Merge #1360 #1361

1360: stm32/rcc: add i2s pll on some f4 micros r=Dirbaio a=xoviat

Adds the i2s pll on some f4 micros. 

1361: Executor: Replace unnecessary atomics in runqueue r=Dirbaio a=GrantM11235

Only the head pointer needs to be atomic. The `RunQueueItem` pointers are only loaded and stored, and never concurrently

Co-authored-by: xoviat <xoviat@users.noreply.github.com>
Co-authored-by: Grant Miller <GrantM11235@gmail.com>

embassy-executor/src/raw/run_queue.rs

@@ -4,15 +4,16 @@ use core::ptr::NonNull;
 use atomic_polyfill::{AtomicPtr, Ordering};
 
 use super::{TaskHeader, TaskRef};
+use crate::raw::util::SyncUnsafeCell;
 
 pub(crate) struct RunQueueItem {
-    next: AtomicPtr<TaskHeader>,
+    next: SyncUnsafeCell<Option<TaskRef>>,
 }
 
 impl RunQueueItem {
     pub const fn new() -> Self {
         Self {
-            next: AtomicPtr::new(ptr::null_mut()),
+            next: SyncUnsafeCell::new(None),
         }
     }
 }
@@ -51,7 +52,12 @@ impl RunQueue {
         self.head
             .fetch_update(Ordering::SeqCst, Ordering::SeqCst, |prev| {
                 was_empty = prev.is_null();
-                task.header().run_queue_item.next.store(prev, Ordering::Relaxed);
+                unsafe {
+                    // safety: the pointer is either null or valid
+                    let prev = NonNull::new(prev).map(|ptr| TaskRef::from_ptr(ptr.as_ptr()));
+                    // safety: there are no concurrent accesses to `next`
+                    task.header().run_queue_item.next.set(prev);
+                }
                 Some(task.as_ptr() as *mut _)
             })
             .ok();
@@ -64,18 +70,19 @@ impl RunQueue {
     /// and will be processed by the *next* call to `dequeue_all`, *not* the current one.
     pub(crate) fn dequeue_all(&self, on_task: impl Fn(TaskRef)) {
         // Atomically empty the queue.
-        let mut ptr = self.head.swap(ptr::null_mut(), Ordering::AcqRel);
+        let ptr = self.head.swap(ptr::null_mut(), Ordering::AcqRel);
+
+        // safety: the pointer is either null or valid
+        let mut next = unsafe { NonNull::new(ptr).map(|ptr| TaskRef::from_ptr(ptr.as_ptr())) };
 
         // Iterate the linked list of tasks that were previously in the queue.
-        while let Some(task) = NonNull::new(ptr) {
+        while let Some(task) = next {
-            let task = unsafe { TaskRef::from_ptr(task.as_ptr()) };
             // If the task re-enqueues itself, the `next` pointer will get overwritten.
             // Therefore, first read the next pointer, and only then process the task.
-            let next = task.header().run_queue_item.next.load(Ordering::Relaxed);
+            // safety: there are no concurrent accesses to `next`
+            next = unsafe { task.header().run_queue_item.next.get() };
 
             on_task(task);
-
-            ptr = next
         }
     }
 }

embassy-stm32/src/rcc/f4.rs

@@ -29,10 +29,66 @@ pub struct Config {
     pub pclk1: Option<Hertz>,
     pub pclk2: Option<Hertz>,
 
+    #[cfg(not(any(stm32f410, stm32f411, stm32f412, stm32f413, stm32f423, stm32f446)))]
+    pub plli2s: Option<Hertz>,
+
     pub pll48: bool,
 }
 
