vals::Frxth::QUARTER,
+                WordSize::SixteenBit => vals::Frxth::HALF,
+            }
+        }
+
+        #[cfg(spi_v3)]
+        pub fn dsize(&self) -> u8 {
+            match self {
+                WordSize::EightBit => 0b0111,
+                WordSize::SixteenBit => 0b1111,
+            }
+        }
+
+        #[cfg(spi_v3)]
+        pub fn _frxth(&self) -> vals::Fthlv {
+            match self {
+                WordSize::EightBit => vals::Fthlv::ONEFRAME,
+                WordSize::SixteenBit => vals::Fthlv::ONEFRAME,
+            }
+        }
+    }
 }
 
+pub trait Word: Copy + 'static + sealed::Word {}
+
+impl Word for u8 {}
+impl Word for u16 {}
+
 pub trait Instance: sealed::Instance + RccPeripheral {}
 pub trait SckPin<T: Instance>: sealed::SckPin<T> {}
 pub trait MosiPin<T: Instance>: sealed::MosiPin<T> {}
diff --git a/embassy-stm32/src/spi/v1.rs b/embassy-stm32/src/spi/v1.rs
index ef7e0f59c..68e5a7348 100644
--- a/embassy-stm32/src/spi/v1.rs
+++ b/embassy-stm32/src/spi/v1.rs
@@ -1,13 +1,12 @@
 #![macro_use]
 
-pub use embedded_hal::blocking;
-pub use embedded_hal::spi::{Mode, Phase, Polarity, MODE_0, MODE_1, MODE_2, MODE_3};
 use futures::future::join;
 
 use super::*;
+use crate::dma::{slice_ptr_parts, slice_ptr_parts_mut, Transfer};
 
 impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
-    pub(super) async fn write_dma_u8(&mut self, write: &[u8]) -> Result<(), Error>
+    pub(super) async fn write_dma_u8(&mut self, write: *const [u8]) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
     {
@@ -18,9 +17,10 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         }
         self.set_word_size(WordSize::EightBit);
 
-        let request = self.txdma.request();
-        let dst = T::regs().tx_ptr();
-        let f = crate::dma::write(&mut self.txdma, request, write, dst);
+        let tx_request = self.txdma.request();
+        let tx_dst = T::regs().tx_ptr();
+        unsafe { self.txdma.start_write(tx_request, write, tx_dst) }
+        let tx_f = Transfer::new(&mut self.txdma);
 
         unsafe {
             T::regs().cr2().modify(|reg| {
@@ -31,14 +31,14 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
             });
         }
 
-        f.await;
+        tx_f.await;
 
         finish_dma(T::regs());
 
         Ok(())
     }
 
-    pub(super) async fn read_dma_u8(&mut self, read: &mut [u8]) -> Result<(), Error>
+    pub(super) async fn read_dma_u8(&mut self, read: *mut [u8]) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
         Rx: RxDmaChannel<T>,
@@ -53,11 +53,12 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         }
         self.set_word_size(WordSize::EightBit);
 
-        let clock_byte_count = read.len();
+        let (_, clock_byte_count) = slice_ptr_parts_mut(read);
 
         let rx_request = self.rxdma.request();
         let rx_src = T::regs().rx_ptr();
-        let rx_f = crate::dma::read(&mut self.rxdma, rx_request, rx_src, read);
+        unsafe { self.rxdma.start_read(rx_request, rx_src, read) };
+        let rx_f = Transfer::new(&mut self.rxdma);
 
         let tx_request = self.txdma.request();
         let tx_dst = T::regs().tx_ptr();
@@ -86,16 +87,18 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         Ok(())
     }
 
