diff --git a/embassy-stm32/src/dma/bdma.rs b/embassy-stm32/src/dma/bdma.rs
index e2da2a8e..46670e1b 100644
--- a/embassy-stm32/src/dma/bdma.rs
+++ b/embassy-stm32/src/dma/bdma.rs
@@ -47,6 +47,7 @@ pub(crate) unsafe fn do_transfer(
     peri_addr: *const u8,
     mem_addr: *mut u8,
     mem_len: usize,
+    incr_mem: bool,
     #[cfg(dmamux)] dmamux_regs: pac::dmamux::Dmamux,
     #[cfg(dmamux)] dmamux_ch_num: u8,
 ) -> impl Future<Output = ()> {
@@ -182,6 +183,7 @@ pac::dma_channels! {
                         src,
                         buf.as_mut_ptr(),
                         buf.len(),
+                        true,
                         #[cfg(dmamux)]
                         <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_REGS,
                         #[cfg(dmamux)]
@@ -206,6 +208,33 @@ pac::dma_channels! {
                         dst,
                         buf.as_ptr() as *mut u8,
                         buf.len(),
+                        true,
+                        #[cfg(dmamux)]
+                        <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_REGS,
+                        #[cfg(dmamux)]
+                        <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_CH_NUM,
+                    )
+                }
+            }
+
+            fn write_x<'a>(
+                &'a mut self,
+                request: Request,
+                word: &u8,
+                count: usize,
+                dst: *mut u8,
+            ) -> Self::WriteFuture<'a> {
+                unsafe {
+                    do_transfer(
+                        crate::pac::$dma_peri,
+                        $channel_num,
+                        (dma_num!($dma_peri) * 8) + $channel_num,
+                        request,
+                        vals::Dir::FROMMEMORY,
+                        dst,
+                        word as *const u8 as *mut u8,
+                        count,
+                        false,
                         #[cfg(dmamux)]
                         <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_REGS,
                         #[cfg(dmamux)]
diff --git a/embassy-stm32/src/dma/dma.rs b/embassy-stm32/src/dma/dma.rs
index cf0889a3..9ac6df15 100644
--- a/embassy-stm32/src/dma/dma.rs
+++ b/embassy-stm32/src/dma/dma.rs
@@ -48,6 +48,7 @@ pub(crate) unsafe fn do_transfer(
     peri_addr: *const u8,
     mem_addr: *mut u8,
     mem_len: usize,
+    incr_mem: bool,
     #[cfg(dmamux)] dmamux_regs: pac::dmamux::Dmamux,
     #[cfg(dmamux)] dmamux_ch_num: u8,
 ) -> impl Future<Output = ()> {
@@ -187,6 +188,7 @@ pac::dma_channels! {
                         src,
                         buf.as_mut_ptr(),
                         buf.len(),
+                        true,
                         #[cfg(dmamux)]
                         <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_REGS,
                         #[cfg(dmamux)]
@@ -211,6 +213,33 @@ pac::dma_channels! {
                         dst,
                         buf.as_ptr() as *mut u8,
                         buf.len(),
+                        true,
+                        #[cfg(dmamux)]
+                        <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_REGS,
+                        #[cfg(dmamux)]
+                        <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_CH_NUM,
+                    )
+                }
+            }
+
+            fn write_x<'a>(
+                &'a mut self,
+                request: Request,
+                word: &u8,
+                num: usize,
+                dst: *mut u8,
+            ) -> Self::WriteFuture<'a> {
+                unsafe {
+                    do_transfer(
+                        crate::pac::$dma_peri,
+                        $channel_num,
+                        (dma_num!($dma_peri) * 8) + $channel_num,
+                        request,
+                        vals::Dir::MEMORYTOPERIPHERAL,
+                        dst,
+                        word as *const u8 as *mut u8,
+                        num,
+                        false,
                         #[cfg(dmamux)]
                         <Self as super::dmamux::sealed::MuxChannel>::DMAMUX_REGS,
                         #[cfg(dmamux)]
diff --git a/embassy-stm32/src/dma/mod.rs b/embassy-stm32/src/dma/mod.rs
index fbf82b87..60f6a302 100644
--- a/embassy-stm32/src/dma/mod.rs
+++ b/embassy-stm32/src/dma/mod.rs
@@ -42,6 +42,14 @@ pub trait Channel: sealed::Channel {
         buf: &'a [u8],
         dst: *mut u8,
     ) -> Self::WriteFuture<'a>;
+
+    fn write_x<'a>(
+        &'a mut self,
+        request: Request,
+        word: &u8,
+        num: usize,
+        dst: *mut u8,
+    ) -> Self::WriteFuture<'a>;
 }
 
