diff options
Diffstat (limited to 'rust/kernel/sync')
| -rw-r--r-- | rust/kernel/sync/arc.rs | 271 | ||||
| -rw-r--r-- | rust/kernel/sync/aref.rs | 199 | ||||
| -rw-r--r-- | rust/kernel/sync/atomic.rs | 850 | ||||
| -rw-r--r-- | rust/kernel/sync/atomic/internal.rs | 352 | ||||
| -rw-r--r-- | rust/kernel/sync/atomic/ordering.rs | 104 | ||||
| -rw-r--r-- | rust/kernel/sync/atomic/predefine.rs | 335 | ||||
| -rw-r--r-- | rust/kernel/sync/barrier.rs | 61 | ||||
| -rw-r--r-- | rust/kernel/sync/completion.rs | 114 | ||||
| -rw-r--r-- | rust/kernel/sync/condvar.rs | 39 | ||||
| -rw-r--r-- | rust/kernel/sync/lock.rs | 88 | ||||
| -rw-r--r-- | rust/kernel/sync/lock/global.rs | 16 | ||||
| -rw-r--r-- | rust/kernel/sync/lock/mutex.rs | 7 | ||||
| -rw-r--r-- | rust/kernel/sync/lock/spinlock.rs | 7 | ||||
| -rw-r--r-- | rust/kernel/sync/locked_by.rs | 2 | ||||
| -rw-r--r-- | rust/kernel/sync/poll.rs | 71 | ||||
| -rw-r--r-- | rust/kernel/sync/rcu.rs | 5 | ||||
| -rw-r--r-- | rust/kernel/sync/refcount.rs | 116 | ||||
| -rw-r--r-- | rust/kernel/sync/set_once.rs | 133 |
18 files changed, 2633 insertions, 137 deletions
diff --git a/rust/kernel/sync/arc.rs b/rust/kernel/sync/arc.rs index 3cefda7a4372..5ac4961b7cd2 100644 --- a/rust/kernel/sync/arc.rs +++ b/rust/kernel/sync/arc.rs @@ -8,7 +8,7 @@ //! threads. //! //! It is different from the standard library's [`Arc`] in a few ways: -//! 1. It is backed by the kernel's `refcount_t` type. +//! 1. It is backed by the kernel's [`Refcount`] type. //! 2. It does not support weak references, which allows it to be half the size. //! 3. It saturates the reference count instead of aborting when it goes over a threshold. //! 4. It does not provide a `get_mut` method, so the ref counted object is pinned. @@ -18,21 +18,23 @@ use crate::{ alloc::{AllocError, Flags, KBox}, - bindings, - init::{self, InPlaceInit, Init, PinInit}, + ffi::c_void, + fmt, + init::InPlaceInit, + sync::Refcount, try_init, - types::{ForeignOwnable, Opaque}, + types::ForeignOwnable, }; use core::{ alloc::Layout, - fmt, + borrow::{Borrow, BorrowMut}, marker::PhantomData, mem::{ManuallyDrop, MaybeUninit}, ops::{Deref, DerefMut}, pin::Pin, ptr::NonNull, }; -use macros::pin_data; +use pin_init::{self, pin_data, InPlaceWrite, Init, PinInit}; mod std_vendor; @@ -126,7 +128,7 @@ mod std_vendor; /// # Ok::<(), Error>(()) /// ``` #[repr(transparent)] -#[cfg_attr(CONFIG_RUSTC_HAS_COERCE_POINTEE, derive(core::marker::CoercePointee))] +#[derive(core::marker::CoercePointee)] pub struct Arc<T: ?Sized> { ptr: NonNull<ArcInner<T>>, // NB: this informs dropck that objects of type `ArcInner<T>` may be used in `<Arc<T> as @@ -135,7 +137,7 @@ pub struct Arc<T: ?Sized> { // meaningful with respect to dropck - but this may change in the future so this is left here // out of an abundance of caution. // - // See https://doc.rust-lang.org/nomicon/phantom-data.html#generic-parameters-and-drop-checking + // See <https://doc.rust-lang.org/nomicon/phantom-data.html#generic-parameters-and-drop-checking> // for more detail on the semantics of dropck in the presence of `PhantomData`. _p: PhantomData<ArcInner<T>>, } @@ -143,7 +145,7 @@ pub struct Arc<T: ?Sized> { #[pin_data] #[repr(C)] struct ArcInner<T: ?Sized> { - refcount: Opaque<bindings::refcount_t>, + refcount: Refcount, data: T, } @@ -155,7 +157,7 @@ impl<T: ?Sized> ArcInner<T> { /// `ptr` must have been returned by a previous call to [`Arc::into_raw`], and the `Arc` must /// not yet have been destroyed. unsafe fn container_of(ptr: *const T) -> NonNull<ArcInner<T>> { - let refcount_layout = Layout::new::<bindings::refcount_t>(); + let refcount_layout = Layout::new::<Refcount>(); // SAFETY: The caller guarantees that the pointer is valid. let val_layout = Layout::for_value(unsafe { &*ptr }); // SAFETY: We're computing the layout of a real struct that existed when compiling this @@ -180,15 +182,6 @@ impl<T: ?Sized> ArcInner<T> { } } -// This is to allow coercion from `Arc<T>` to `Arc<U>` if `T` can be converted to the -// dynamically-sized