#![macro_use] use core::future::poll_fn; use core::sync::atomic::{compiler_fence, Ordering}; use core::task::Poll; use embassy_embedded_hal::SetConfig; use embassy_hal_common::{into_ref, PeripheralRef}; pub use embedded_hal_02::spi::{Mode, Phase, Polarity, MODE_0, MODE_1, MODE_2, MODE_3}; use crate::chip::FORCE_COPY_BUFFER_SIZE; use crate::gpio::sealed::Pin as _; use crate::gpio::{self, AnyPin, Pin as GpioPin, PselBits}; use crate::interrupt::{Interrupt, InterruptExt}; use crate::util::{slice_in_ram_or, slice_ptr_parts, slice_ptr_parts_mut}; use crate::{pac, Peripheral}; #[derive(Debug, Clone, Copy, PartialEq, Eq)] #[cfg_attr(feature = "defmt", derive(defmt::Format))] #[non_exhaustive] pub enum Error { TxBufferTooLong, RxBufferTooLong, /// EasyDMA can only read from data memory, read only buffers in flash will fail. DMABufferNotInDataMemory, } /// Interface for the SPIS peripheral using EasyDMA to offload the transmission and reception workload. /// /// For more details about EasyDMA, consult the module documentation. pub struct Spis<'d, T: Instance> { _p: PeripheralRef<'d, T>, } #[non_exhaustive] pub struct Config { pub mode: Mode, pub orc: u8, pub def: u8, pub auto_acquire: bool, } impl Default for Config { fn default() -> Self { Self { mode: MODE_0, orc: 0x00, def: 0x00, auto_acquire: true, } } } impl<'d, T: Instance> Spis<'d, T> { pub fn new( spis: impl Peripheral

+ 'd, irq: impl Peripheral

+ 'd, cs: impl Peripheral

+ 'd, sck: impl Peripheral

+ 'd, miso: impl Peripheral

+ 'd, mosi: impl Peripheral

+ 'd, config: Config, ) -> Self { into_ref!(cs, sck, miso, mosi); Self::new_inner( spis, irq, cs.map_into(), sck.map_into(), Some(miso.map_into()), Some(mosi.map_into()), config, ) } pub fn new_txonly( spis: impl Peripheral

+ 'd, irq: impl Peripheral

+ 'd, cs: impl Peripheral

+ 'd, sck: impl Peripheral

+ 'd, mosi: impl Peripheral

+ 'd, config: Config, ) -> Self { into_ref!(cs, sck, mosi); Self::new_inner( spis, irq, cs.map_into(), sck.map_into(), None, Some(mosi.map_into()), config, ) } pub fn new_rxonly( spis: impl Peripheral

+ 'd, irq: impl Peripheral

+ 'd, cs: impl Peripheral

+ 'd, sck: impl Peripheral

+ 'd, miso: impl Peripheral

+ 'd, config: Config, ) -> Self { into_ref!(cs, sck, miso); Self::new_inner( spis, irq, cs.map_into(), sck.map_into(), Some(miso.map_into()), None, config, ) } fn new_inner( spis: impl Peripheral