-    pub(super) async fn read_write_dma_u8(
+    pub(super) async fn transfer_dma_u8(
         &mut self,
-        read: &mut [u8],
-        write: &[u8],
+        read: *mut [u8],
+        write: *const [u8],
     ) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
         Rx: RxDmaChannel<T>,
     {
-        assert!(read.len() >= write.len());
+        let (_, rx_len) = slice_ptr_parts(read);
+        let (_, tx_len) = slice_ptr_parts(write);
+        assert_eq!(rx_len, tx_len);
 
         unsafe {
             T::regs().cr1().modify(|w| {
@@ -109,16 +112,13 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
 
         let rx_request = self.rxdma.request();
         let rx_src = T::regs().rx_ptr();
-        let rx_f = crate::dma::read(
-            &mut self.rxdma,
-            rx_request,
-            rx_src,
-            &mut read[0..write.len()],
-        );
+        unsafe { self.rxdma.start_read(rx_request, rx_src, read) };
+        let rx_f = Transfer::new(&mut self.rxdma);
 
         let tx_request = self.txdma.request();
         let tx_dst = T::regs().tx_ptr();
-        let tx_f = crate::dma::write(&mut self.txdma, tx_request, write, tx_dst);
+        unsafe { self.txdma.start_write(tx_request, write, tx_dst) }
+        let tx_f = Transfer::new(&mut self.txdma);
 
         unsafe {
             T::regs().cr2().modify(|reg| {
diff --git a/embassy-stm32/src/spi/v2.rs b/embassy-stm32/src/spi/v2.rs
index f2ba3ac67..78bb1192b 100644
--- a/embassy-stm32/src/spi/v2.rs
+++ b/embassy-stm32/src/spi/v2.rs
@@ -1,12 +1,12 @@
 #![macro_use]
 
-pub use embedded_hal::spi::{Mode, Phase, Polarity, MODE_0, MODE_1, MODE_2, MODE_3};
 use futures::future::join;
 
 use super::*;
+use crate::dma::{slice_ptr_parts, slice_ptr_parts_mut, Transfer};
 
 impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
-    pub(super) async fn write_dma_u8(&mut self, write: &[u8]) -> Result<(), Error>
+    pub(super) async fn write_dma_u8(&mut self, write: *const [u8]) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
     {
@@ -22,9 +22,10 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         }
         self.set_word_size(WordSize::EightBit);
 
-        let request = self.txdma.request();
-        let dst = T::regs().tx_ptr();
-        let f = crate::dma::write(&mut self.txdma, request, write, dst);
+        let tx_request = self.txdma.request();
+        let tx_dst = T::regs().tx_ptr();
+        unsafe { self.txdma.start_write(tx_request, write, tx_dst) }
+        let tx_f = Transfer::new(&mut self.txdma);
 
         unsafe {
             T::regs().cr2().modify(|reg| {
@@ -35,14 +36,14 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
             });
         }
 
-        f.await;
+        tx_f.await;
 
         finish_dma(T::regs());
 
         Ok(())
     }
 
-    pub(super) async fn read_dma_u8(&mut self, read: &mut [u8]) -> Result<(), Error>
+    pub(super) async fn read_dma_u8(&mut self, read: *mut [u8]) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
         Rx: RxDmaChannel<T>,
@@ -57,11 +58,12 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         }
         self.set_word_size(WordSize::EightBit);
 
-        let clock_byte_count = read.len();
+        let (_, clock_byte_count) = slice_ptr_parts_mut(read);
 
         let rx_request = self.rxdma.request();
         let rx_src = T::regs().rx_ptr();
-        let rx_f = crate::dma::read(&mut self.rxdma, rx_request, rx_src, read);
+        unsafe { self.rxdma.start_read(rx_request, rx_src, read) };
+        let rx_f = Transfer::new(&mut self.rxdma);
 
         let tx_request = self.txdma.request();
         let tx_dst = T::regs().tx_ptr();
@@ -90,16 +92,18 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         Ok(())
     }
 