type (DST) `U`. -#[cfg(not(CONFIG_RUSTC_HAS_COERCE_POINTEE))] -impl<T: ?Sized + core::marker::Unsize<U>, U: ?Sized> core::ops::CoerceUnsized<Arc<U>> for Arc<T> {} - -// This is to allow `Arc<U>` to be dispatched on when `Arc<T>` can be coerced into `Arc<U>`. -#[cfg(not(CONFIG_RUSTC_HAS_COERCE_POINTEE))] -impl<T: ?Sized + core::marker::Unsize<U>, U: ?Sized> core::ops::DispatchFromDyn<Arc<U>> for Arc<T> {} - // SAFETY: It is safe to send `Arc<T>` to another thread when the underlying `T` is `Sync` because // it effectively means sharing `&T` (which is safe because `T` is `Sync`); additionally, it needs // `T` to be `Send` because any thread that has an `Arc<T>` may ultimately access `T` using a @@ -202,13 +195,32 @@ unsafe impl<T: ?Sized + Sync + Send> Send for Arc<T> {} // the reference count reaches zero and `T` is dropped. unsafe impl<T: ?Sized + Sync + Send> Sync for Arc<T> {} +impl<T> InPlaceInit<T> for Arc<T> { + type PinnedSelf = Self; + + #[inline] + fn try_pin_init<E>(init: impl PinInit<T, E>, flags: Flags) -> Result<Self::PinnedSelf, E> + where + E: From<AllocError>, + { + UniqueArc::try_pin_init(init, flags).map(|u| u.into()) + } + + #[inline] + fn try_init<E>(init: impl Init<T, E>, flags: Flags) -> Result<Self, E> + where + E: From<AllocError>, + { + UniqueArc::try_init(init, flags).map(|u| u.into()) + } +} + impl<T> Arc<T> { /// Constructs a new reference counted instance of `T`. pub fn new(contents: T, flags: Flags) -> Result<Self, AllocError> { // INVARIANT: The refcount is initialised to a non-zero value. let value = ArcInner { - // SAFETY: There are no safety requirements for this FFI call. - refcount: Opaque::new(unsafe { bindings::REFCOUNT_INIT(1) }), + refcount: Refcount::new(1), data: contents, }; @@ -219,6 +231,9 @@ impl<T> Arc<T> { // `Arc` object. Ok(unsafe { Self::from_inner(inner) }) } + + /// The offset that the value is stored at. + pub const DATA_OFFSET: usize = core::mem::offset_of!(ArcInner<T>, data); } impl<T: ?Sized> Arc<T> { @@ -246,6 +261,15 @@ impl<T: ?Sized> Arc<T> { unsafe { core::ptr::addr_of!((*ptr).data) } } + /// Return a raw pointer to the data in this arc. + pub fn as_ptr(this: &Self) -> *const T { + let ptr = this.ptr.as_ptr(); + + // SAFETY: As `ptr` points to a valid allocation of type `ArcInner`, + // field projection to `data`is within bounds of the allocation. + unsafe { core::ptr::addr_of!((*ptr).data) } + } + /// Recreates an [`Arc`] instance previously deconstructed via [`Arc::into_raw`]. /// /// # Safety @@ -290,7 +314,7 @@ impl<T: ?Sized> Arc<T> { /// use kernel::sync::{Arc, UniqueArc}; /// /// let arc = Arc::new(42, GFP_KERNEL)?; - /// let unique_arc = arc.into_unique_or_drop(); + /// let unique_arc = Arc::into_unique_or_drop(arc); /// /// // The above conversion should succeed since refcount of `arc` is 1. /// assert!(unique_arc.is_some()); @@ -306,35 +330,30 @@ impl<T: ?Sized> Arc<T> { /// let arc = Arc::new(42, GFP_KERNEL)?; /// let another = arc.clone(); /// - /// let unique_arc = arc.into_unique_or_drop(); + /// let unique_arc = Arc::into_unique_or_drop(arc); /// /// // The above conversion should fail since refcount of `arc` is >1. /// assert!(unique_arc.is_none()); /// /// # Ok::<(), Error>(()) /// ``` - pub fn into_unique_or_drop(self) -> Option<Pin<UniqueArc<T>>> { + pub fn into_unique_or_drop(this: Self) -> Option<Pin<UniqueArc<T>>> { // We will manually manage the refcount in this method, so we disable the destructor. - let me = ManuallyDrop::new(self); + let this = ManuallyDrop::new(this); // SAFETY: We own a refcount, so the pointer is still valid. - let refcount = unsafe { me.ptr.as_ref() }.refcount.get(); + let refcount = unsafe { &this.ptr.as_ref().refcount }; // If the refcount reaches a non-zero value, then we have destroyed this `Arc` and will // return without further touching the `Arc`. If the refcount reaches zero, then there are // no other arcs, and we can create a `UniqueArc`. - // - // SAFETY: We own a refcount, so the pointer is not dangling. - let is_zero = unsafe { bindings::refcount_dec_and_test(refcount) }; - if is_zero { - // SAFETY: We have exclusive access to the arc, so we can perform unsynchronized - // accesses to the refcount. - unsafe { core::ptr::write(refcount, bindings::REFCOUNT_INIT(1)) }; + if refcount.dec_and_test() { + refcount.set(1); // INVARIANT: We own the only refcount to this arc, so we may create a `UniqueArc`. We // must pin the `UniqueArc` because the values was previously in an `Arc`, and they pin // their values. Some(Pin::from(UniqueArc { - inner: ManuallyDrop::into_inner(me), + inner: ManuallyDrop::into_inner(this), })) } else { None @@ -342,15 +361,19 @@ impl<T: ?Sized> Arc<T> { } } -impl<T: 'static> ForeignOwnable for Arc<T> { +// SAFETY: The pointer returned by `into_foreign` was originally allocated as an +// `KBox<ArcInner<T>>`, so that type is what determines the alignment. +unsafe impl<T: 'static> ForeignOwnable for Arc<T> { + const FOREIGN_ALIGN: usize = <KBox<ArcInner<T>> as ForeignOwnable>::FOREIGN_ALIGN; + type Borrowed<'a> = ArcBorrow<'a, T>; type BorrowedMut<'a> = Self::Borrowed<'a>; - fn into_foreign(self) -> *mut crate::ffi::c_void { + fn into_foreign(self) -> *mut c_void { ManuallyDrop::new(self).ptr.as_ptr().cast() } - unsafe fn from_foreign(ptr: *mut crate::ffi::c_void) -> Self { + unsafe fn from_foreign(ptr: *mut c_void) -> Self { // SAFETY: The safety requirements of this function ensure that `ptr` comes from a previous // call to `Self::into_foreign`. let inner = unsafe { NonNull::new_unchecked(ptr.cast::<ArcInner<T>>()) }; @@ -361,7 +384,7 @@ impl<T: 'static> ForeignOwnable for Arc<T> { unsafe { Self::from_inner(inner) } } - unsafe fn borrow<'a>(ptr: *mut crate::ffi::c_void) -> ArcBorrow<'a, T> { + unsafe fn borrow<'a>(ptr: *mut c_void) -> ArcBorrow<'a, T> { // SAFETY: The safety requirements of this function ensure that `ptr` comes from a previous // call to `Self::into_foreign`. let inner = unsafe { NonNull::new_unchecked(ptr.cast::<ArcInner<T>>()) }; @@ -371,10 +394,10 @@ impl<T: 'static> ForeignOwnable for Arc<T> { unsafe { ArcBorrow::new(inner) } } - unsafe fn borrow_mut<'a>(ptr: *mut crate::ffi::c_void) -> ArcBorrow<'a, T> { + unsafe fn borrow_mut<'a>(ptr: *mut c_void) -> ArcBorrow<'a, T> { // SAFETY: The safety requirements for `borrow_mut` are a superset of the safety // requirements for `borrow`. - unsafe { Self::borrow(ptr) } + unsafe { <Self as ForeignOwnable>::borrow(ptr) } } } @@ -394,16 +417,37 @@ impl<T: ?Sized> AsRef<T> for Arc<T> { } } +/// # Examples +/// +/// ``` +/// # use core::borrow::Borrow; +/// # use kernel::sync::Arc; +/// struct Foo<B: Borrow<u32>>(B); +/// +/// // Owned instance. +/// let owned = Foo(1); +/// +/// // Shared instance. +/// let arc = Arc::new(1, GFP_KERNEL)?; +/// let shared = Foo(arc.clone()); +/// +/// let i = 1; +/// // Borrowed from `i`. +/// let borrowed = Foo(&i); +/// # Ok::<(), Error>(()) +/// ``` +impl<T: ?Sized> Borrow<T> for Arc<T> { + fn borrow(&self) -> &T { + self.deref() + } +} + impl<T: ?Sized> Clone for Arc<T> { fn clone(&self) -> Self { - // SAFETY: By the type invariant, there is necessarily a reference to the object, so it is - // safe to dereference it. - let refcount = unsafe { self.ptr.as_ref() }.refcount.get(); - - // INVARIANT: C `refcount_inc` saturates the refcount, so it cannot overflow to zero. + // INVARIANT: `Refcount` saturates the refcount, so it cannot overflow to zero. // SAFETY: By the type invariant, there is necessarily a reference to the object, so it is // safe to increment the refcount. - unsafe { bindings::refcount_inc(refcount) }; + unsafe { self.ptr.as_ref() }.refcount.inc(); // SAFETY: We just incremented the refcount. This increment is now owned by the new `Arc`. unsafe { Self::from_inner(self.ptr) } @@ -412,16 +456,10 @@ impl<T: ?Sized> Clone for Arc<T> { impl<T: ?Sized> Drop for Arc<T> { fn drop(&mut self) { - // SAFETY: By the type invariant, there is necessarily a reference to the object. We cannot - // touch `refcount` after it's decremented to a non-zero value