+ 'd, irq: impl Peripheral

+ 'd, cs: PeripheralRef<'d, AnyPin>, sck: PeripheralRef<'d, AnyPin>, miso: Option>, mosi: Option>, config: Config, ) -> Self { into_ref!(cs, spis, irq); let r = T::regs(); // Configure pins. sck.conf().write(|w| w.input().connect().drive().h0h1()); cs.conf().write(|w| w.input().connect().drive().h0h1()); if let Some(mosi) = &mosi { mosi.conf().write(|w| w.input().connect().drive().h0h1()); } if let Some(miso) = &miso { miso.conf().write(|w| w.dir().output().drive().h0h1()); } match config.mode.polarity { Polarity::IdleHigh => { sck.set_high(); if let Some(mosi) = &mosi { mosi.set_high(); } } Polarity::IdleLow => { sck.set_low(); if let Some(mosi) = &mosi { mosi.set_low(); } } } // Select pins. r.psel.sck.write(|w| unsafe { w.bits(sck.psel_bits()) }); r.psel.csn.write(|w| unsafe { w.bits(cs.psel_bits()) }); r.psel.mosi.write(|w| unsafe { w.bits(mosi.psel_bits()) }); r.psel.miso.write(|w| unsafe { w.bits(miso.psel_bits()) }); // Enable SPIS instance. r.enable.write(|w| w.enable().enabled()); // Configure mode. let mode = config.mode; r.config.write(|w| { match mode { MODE_0 => { w.order().msb_first(); w.cpol().active_high(); w.cpha().leading(); } MODE_1 => { w.order().msb_first(); w.cpol().active_high(); w.cpha().trailing(); } MODE_2 => { w.order().msb_first(); w.cpol().active_low(); w.cpha().leading(); } MODE_3 => { w.order().msb_first(); w.cpol().active_low(); w.cpha().trailing(); } } w }); // Set over-read character. let orc = config.orc; r.orc.write(|w| unsafe { w.orc().bits(orc) }); // Set default character. let def = config.def; r.def.write(|w| unsafe { w.def().bits(def) }); // Configure auto-acquire on 'transfer end' event. if config.auto_acquire { r.shorts.write(|w| w.end_acquire().bit(true)); } // Disable all events interrupts. r.intenclr.write(|w| unsafe { w.bits(0xFFFF_FFFF) }); irq.set_handler(Self::on_interrupt); irq.unpend(); irq.enable(); Self { _p: spis } } fn on_interrupt(_: *mut ()) { let r = T::regs(); let s = T::state(); if r.events_end.read().bits() != 0 { s.end_waker.wake(); r.intenclr.write(|w| w.end().clear()); } if r.events_acquired.read().bits() != 0 { s.acquire_waker.wake(); r.intenclr.write(|w| w.acquired().clear()); } } fn prepare(&mut self, rx: *mut [u8], tx: *const [u8]) -> Result<(), Error> { slice_in_ram_or(tx, Error::DMABufferNotInDataMemory)?; // NOTE: RAM slice check for rx is not necessary, as a mutable // slice can only be built from data located in RAM. compiler_fence(Ordering::SeqCst); let r = T::regs(); // Set up the DMA write. let (ptr, len) = slice_ptr_parts(tx); r.txd.ptr.write(|w| unsafe { w.ptr().bits(ptr as _) }); r.txd.maxcnt.write(|w| unsafe { w.maxcnt().bits(len as _) }); // Set up the DMA read. let (ptr, len) = slice_ptr_parts_mut(rx); r.rxd.ptr.write(|w| unsafe { w.ptr().bits(ptr as _) }); r.rxd.maxcnt.write(|w| unsafe { w.maxcnt().bits(len as _) }); // Reset and enable the end event. r.events_end.reset(); r.intenset.write(|w| w.end().set()); // Release the semaphore. r.tasks_release.write(|w| unsafe { w.bits(1) }); Ok(()) } fn blocking_inner_from_ram(&mut self, rx: *mut [u8], tx: *const [u8]) -> Result<(usize, usize), Error> { compiler_fence(Ordering::SeqCst); let r = T::regs(); // Acquire semaphore. if r.semstat.read().bits() != 1 { r.events_acquired.reset(); r.tasks_acquire.write(|w| unsafe { w.bits(1) }); // Wait until CPU has acquired the semaphore. while r.semstat.read().bits() != 1 {} } self.prepare(rx, tx)?; // Wait for 'end' event. while r.events_end.read().bits() == 0 {} let n_rx = r.rxd.amount.read().bits() as usize; let n_tx = r.txd.amount.read().bits() as usize; compiler_fence(Ordering::SeqCst); Ok((n_rx, n_tx)) } fn blocking_inner(&mut self, rx: &mut [u8], tx: &[u8]) -> Result<(usize, usize), Error> { match self.blocking_inner_from_ram(rx, tx) { Ok(n) => Ok(n), Err(Error::DMABufferNotInDataMemory) => { trace!("Copying SPIS tx buffer into RAM for DMA"); let tx_ram_buf = &mut [0; FORCE_COPY_BUFFER_SIZE][..tx.len()]; tx_ram_buf.copy_from_slice(tx); self.blocking_inner_from_ram(rx, tx_ram_buf) } Err(error) => Err(error), } } async fn async_inner_from_ram(&mut self, rx: *mut [u8], tx: *const [u8]) -> Result<(usize, usize), Error> { let r = T::regs(); let s = T::state(); // Clear status register. r.status.write(|w| w.overflow().clear().overread().clear()); // Acquire semaphore. if r.semstat.read().bits() != 1 { // Reset and enable the acquire event. r.events_acquired.reset(); r.intenset.write(|w| w.acquired().set()); // Request acquiring the SPIS semaphore. r.tasks_acquire.write(|w| unsafe { w.bits(1) }); // Wait until CPU has acquired the semaphore. poll_fn(|cx| { s.acquire_waker.register(cx.waker()); if r.semstat.read().bits() == 1 { return Poll::Ready(()); } Poll::Pending }) .await; } self.prepare(rx, tx)?; // Wait for 'end' event. poll_fn(|cx| { s.end_waker.register(cx.waker()); if r.events_end.read().bits() != 0 { return Poll::Ready(()); } Poll::Pending }) .await; let n_rx = r.rxd.amount.read().bits() as usize; let n_tx = r.txd.amount.read().bits() as usize; compiler_fence(Ordering::SeqCst); Ok((n_rx, n_tx)) } async fn async_inner(&mut self, rx: &mut [u8], tx: &[u8]) -> Result<(usize, usize), Error> { match self.async_inner_from_ram(rx, tx).await { Ok(n) => Ok(n), Err(Error::DMABufferNotInDataMemory) => { trace!