-    pub(super) async fn read_write_dma_u8(
+    pub(super) async fn transfer_dma_u8(
         &mut self,
-        read: &mut [u8],
-        write: &[u8],
+        read: *mut [u8],
+        write: *const [u8],
     ) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
         Rx: RxDmaChannel<T>,
     {
-        assert!(read.len() >= write.len());
+        let (_, rx_len) = slice_ptr_parts(read);
+        let (_, tx_len) = slice_ptr_parts(write);
+        assert_eq!(rx_len, tx_len);
 
         unsafe {
             T::regs().cr1().modify(|w| {
@@ -118,16 +122,13 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
 
         let rx_request = self.rxdma.request();
         let rx_src = T::regs().rx_ptr();
-        let rx_f = crate::dma::read(
-            &mut self.rxdma,
-            rx_request,
-            rx_src,
-            &mut read[0..write.len()],
-        );
+        unsafe { self.rxdma.start_read(rx_request, rx_src, read) };
+        let rx_f = Transfer::new(&mut self.rxdma);
 
         let tx_request = self.txdma.request();
         let tx_dst = T::regs().tx_ptr();
-        let tx_f = crate::dma::write(&mut self.txdma, tx_request, write, tx_dst);
+        unsafe { self.txdma.start_write(tx_request, write, tx_dst) }
+        let tx_f = Transfer::new(&mut self.txdma);
 
         unsafe {
             T::regs().cr2().modify(|reg| {
diff --git a/embassy-stm32/src/spi/v3.rs b/embassy-stm32/src/spi/v3.rs
index b42eeed89..50650da1b 100644
--- a/embassy-stm32/src/spi/v3.rs
+++ b/embassy-stm32/src/spi/v3.rs
@@ -1,12 +1,12 @@
 #![macro_use]
 
-pub use embedded_hal::spi::{Mode, Phase, Polarity, MODE_0, MODE_1, MODE_2, MODE_3};
 use futures::future::join;
 
 use super::*;
+use crate::dma::{slice_ptr_parts, slice_ptr_parts_mut, Transfer};
 
 impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
-    pub(super) async fn write_dma_u8(&mut self, write: &[u8]) -> Result<(), Error>
+    pub(super) async fn write_dma_u8(&mut self, write: *const [u8]) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
     {
@@ -22,9 +22,10 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
             }
         }
 
-        let request = self.txdma.request();
-        let dst = T::regs().tx_ptr();
-        let f = crate::dma::write(&mut self.txdma, request, write, dst);
+        let tx_request = self.txdma.request();
+        let tx_dst = T::regs().tx_ptr();
+        unsafe { self.txdma.start_write(tx_request, write, tx_dst) }
+        let tx_f = Transfer::new(&mut self.txdma);
 
         unsafe {
             T::regs().cfg1().modify(|reg| {
@@ -38,14 +39,14 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
             });
         }
 
-        f.await;
+        tx_f.await;
 
         finish_dma(T::regs());
 
         Ok(())
     }
 
-    pub(super) async fn read_dma_u8(&mut self, read: &mut [u8]) -> Result<(), Error>
+    pub(super) async fn read_dma_u8(&mut self, read: *mut [u8]) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
         Rx: RxDmaChannel<T>,
@@ -60,11 +61,12 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
             });
         }
 
-        let clock_byte_count = read.len();
+        let (_, clock_byte_count) = slice_ptr_parts_mut(read);
 
         let rx_request = self.rxdma.request();
         let rx_src = T::regs().rx_ptr();
-        let rx_f = crate::dma::read(&mut self.rxdma, rx_request, rx_src, read);
+        unsafe { self.rxdma.start_read(rx_request, rx_src, read) };
+        let rx_f = Transfer::new(&mut self.rxdma);
 
         let tx_request = self.txdma.request();
         let tx_dst = T::regs().tx_ptr();
@@ -96,16 +98,18 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
         Ok(())
     }
 