because another thread/CPU - // may concurrently decrement it to zero and free it. It is ok to have a raw pointer to - // freed/invalid memory as long as it is never dereferenced. - let refcount = unsafe { self.ptr.as_ref() }.refcount.get(); - // INVARIANT: If the refcount reaches zero, there are no other instances of `Arc`, and // this instance is being dropped, so the broken invariant is not observable. - // SAFETY: Also by the type invariant, we are allowed to decrement the refcount. - let is_zero = unsafe { bindings::refcount_dec_and_test(refcount) }; + // SAFETY: By the type invariant, there is necessarily a reference to the object. + let is_zero = unsafe { self.ptr.as_ref() }.refcount.dec_and_test(); if is_zero { // The count reached zero, we must free the memory. // @@ -460,7 +498,7 @@ impl<T: ?Sized> From<Pin<UniqueArc<T>>> for Arc<T> { /// There are no mutable references to the underlying [`Arc`], and it remains valid for the /// lifetime of the [`ArcBorrow`] instance. /// -/// # Example +/// # Examples /// /// ``` /// use kernel::sync::{Arc, ArcBorrow}; @@ -500,20 +538,12 @@ impl<T: ?Sized> From<Pin<UniqueArc<T>>> for Arc<T> { /// # Ok::<(), Error>(()) /// ``` #[repr(transparent)] -#[cfg_attr(CONFIG_RUSTC_HAS_COERCE_POINTEE, derive(core::marker::CoercePointee))] +#[derive(core::marker::CoercePointee)] pub struct ArcBorrow<'a, T: ?Sized + 'a> { inner: NonNull<ArcInner<T>>, _p: PhantomData<&'a ()>, } -// This is to allow `ArcBorrow<U>` to be dispatched on when `ArcBorrow<T>` can be coerced into -// `ArcBorrow<U>`. -#[cfg(not(CONFIG_RUSTC_HAS_COERCE_POINTEE))] -impl<T: ?Sized + core::marker::Unsize<U>, U: ?Sized> core::ops::DispatchFromDyn<ArcBorrow<'_, U>> - for ArcBorrow<'_, T> -{ -} - impl<T: ?Sized> Clone for ArcBorrow<'_, T> { fn clone(&self) -> Self { *self @@ -539,11 +569,11 @@ impl<T: ?Sized> ArcBorrow<'_, T> { } /// Creates an [`ArcBorrow`] to an [`Arc`] that has previously been deconstructed with - /// [`Arc::into_raw`]. + /// [`Arc::into_raw`] or [`Arc::as_ptr`]. /// /// # Safety /// - /// * The provided pointer must originate from a call to [`Arc::into_raw`]. + /// * The provided pointer must originate from a call to [`Arc::into_raw`] or [`Arc::as_ptr`]. /// * For the duration of the lifetime annotated on this `ArcBorrow`, the reference count must /// not hit zero. /// * For the duration of the lifetime annotated on this `ArcBorrow`, there must not be a @@ -659,6 +689,50 @@ pub struct UniqueArc<T: ?Sized> { inner: Arc<T>, } +impl<T> InPlaceInit<T> for UniqueArc<T> { + type PinnedSelf = Pin<Self>; + + #[inline] + fn try_pin_init<E>(init: impl PinInit<T, E>, flags: Flags) -> Result<Self::PinnedSelf, E> + where + E: From<AllocError>, + { + UniqueArc::new_uninit(flags)?.write_pin_init(init) + } + + #[inline] + fn try_init<E>(init: impl Init<T, E>, flags: Flags) -> Result<Self, E> + where + E: From<AllocError>, + { + UniqueArc::new_uninit(flags)?.write_init(init) + } +} + +impl<T> InPlaceWrite<T> for UniqueArc<MaybeUninit<T>> { + type Initialized = UniqueArc<T>; + + #[inline] + fn write_init<E>(mut self, init: impl Init<T, E>) -> Result<Self::Initialized, E> { + let slot = self.as_mut_ptr(); + // SAFETY: When init errors/panics, slot will get deallocated but not dropped, + // slot is valid. + unsafe { init.__init(slot)? }; + // SAFETY: All fields have been initialized. + Ok(unsafe { self.assume_init() }) + } + + #[inline] + fn write_pin_init<E>(mut self, init: impl PinInit<T, E>) -> Result<Pin<Self::Initialized>, E> { + let slot = self.as_mut_ptr(); + // SAFETY: When init errors/panics, slot will get deallocated but not dropped, + // slot is valid and will not be moved, because we pin it later. + unsafe { init.__pinned_init(slot)? }; + // SAFETY: All fields have been initialized. + Ok(unsafe { self.assume_init() }.into()) + } +} + impl<T> UniqueArc<T> { /// Tries to allocate a new [`UniqueArc`] instance. pub fn new(value: T, flags: Flags) -> Result<Self, AllocError> { @@ -673,9 +747,8 @@ impl<T> UniqueArc<T> { // INVARIANT: The refcount is initialised to a non-zero value. let inner = KBox::try_init::<AllocError>( try_init!