("Copying SPIS tx buffer into RAM for DMA"); let tx_ram_buf = &mut [0; FORCE_COPY_BUFFER_SIZE][..tx.len()]; tx_ram_buf.copy_from_slice(tx); self.async_inner_from_ram(rx, tx_ram_buf).await } Err(error) => Err(error), } } /// Reads data from the SPI bus without sending anything. Blocks until the buffer has been filled. /// Returns number of bytes read. pub fn blocking_read(&mut self, data: &mut [u8]) -> Result { self.blocking_inner(data, &[]).map(|n| n.0) } /// Simultaneously sends and receives data. Blocks until the transmission is completed. /// If necessary, the write buffer will be copied into RAM (see struct description for detail). /// Returns number of bytes transferred `(n_rx, n_tx)`. pub fn blocking_transfer(&mut self, read: &mut [u8], write: &[u8]) -> Result<(usize, usize), Error> { self.blocking_inner(read, write) } /// Same as [`blocking_transfer`](Spis::blocking_transfer) but will fail instead of copying data into RAM. Consult the module level documentation to learn more. /// Returns number of bytes transferred `(n_rx, n_tx)`. pub fn blocking_transfer_from_ram(&mut self, read: &mut [u8], write: &[u8]) -> Result<(usize, usize), Error> { self.blocking_inner(read, write) } /// Simultaneously sends and receives data. /// Places the received data into the same buffer and blocks until the transmission is completed. /// Returns number of bytes transferred. pub fn blocking_transfer_in_place(&mut self, data: &mut [u8]) -> Result { self.blocking_inner_from_ram(data, data).map(|n| n.0) } /// Sends data, discarding any received data. Blocks until the transmission is completed. /// If necessary, the write buffer will be copied into RAM (see struct description for detail). /// Returns number of bytes written. pub fn blocking_write(&mut self, data: &[u8]) -> Result { self.blocking_inner(&mut [], data).map(|n| n.1) } /// Same as [`blocking_write`](Spis::blocking_write) but will fail instead of copying data into RAM. Consult the module level documentation to learn more. /// Returns number of bytes written. pub fn blocking_write_from_ram(&mut self, data: &[u8]) -> Result { self.blocking_inner(&mut [], data).map(|n| n.1) } /// Reads data from the SPI bus without sending anything. /// Returns number of bytes read. pub async fn read(&mut self, data: &mut [u8]) -> Result { self.async_inner(data, &[]).await.map(|n| n.0) } /// Simultaneously sends and receives data. /// If necessary, the write buffer will be copied into RAM (see struct description for detail). /// Returns number of bytes transferred `(n_rx, n_tx)`. pub async fn transfer(&mut self, read: &mut [u8], write: &[u8]) -> Result<(usize, usize), Error> { self.async_inner(read, write).await } /// Same as [`transfer`](Spis::transfer) but will fail instead of copying data into RAM. Consult the module level documentation to learn more. /// Returns number of bytes transferred `(n_rx, n_tx)`. pub async fn transfer_from_ram(&mut self, read: &mut [u8], write: &[u8]) -> Result<(usize, usize), Error> { self.async_inner_from_ram(read, write).await } /// Simultaneously sends and receives data. Places the received data into the same buffer. /// Returns number of bytes transferred. pub async fn transfer_in_place(&mut self, data: &mut [u8]) -> Result { self.async_inner_from_ram(data, data).await.map(|n| n.0) } /// Sends data, discarding any received data. /// If necessary, the write buffer will be copied into RAM (see struct description for detail). /// Returns number of bytes written. pub async fn write(&mut self, data: &[u8]) -> Result { self.async_inner(&mut [], data).await.map(|n| n.1) } /// Same as [`write`](Spis::write) but will fail instead of copying data into RAM. Consult the module level documentation to learn more. /// Returns number of bytes written. pub async fn write_from_ram(&mut self, data: &[u8]) -> Result { self.async_inner_from_ram(&mut [], data).await.map(|n| n.1) } /// Checks if last transaction overread. pub fn is_overread(&mut self) -> bool { T::regs().status.read().overread().is_present() } /// Checks if last transaction overflowed. pub fn is_overflow(&mut self) -> bool { T::regs().status.read().overflow().is_present() } } impl<'d, T: Instance> Drop for Spis<'d, T> { fn drop(&mut self) { trace!("spis drop"); // Disable let r = T::regs(); r.enable.write(|w| w.enable().disabled()); gpio::deconfigure_pin(r.psel.sck.read().bits()); gpio::deconfigure_pin(r.psel.csn.read().bits()); gpio::deconfigure_pin(r.psel.miso.read().bits()); gpio::deconfigure_pin(r.psel.mosi.read().bits()); trace!("spis drop: done"); } } pub(crate) mod sealed { use embassy_sync::waitqueue::AtomicWaker; use super::*; pub struct State { pub end_waker: AtomicWaker, pub acquire_waker: AtomicWaker, } impl State { pub const fn new() -> Self { Self { end_waker: AtomicWaker::new(), acquire_waker: AtomicWaker::new(), } } } pub trait Instance { fn regs() -> &'static pac::spis0::RegisterBlock; fn state() -> &'static State; } } pub trait Instance: Peripheral