-unsafe fn setup_pll(pllsrcclk: u32, use_hse: bool, pllsysclk: Option<u32>, pll48clk: bool) -> PllResults {
+#[cfg(stm32f410)]
+unsafe fn setup_i2s_pll(_vco_in: u32, _plli2s: Option<u32>) -> Option<u32> {
+    None
+}
+
+// Not currently implemented, but will be in the future
+#[cfg(any(stm32f411, stm32f412, stm32f413, stm32f423, stm32f446))]
+unsafe fn setup_i2s_pll(_vco_in: u32, _plli2s: Option<u32>) -> Option<u32> {
+    None
+}
+
+#[cfg(not(any(stm32f410, stm32f411, stm32f412, stm32f413, stm32f423, stm32f446)))]
+unsafe fn setup_i2s_pll(vco_in: u32, plli2s: Option<u32>) -> Option<u32> {
+    let min_div = 2;
+    let max_div = 7;
+    let target = match plli2s {
+        Some(target) => target,
+        None => return None,
+    };
+
+    // We loop through the possible divider values to find the best configuration. Looping
+    // through all possible "N" values would result in more iterations.
+    let (n, outdiv, output, _error) = (min_div..=max_div)
+        .filter_map(|outdiv| {
+            let target_vco_out = match target.checked_mul(outdiv) {
+                Some(x) => x,
+                None => return None,
+            };
+            let n = (target_vco_out + (vco_in >> 1)) / vco_in;
+            let vco_out = vco_in * n;
+            if !(100_000_000..=432_000_000).contains(&vco_out) {
+                return None;
+            }
+            let output = vco_out / outdiv;
+            let error = (output as i32 - target as i32).unsigned_abs();
+            Some((n, outdiv, output, error))
+        })
+        .min_by_key(|(_, _, _, error)| *error)?;
+
+    RCC.plli2scfgr().modify(|w| {
+        w.set_plli2sn(n as u16);
+        w.set_plli2sr(outdiv as u8);
+    });
+
+    Some(output)
+}
+
+unsafe fn setup_pll(
+    pllsrcclk: u32,
+    use_hse: bool,
+    pllsysclk: Option<u32>,
+    plli2s: Option<u32>,
+    pll48clk: bool,
+) -> PllResults {
     use crate::pac::rcc::vals::{Pllp, Pllsrc};
 
     let sysclk = pllsysclk.unwrap_or(pllsrcclk);
@@ -43,6 +99,7 @@ unsafe fn setup_pll(pllsrcclk: u32, use_hse: bool, pllsysclk: Option<u32>, pll48
             use_pll: false,
             pllsysclk: None,
             pll48clk: None,
+            plli2sclk: None,
         };
     }
     // Input divisor from PLL source clock, must result to frequency in
@@ -101,6 +158,7 @@ unsafe fn setup_pll(pllsrcclk: u32, use_hse: bool, pllsysclk: Option<u32>, pll48
         use_pll: true,
         pllsysclk: Some(real_pllsysclk),
         pll48clk: if pll48clk { Some(real_pll48clk) } else { None },
+        plli2sclk: setup_i2s_pll(vco_in, plli2s),
     }
 }
 
@@ -286,6 +344,10 @@ pub(crate) unsafe fn init(config: Config) {
         pllsrcclk,
         config.hse.is_some(),
         if sysclk_on_pll { Some(sysclk) } else { None },
+        #[cfg(not(any(stm32f410, stm32f411, stm32f412, stm32f413, stm32f423, stm32f446)))]
+        config.plli2s.map(|i2s| i2s.0),
+        #[cfg(any(stm32f410, stm32f411, stm32f412, stm32f413, stm32f423, stm32f446))]
+        None,
         config.pll48,
     );
 
@@ -376,6 +438,13 @@ pub(crate) unsafe fn init(config: Config) {
         while !RCC.cr().read().pllrdy() {}
     }
 
+    #[cfg(not(stm32f410))]
+    if plls.plli2sclk.is_some() {
+        RCC.cr().modify(|w| w.set_plli2son(true));
+
+        while !RCC.cr().read().plli2srdy() {}
+    }
+
     RCC.cfgr().modify(|w| {
         w.set_ppre2(Ppre(ppre2_bits));
         w.set_ppre1(Ppre(ppre1_bits));
@@ -409,6 +478,12 @@ pub(crate) unsafe fn init(config: Config) {
         ahb3: Hertz(hclk),
 
         pll48: plls.pll48clk.map(Hertz),
+
+        #[cfg(not(stm32f410))]
+        plli2s: plls.plli2sclk.map(Hertz),
+
+        #[cfg(any(stm32f427, stm32f429, stm32f437, stm32f439, stm32f446, stm32f469, stm32f479))]
+        pllsai: None,
     });
 }
 
@@ -416,6 +491,8 @@ struct PllResults {
     use_pll: bool,
     pllsysclk: Option<u32>,
     pll48clk: Option<u32>,
+    #[allow(dead_code)]
+    plli2sclk: Option<u32>,
 }
 
 mod max {

embassy-stm32/src/rcc/mod.rs

@@ -60,6 +60,12 @@ pub struct Clocks {
     #[cfg(any(rcc_f2, rcc_f4, rcc_f410, rcc_f7))]
     pub pll48: Option<Hertz>,
 
+    #[cfg(all(rcc_f4, not(stm32f410)))]
+    pub plli2s: Option<Hertz>,
+
+    #[cfg(any(stm32f427, stm32f429, stm32f437, stm32f439, stm32f446, stm32f469, stm32f479))]
+    pub pllsai: Option<Hertz>,
+
     #[cfg(stm32f1)]
     pub adc: Hertz,