(ArcInner { - // SAFETY: There are no safety requirements for this FFI call. - refcount: Opaque::new(unsafe { bindings::REFCOUNT_INIT(1) }), - data <- init::uninit::<T, AllocError>(), + refcount: Refcount::new(1), + data <- pin_init::uninit::<T, AllocError>(), }? AllocError), flags, )?; @@ -687,6 +760,14 @@ impl<T> UniqueArc<T> { } } +impl<T: ?Sized> UniqueArc<T> { + /// Return a raw pointer to the data in this [`UniqueArc`]. + #[inline] + pub fn as_ptr(this: &Self) -> *const T { + Arc::as_ptr(&this.inner) + } +} + impl<T> UniqueArc<MaybeUninit<T>> { /// Converts a `UniqueArc<MaybeUninit<T>>` into a `UniqueArc<T>` by writing a value into it. pub fn write(mut self, value: T) -> UniqueArc<T> { @@ -711,6 +792,7 @@ impl<T> UniqueArc<MaybeUninit<T>> { } /// Initialize `self` using the given initializer. + #[inline] pub fn init_with<E>(mut self, init: impl Init<T, E>) -> core::result::Result<UniqueArc<T>, E> { // SAFETY: The supplied pointer is valid for initialization. match unsafe { init.__init(self.as_mut_ptr()) } { @@ -721,6 +803,7 @@ impl<T> UniqueArc<MaybeUninit<T>> { } /// Pin-initialize `self` using the given pin-initializer. + #[inline] pub fn pin_init_with<E>( mut self, init: impl PinInit<T, E>, @@ -760,6 +843,56 @@ impl<T: ?Sized> DerefMut for UniqueArc<T> { } } +/// # Examples +/// +/// ``` +/// # use core::borrow::Borrow; +/// # use kernel::sync::UniqueArc; +/// struct Foo<B: Borrow<u32>>(B); +/// +/// // Owned instance. +/// let owned = Foo(1); +/// +/// // Owned instance using `UniqueArc`. +/// let arc = UniqueArc::new(1, GFP_KERNEL)?; +/// let shared = Foo(arc); +/// +/// let i = 1; +/// // Borrowed from `i`. +/// let borrowed = Foo(&i); +/// # Ok::<(), Error>(()) +/// ``` +impl<T: ?Sized> Borrow<T> for UniqueArc<T> { + fn borrow(&self) -> &T { + self.deref() + } +} + +/// # Examples +/// +/// ``` +/// # use core::borrow::BorrowMut; +/// # use kernel::sync::UniqueArc; +/// struct Foo<B: BorrowMut<u32>>(B); +/// +/// // Owned instance. +/// let owned = Foo(1); +/// +/// // Owned instance using `UniqueArc`. +/// let arc = UniqueArc::new(1, GFP_KERNEL)?; +/// let shared = Foo(arc); +/// +/// let mut i = 1; +/// // Borrowed from `i`. +/// let borrowed = Foo(&mut i); +/// # Ok::<(), Error>(()) +/// ``` +impl<T: ?Sized> BorrowMut<T> for UniqueArc<T> { + fn borrow_mut(&mut self) -> &mut T { + self.deref_mut() + } +} + impl<T: fmt::Display + ?Sized> fmt::Display for UniqueArc<T> { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { fmt::Display::fmt(self.deref(), f) diff --git a/rust/kernel/sync/aref.rs b/rust/kernel/sync/aref.rs new file mode 100644 index 000000000000..b721b2e00b98 --- /dev/null +++ b/rust/kernel/sync/aref.rs @@ -0,0 +1,199 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Internal reference counting support. +//! +//! Many C types already have their own reference counting mechanism (e.g. by storing a +//! `refcount_t`). This module provides support for directly using their internal reference count +//! from Rust; instead of making users have to use an additional Rust-reference count in the form of +//! [`Arc`]. +//! +//! The smart pointer [`ARef<T>`] acts similarly to [`Arc<T>`] in that it holds a refcount on the +//! underlying object, but this refcount is internal to the object. It essentially is a Rust +//! implementation of the `get_` and `put_` pattern used in C for reference counting. +//! +//! To make use of [`ARef<MyType>`], `MyType` needs to implement [`AlwaysRefCounted`]. It is a trait +//! for accessing the internal reference count of an object of the `MyType` type. +//! +//! [`Arc`]: crate::sync::Arc +//! [`Arc<T>`]: crate::sync::Arc + +use core::{ + marker::PhantomData, + mem::ManuallyDrop, + ops::Deref, + ptr::NonNull, // +}; + +/// Types that are _always_ reference counted. +/// +/// It allows such types to define their own custom ref increment and decrement functions. +/// Additionally, it allows users to convert