-    pub(super) async fn read_write_dma_u8(
+    pub(super) async fn transfer_dma_u8(
         &mut self,
-        read: &mut [u8],
-        write: &[u8],
+        read: *mut [u8],
+        write: *const [u8],
     ) -> Result<(), Error>
     where
         Tx: TxDmaChannel<T>,
         Rx: RxDmaChannel<T>,
     {
-        assert!(read.len() >= write.len());
+        let (_, rx_len) = slice_ptr_parts(read);
+        let (_, tx_len) = slice_ptr_parts(write);
+        assert_eq!(rx_len, tx_len);
 
         self.set_word_size(WordSize::EightBit);
         unsafe {
@@ -124,16 +128,13 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
 
         let rx_request = self.rxdma.request();
         let rx_src = T::regs().rx_ptr();
-        let rx_f = crate::dma::read(
-            &mut self.rxdma,
-            rx_request,
-            rx_src,
-            &mut read[0..write.len()],
-        );
+        unsafe { self.rxdma.start_read(rx_request, rx_src, read) };
+        let rx_f = Transfer::new(&mut self.rxdma);
 
         let tx_request = self.txdma.request();
         let tx_dst = T::regs().tx_ptr();
-        let tx_f = crate::dma::write(&mut self.txdma, tx_request, write, tx_dst);
+        unsafe { self.txdma.start_write(tx_request, write, tx_dst) }
+        let tx_f = Transfer::new(&mut self.txdma);
 
         unsafe {
             T::regs().cfg1().modify(|reg| {
diff --git a/embassy-stm32/src/subghz/mod.rs b/embassy-stm32/src/subghz/mod.rs
index 7d74569f3..87f376c40 100644
--- a/embassy-stm32/src/subghz/mod.rs
+++ b/embassy-stm32/src/subghz/mod.rs
@@ -82,14 +82,10 @@ use crate::{
     pac,
     peripherals::SUBGHZSPI,
     rcc::sealed::RccPeripheral,
-    spi::{BitOrder, Config as SpiConfig, MisoPin, MosiPin, SckPin, Spi},
+    spi::{BitOrder, Config as SpiConfig, MisoPin, MosiPin, SckPin, Spi, MODE_0},
     time::Hertz,
 };
 use embassy::util::Unborrow;
-use embedded_hal::{
-    blocking::spi::{Transfer, Write},
-    spi::MODE_0,
-};
 
 /// Passthrough for SPI errors (for now)
 pub type Error = crate::spi::Error;
@@ -255,8 +251,8 @@ impl<'d> SubGhz<'d, NoDma, NoDma> {
         self.poll_not_busy();
         {
             let _nss: Nss = Nss::new();
-            self.spi.write(&[opcode as u8])?;
-            self.spi.transfer(data)?;
+            self.spi.blocking_write(&[opcode as u8])?;
+            self.spi.blocking_transfer_in_place(data)?;
         }
         self.poll_not_busy();
         Ok(())
@@ -280,7 +276,7 @@ impl<'d> SubGhz<'d, NoDma, NoDma> {
         self.poll_not_busy();
         {
             let _nss: Nss = Nss::new();
-            self.spi.write(data)?;
+            self.spi.blocking_write(data)?;
         }
         self.poll_not_busy();
         Ok(())
@@ -290,8 +286,9 @@ impl<'d> SubGhz<'d, NoDma, NoDma> {
         self.poll_not_busy();
         {
             let _nss: Nss = Nss::new();
-            self.spi.write(&[OpCode::WriteBuffer as u8, offset])?;
-            self.spi.write(data)?;
+            self.spi
+                .blocking_write(&[OpCode::WriteBuffer as u8, offset])?;
+            self.spi.blocking_write(data)?;
         }
         self.poll_not_busy();
 
@@ -308,9 +305,10 @@ impl<'d> SubGhz<'d, NoDma, NoDma> {
         self.poll_not_busy();
         {
             let _nss: Nss = Nss::new();
-            self.spi.write(&[OpCode::ReadBuffer as u8, offset])?;
-            self.spi.transfer(&mut status_buf)?;
-            self.spi.transfer(buf)?;
+            self.spi
+                .blocking_write(&[OpCode::ReadBuffer as u8, offset])?;
+            self.spi.blocking_transfer_in_place(&mut status_buf)?;
+            self.spi.blocking_transfer_in_place(buf)?;
         }
         self.poll_not_busy();
 