+ sealed::Instance + 'static { type Interrupt: Interrupt; } macro_rules! impl_spis { ($type:ident, $pac_type:ident, $irq:ident) => { impl crate::spis::sealed::Instance for peripherals::$type { fn regs() -> &'static pac::spis0::RegisterBlock { unsafe { &*pac::$pac_type::ptr() } } fn state() -> &'static crate::spis::sealed::State { static STATE: crate::spis::sealed::State = crate::spis::sealed::State::new(); &STATE } } impl crate::spis::Instance for peripherals::$type { type Interrupt = crate::interrupt::$irq; } }; } // ==================== impl<'d, T: Instance> SetConfig for Spis<'d, T> { type Config = Config; fn set_config(&mut self, config: &Self::Config) { let r = T::regs(); // Configure mode. let mode = config.mode; r.config.write(|w| { match mode { MODE_0 => { w.order().msb_first(); w.cpol().active_high(); w.cpha().leading(); } MODE_1 => { w.order().msb_first(); w.cpol().active_high(); w.cpha().trailing(); } MODE_2 => { w.order().msb_first(); w.cpol().active_low(); w.cpha().leading(); } MODE_3 => { w.order().msb_first(); w.cpol().active_low(); w.cpha().trailing(); } } w }); // Set over-read character. let orc = config.orc; r.orc.write(|w| unsafe { w.orc().bits(orc) }); // Set default character. let def = config.def; r.def.write(|w| unsafe { w.def().bits(def) }); // Configure auto-acquire on 'transfer end' event. let auto_acquire = config.auto_acquire; r.shorts.write(|w| w.end_acquire().bit(auto_acquire)); } }