from a shared reference `&T` to an owned reference +/// [`ARef<T>`]. +/// +/// This is usually implemented by wrappers to existing structures on the C side of the code. For +/// Rust code, the recommendation is to use [`Arc`](crate::sync::Arc) to create reference-counted +/// instances of a type. +/// +/// # Safety +/// +/// Implementers must ensure that increments to the reference count keep the object alive in memory +/// at least until matching decrements are performed. +/// +/// Implementers must also ensure that all instances are reference-counted. (Otherwise they +/// won't be able to honour the requirement that [`AlwaysRefCounted::inc_ref`] keep the object +/// alive.) +pub unsafe trait AlwaysRefCounted { + /// Increments the reference count on the object. + fn inc_ref(&self); + + /// Decrements the reference count on the object. + /// + /// Frees the object when the count reaches zero. + /// + /// # Safety + /// + /// Callers must ensure that there was a previous matching increment to the reference count, + /// and that the object is no longer used after its reference count is decremented (as it may + /// result in the object being freed), unless the caller owns another increment on the refcount + /// (e.g., it calls [`AlwaysRefCounted::inc_ref`] twice, then calls + /// [`AlwaysRefCounted::dec_ref`] once). + unsafe fn dec_ref(obj: NonNull<Self>); +} + +/// An owned reference to an always-reference-counted object. +/// +/// The object's reference count is automatically decremented when an instance of [`ARef`] is +/// dropped. It is also automatically incremented when a new instance is created via +/// [`ARef::clone`]. +/// +/// # Invariants +/// +/// The pointer stored in `ptr` is non-null and valid for the lifetime of the [`ARef`] instance. In +/// particular, the [`ARef`] instance owns an increment on the underlying object's reference count. +pub struct ARef<T: AlwaysRefCounted> { + ptr: NonNull<T>, + _p: PhantomData<T>, +} + +// SAFETY: It is safe to send `ARef<T>` to another thread when the underlying `T` is `Sync` because +// it effectively means sharing `&T` (which is safe because `T` is `Sync`); additionally, it needs +// `T` to be `Send` because any thread that has an `ARef<T>` may ultimately access `T` using a +// mutable reference, for example, when the reference count reaches zero and `T` is dropped. +unsafe impl<T: AlwaysRefCounted + Sync + Send> Send for ARef<T> {} + +// SAFETY: It is safe to send `&ARef<T>` to another thread when the underlying `T` is `Sync` +// because it effectively means sharing `&T` (which is safe because `T` is `Sync`); additionally, +// it needs `T` to be `Send` because any thread that has a `&ARef<T>` may clone it and get an +// `ARef<T>` on that thread, so the thread may ultimately access `T` using a mutable reference, for +// example, when the reference count reaches zero and `T` is dropped. +unsafe impl<T: AlwaysRefCounted + Sync + Send> Sync for ARef<T> {} + +// Even if `T` is pinned, pointers to `T` can still move. +impl<T: AlwaysRefCounted> Unpin for ARef<T> {} + +impl<T: AlwaysRefCounted> ARef<T> { + /// Creates a new instance of [`ARef`]. + /// + /// It takes over an increment of the reference count on the underlying object. + /// + /// # Safety + /// + /// Callers must ensure that the reference count was incremented at least once, and that they + /// are properly relinquishing one increment. That is, if there is only one increment, callers + /// must not use the underlying object anymore -- it is only safe to do so via the newly + /// created [`ARef`]. + pub unsafe fn from_raw(ptr: NonNull<T>) -> Self { + // INVARIANT: The safety requirements guarantee that the new instance now owns the + // increment on the refcount. + Self { + ptr, + _p: PhantomData, + } + } + + /// Consumes the `ARef`, returning a raw pointer. + /// + /// This function does not change the refcount. After calling this function, the caller is + /// responsible for the refcount previously managed by the `ARef`. + /// + /// # Examples + /// + /// ``` + /// use core::ptr::NonNull; + /// use kernel::sync::aref::{ARef, AlwaysRefCounted}; + /// + /// struct Empty {} + /// + /// # // SAFETY: TODO. + /// unsafe impl AlwaysRefCounted for Empty { + /// fn inc_ref(&self) {} + /// unsafe fn dec_ref(_obj: NonNull<Self>) {} + /// } + /// + /// let mut data = Empty {}; + /// let ptr = NonNull::<Empty>::new(&mut data).unwrap(); + /// # // SAFETY: TODO. + /// let data_ref: ARef<Empty> = unsafe { ARef::from_raw(ptr) }; + /// let raw_ptr: NonNull<Empty> = ARef::into_raw(data_ref); + /// + /// assert_eq!