@@ -342,8 +340,8 @@ impl<'d> SubGhz<'d, NoDma, NoDma> {
         {
             let _nss: Nss = Nss::new();
             self.spi
-                .write(&[OpCode::WriteRegister as u8, addr[0], addr[1]])?;
-            self.spi.write(data)?;
+                .blocking_write(&[OpCode::WriteRegister as u8, addr[0], addr[1]])?;
+            self.spi.blocking_write(data)?;
         }
         self.poll_not_busy();
 
diff --git a/examples/stm32f4/src/bin/spi.rs b/examples/stm32f4/src/bin/spi.rs
index b66eb9582..6b04f1fed 100644
--- a/examples/stm32f4/src/bin/spi.rs
+++ b/examples/stm32f4/src/bin/spi.rs
@@ -10,7 +10,6 @@ use embassy_stm32::dma::NoDma;
 use embassy_stm32::gpio::{Level, Output, Speed};
 use embassy_stm32::spi::{Config, Spi};
 use embassy_stm32::time::Hertz;
-use embedded_hal::blocking::spi::Transfer;
 use example_common::*;
 
 #[entry]
@@ -35,7 +34,7 @@ fn main() -> ! {
     loop {
         let mut buf = [0x0Au8; 4];
         cs.set_low();
-        unwrap!(spi.transfer(&mut buf));
+        unwrap!(spi.blocking_transfer_in_place(&mut buf));
         cs.set_high();
         info!("xfer {=[u8]:x}", buf);
     }
diff --git a/examples/stm32f4/src/bin/spi_dma.rs b/examples/stm32f4/src/bin/spi_dma.rs
index b3bf6fc28..9171f7516 100644
--- a/examples/stm32f4/src/bin/spi_dma.rs
+++ b/examples/stm32f4/src/bin/spi_dma.rs
@@ -10,7 +10,6 @@ use embassy::executor::Spawner;
 use embassy_stm32::spi::{Config, Spi};
 use embassy_stm32::time::Hertz;
 use embassy_stm32::Peripherals;
-use embassy_traits::spi::FullDuplex;
 use example_common::*;
 use heapless::String;
 
@@ -33,7 +32,7 @@ async fn main(_spawner: Spawner, p: Peripherals) {
         let mut write: String<128> = String::new();
         let mut read = [0; 128];
         core::write!(&mut write, "Hello DMA World {}!\r\n", n).unwrap();
-        spi.read_write(&mut read[0..write.len()], write.as_bytes())
+        spi.transfer(&mut read[0..write.len()], write.as_bytes())
             .await
             .ok();
         info!("read via spi+dma: {}", from_utf8(&read).unwrap());
diff --git a/examples/stm32h7/src/bin/spi.rs b/examples/stm32h7/src/bin/spi.rs
index 0b375b0d0..17e64da79 100644
--- a/examples/stm32h7/src/bin/spi.rs
+++ b/examples/stm32h7/src/bin/spi.rs
@@ -10,7 +10,6 @@ use embassy::executor::Executor;
 use embassy::util::Forever;
 use embassy_stm32::dma::NoDma;
 use embassy_stm32::spi;
-use embedded_hal::blocking::spi::Transfer;
 use example_common::*;
 
 use core::str::from_utf8;
@@ -25,7 +24,7 @@ async fn main_task(mut spi: spi::Spi<'static, SPI3, NoDma, NoDma>) {
         let mut write: String<128> = String::new();
         core::write!(&mut write, "Hello DMA World {}!\r\n", n).unwrap();
         unsafe {
-            let result = spi.transfer(write.as_bytes_mut());
+            let result = spi.blocking_transfer_in_place(write.as_bytes_mut());
             if let Err(_) = result {
                 defmt::panic!("crap");
             }
diff --git a/examples/stm32l0/src/bin/spi.rs b/examples/stm32l0/src/bin/spi.rs
index d30bb8d7a..8d6e89d91 100644
--- a/examples/stm32l0/src/bin/spi.rs
+++ b/examples/stm32l0/src/bin/spi.rs
@@ -13,7 +13,6 @@ use embassy_stm32::dma::NoDma;
 use embassy_stm32::spi::{Config, Spi};
 use embassy_stm32::time::Hertz;
 use embassy_stm32::Peripherals;
-use embedded_hal::blocking::spi::Transfer;
 