 pub struct NoDma;
diff --git a/embassy-stm32/src/spi/mod.rs b/embassy-stm32/src/spi/mod.rs
index 7e168644..f84d820b 100644
--- a/embassy-stm32/src/spi/mod.rs
+++ b/embassy-stm32/src/spi/mod.rs
@@ -1,7 +1,7 @@
 #![macro_use]
 
-#[cfg_attr(spi_v1, path = "v1.rs")]
-#[cfg_attr(spi_v2, path = "v2.rs")]
+//#[cfg_attr(spi_v1, path = "v1.rs")]
+//#[cfg_attr(spi_v2, path = "v2.rs")]
 #[cfg_attr(spi_v3, path = "v3.rs")]
 mod _version;
 use crate::{dma, peripherals, rcc::RccPeripheral};
diff --git a/embassy-stm32/src/spi/v3.rs b/embassy-stm32/src/spi/v3.rs
index 2df2b49b..a3f9b891 100644
--- a/embassy-stm32/src/spi/v3.rs
+++ b/embassy-stm32/src/spi/v3.rs
@@ -18,6 +18,8 @@ use embassy_extras::unborrow;
 use embassy_traits::spi as traits;
 pub use embedded_hal::spi::{Mode, Phase, Polarity, MODE_0, MODE_1, MODE_2, MODE_3};
 
+use futures::future::join;
+
 impl WordSize {
     fn dsize(&self) -> u8 {
         match self {
@@ -176,18 +178,58 @@ impl<'d, T: Instance, Tx, Rx> Spi<'d, T, Tx, Rx> {
     }
 
     #[allow(unused)]
-    async fn write_dma_u8(&mut self, write: &[u8]) -> Result<(), Error> {
+    async fn write_dma_u8(&mut self, write: &[u8]) -> Result<(), Error>
+    where
+        Tx: TxDmaChannel<T>,
+    {
+        let request = self.txdma.request();
+        let dst = T::regs().txdr().ptr() as *mut u8;
+        let f = self.txdma.write(request, write, dst);
+        unsafe {
+            T::regs().cfg1().modify(|reg| {
+                reg.set_txdmaen(true);
+            });
+        }
+
+        f.await;
+        Ok(())
+    }
+
+    #[allow(unused)]
+    async fn read_dma_u8(&mut self, read: &mut [u8]) -> Result<(), Error>
+    where
+        Tx: TxDmaChannel<T>,
+        Rx: RxDmaChannel<T>,
+    {
         unimplemented!()
     }
 
     #[allow(unused)]
-    async fn read_dma_u8(&mut self, read: &mut [u8]) -> Result<(), Error> {
-        unimplemented!()
-    }
+    async fn read_write_dma_u8(&mut self, read: &mut [u8], write: &[u8]) -> Result<(), Error>
+    where
+        Tx: TxDmaChannel<T>,
+        Rx: RxDmaChannel<T>,
+    {
+        let rx_request = self.rxdma.request();
+        let rx_src = T::regs().rxdr().ptr() as *mut u8;
+        let rx_f = self.rxdma.read(rx_request, rx_src, read);
 
-    #[allow(unused)]
-    async fn read_write_dma_u8(&mut self, read: &mut [u8], write: &[u8]) -> Result<(), Error> {
-        unimplemented!()
+        let tx_request = self.txdma.request();
+        let tx_dst = T::regs().txdr().ptr() as *mut u8;
+        let clock_byte = 0x00;
+        let tx_f = self
+            .txdma
+            .write_x(tx_request, &clock_byte, read.len(), tx_dst);
+
+        unsafe {
+            T::regs().cfg1().modify(|reg| {
+                reg.set_txdmaen(true);
+                reg.set_rxdmaen(true);
+            });
+        }
+
+        let r = join(tx_f, rx_f).await;
+        Ok(())
     }
 }