(ptr, raw_ptr); + /// ``` + pub fn into_raw(me: Self) -> NonNull<T> { + ManuallyDrop::new(me).ptr + } +} + +impl<T: AlwaysRefCounted> Clone for ARef<T> { + fn clone(&self) -> Self { + self.inc_ref(); + // SAFETY: We just incremented the refcount above. + unsafe { Self::from_raw(self.ptr) } + } +} + +impl<T: AlwaysRefCounted> Deref for ARef<T> { + type Target = T; + + fn deref(&self) -> &Self::Target { + // SAFETY: The type invariants guarantee that the object is valid. + unsafe { self.ptr.as_ref() } + } +} + +impl<T: AlwaysRefCounted> From<&T> for ARef<T> { + fn from(b: &T) -> Self { + b.inc_ref(); + // SAFETY: We just incremented the refcount above. + unsafe { Self::from_raw(NonNull::from(b)) } + } +} + +impl<T: AlwaysRefCounted> Drop for ARef<T> { + fn drop(&mut self) { + // SAFETY: The type invariants guarantee that the `ARef` owns the reference we're about to + // decrement. + unsafe { T::dec_ref(self.ptr) }; + } +} + +impl<T, U> PartialEq<ARef<U>> for ARef<T> +where + T: AlwaysRefCounted + PartialEq<U>, + U: AlwaysRefCounted, +{ + #[inline] + fn eq(&self, other: &ARef<U>) -> bool { + T::eq(&**self, &**other) + } +} +impl<T: AlwaysRefCounted + Eq> Eq for ARef<T> {} + +impl<T, U> PartialEq<&'_ U> for ARef<T> +where + T: AlwaysRefCounted + PartialEq<U>, +{ + #[inline] + fn eq(&self, other: &&U) -> bool { + T::eq(&**self, other) + } +} diff --git a/rust/kernel/sync/atomic.rs b/rust/kernel/sync/atomic.rs new file mode 100644 index 000000000000..9cd009d57e35 --- /dev/null +++ b/rust/kernel/sync/atomic.rs @@ -0,0 +1,850 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Atomic primitives. +//! +//! These primitives have the same semantics as their C counterparts: and the precise definitions of +//! semantics can be found at [`LKMM`]. Note that Linux Kernel Memory (Consistency) Model is the +//! only model for Rust code in kernel, and Rust's own atomics should be avoided. +//! +//! # Data races +//! +//! [`LKMM`] atomics have different rules regarding data races: +//! +//! - A normal write from C side is treated as an atomic write if +//! CONFIG_KCSAN_ASSUME_PLAIN_WRITES_ATOMIC=y. +//! - Mixed-size atomic accesses don't cause data races. +//! +//! [`LKMM`]: srctree/tools/memory-model/ + +mod internal; +pub mod ordering; +mod predefine; + +pub use internal::AtomicImpl; +pub use ordering::{Acquire, Full, Relaxed, Release}; + +pub(crate) use internal::{AtomicArithmeticOps, AtomicBasicOps, AtomicExchangeOps}; + +use crate::build_error; +use internal::AtomicRepr; +use ordering::OrderingType; + +/// A memory location which can be safely modified from multiple execution contexts. +/// +/// This has the same size, alignment and bit validity as the underlying type `T`. And it disables +/// niche optimization for the same reason as [`UnsafeCell`]. +/// +/// The atomic operations are implemented in a way that is fully compatible with the [Linux Kernel +/// Memory (Consistency) Model][LKMM], hence they should be modeled as the corresponding +/// [`LKMM`][LKMM] atomic primitives. With the help of [`Atomic::from_ptr()`] and +/// [`Atomic::as_ptr()`], this provides a way to interact with [C-side atomic operations] +/// (including those without the `atomic` prefix, e.g. `READ_ONCE()`, `WRITE_ONCE()`, +/// `smp_load_acquire()` and `smp_store_release()`). +/// +/// # Invariants +/// +/// `self.0` is a valid `T`. +/// +/// [`UnsafeCell`]: core::cell::UnsafeCell +/// [LKMM]: srctree/tools/memory-model/ +/// [C-side