 #[embassy::main]
 async fn main(_spawner: Spawner, p: Peripherals) {
@@ -35,7 +34,7 @@ async fn main(_spawner: Spawner, p: Peripherals) {
     loop {
         let mut buf = [0x0Au8; 4];
         cs.set_low();
-        unwrap!(spi.transfer(&mut buf));
+        unwrap!(spi.blocking_transfer_in_place(&mut buf));
         cs.set_high();
         info!("xfer {=[u8]:x}", buf);
     }
diff --git a/examples/stm32l1/src/bin/spi.rs b/examples/stm32l1/src/bin/spi.rs
index 9d1a2fc87..e97e3ebb4 100644
--- a/examples/stm32l1/src/bin/spi.rs
+++ b/examples/stm32l1/src/bin/spi.rs
@@ -13,7 +13,6 @@ use embassy_stm32::dma::NoDma;
 use embassy_stm32::spi::{Config, Spi};
 use embassy_stm32::time::Hertz;
 use embassy_stm32::Peripherals;
-use embedded_hal::blocking::spi::Transfer;
 
 #[embassy::main]
 async fn main(_spawner: Spawner, p: Peripherals) {
@@ -35,7 +34,7 @@ async fn main(_spawner: Spawner, p: Peripherals) {
     loop {
         let mut buf = [0x0Au8; 4];
         cs.set_low();
-        unwrap!(spi.transfer(&mut buf));
+        unwrap!(spi.blocking_transfer_in_place(&mut buf));
         cs.set_high();
         info!("xfer {=[u8]:x}", buf);
     }
diff --git a/examples/stm32l4/src/bin/spi.rs b/examples/stm32l4/src/bin/spi.rs
index 1b6e3946e..8567d3062 100644
--- a/examples/stm32l4/src/bin/spi.rs
+++ b/examples/stm32l4/src/bin/spi.rs
@@ -9,7 +9,6 @@ use embassy_stm32::dma::NoDma;
 use embassy_stm32::gpio::{Level, Output, Speed};
 use embassy_stm32::spi::{Config, Spi};
 use embassy_stm32::time::Hertz;
-use embedded_hal::blocking::spi::Transfer;
 use example_common::*;
 
 #[cortex_m_rt::entry]
@@ -34,7 +33,7 @@ fn main() -> ! {
     loop {
         let mut buf = [0x0Au8; 4];
         cs.set_low();
-        unwrap!(spi.transfer(&mut buf));
+        unwrap!(spi.blocking_transfer_in_place(&mut buf));
         cs.set_high();
         info!("xfer {=[u8]:x}", buf);
     }
diff --git a/tests/stm32/src/bin/spi.rs b/tests/stm32/src/bin/spi.rs
index 043505c7b..47d0017ac 100644
--- a/tests/stm32/src/bin/spi.rs
+++ b/tests/stm32/src/bin/spi.rs
@@ -10,7 +10,6 @@ use embassy_stm32::dma::NoDma;
 use embassy_stm32::spi::{self, Spi};
 use embassy_stm32::time::Hertz;
 use embassy_stm32::Peripherals;
-use embedded_hal::blocking::spi::Transfer;
 use example_common::*;
 
 #[embassy::main(config = "config()")]
@@ -38,7 +37,7 @@ async fn main(_spawner: Spawner, p: Peripherals) {
     // Arduino pins D11 and D12 (MOSI-MISO) are connected together with a 1K resistor.
     // so we should get the data we sent back.
     let mut buf = data;
-    spi.transfer(&mut buf).unwrap();
+    spi.blocking_transfer_in_place(&mut buf).unwrap();
     assert_eq!(buf, data);
 
     info!("Test OK");