atomic operations]: srctree/Documentation/atomic_t.txt +#[repr(transparent)] +pub struct Atomic<T: AtomicType>(AtomicRepr<T::Repr>); + +// SAFETY: `Atomic<T>` is safe to transfer between execution contexts because of the safety +// requirement of `AtomicType`. +unsafe impl<T: AtomicType> Send for Atomic<T> {} + +// SAFETY: `Atomic<T>` is safe to share among execution contexts because all accesses are atomic. +unsafe impl<T: AtomicType> Sync for Atomic<T> {} + +/// Types that support basic atomic operations. +/// +/// # Round-trip transmutability +/// +/// `T` is round-trip transmutable to `U` if and only if both of these properties hold: +/// +/// - Any valid bit pattern for `T` is also a valid bit pattern for `U`. +/// - Transmuting (e.g. using [`transmute()`]) a value of type `T` to `U` and then to `T` again +/// yields a value that is in all aspects equivalent to the original value. +/// +/// # Safety +/// +/// - [`Self`] must have the same size and alignment as [`Self::Repr`]. +/// - [`Self`] must be [round-trip transmutable] to [`Self::Repr`]. +/// - [`Self`] must be safe to transfer between execution contexts, if it's [`Send`], this is +/// automatically satisfied. The exception is pointer types that are even though marked as +/// `!Send` (e.g. raw pointers and [`NonNull<T>`]) but requiring `unsafe` to do anything +/// meaningful on them. This is because transferring pointer values between execution contexts is +/// safe as long as the actual `unsafe` dereferencing is justified. +/// +/// Note that this is more relaxed than requiring the bi-directional transmutability (i.e. +/// [`transmute()`] is always sound between `U` and `T`) because of the support for atomic +/// variables over unit-only enums, see [Examples]. +/// +/// # Limitations +/// +/// Because C primitives are used to implement the atomic operations, and a C function requires a +/// valid object of a type to operate on (i.e. no `MaybeUninit<_>`), hence at the Rust <-> C +/// surface, only types with all the bits initialized can be passed. As a result, types like `(u8, +/// u16)` (padding bytes are uninitialized) are currently not supported. +/// +/// # Examples +/// +/// A unit-only enum that implements [`AtomicType`]: +/// +/// ``` +/// use kernel::sync::atomic::{AtomicType, Atomic, Relaxed}; +/// +/// #[derive(Clone, Copy, PartialEq, Eq)] +/// #[repr(i32)] +/// enum State { +/// Uninit = 0, +/// Working = 1, +/// Done = 2, +/// }; +/// +/// // SAFETY: `State` and `i32` has the same size and alignment, and it's round-trip +/// // transmutable to `i32`. +/// unsafe impl AtomicType for State { +/// type Repr = i32; +/// } +/// +/// let s = Atomic::new(State::Uninit); +/// +/// assert_eq!(State::Uninit, s.load(Relaxed)); +/// ``` +/// [`transmute()`]: core::mem::transmute +/// [round-trip transmutable]: AtomicType#round-trip-transmutability +/// [Examples]: AtomicType#examples +/// [`NonNull<T>`]: core::ptr::NonNull +pub unsafe trait AtomicType: Sized + Copy { + /// The backing atomic implementation type. + type Repr: AtomicImpl; +} + +/// Types that support atomic add operations. +/// +/// # Safety +/// +// TODO: Properly defines `wrapping_add` in the following comment. +/// `wrapping_add` any value of type `Self::Repr::Delta` obtained by [`Self::rhs_into_delta()`] to +/// any value of type `Self::Repr` obtained through transmuting a value of type `Self` to must +/// yield a value with a bit pattern also valid for `Self`. +pub unsafe trait AtomicAdd<Rhs = Self>: AtomicType { + /// Converts `Rhs` into the `Delta` type of the atomic implementation. + fn rhs_into_delta(rhs: Rhs) -> <Self::Repr as AtomicImpl>::Delta; +} + +#[inline(always)] +const fn into_repr<T: AtomicType>(v: T) -> T::Repr { + // SAFETY: Per the safety requirement of `AtomicType`, `T` is round-trip transmutable to + // `T::Repr`, therefore the transmute operation is sound. + unsafe { core::mem::transmute_copy(&v) } +} + +/// # Safety +/// +/// `r` must be a valid bit pattern of `T`. +#[inline(always)] +const unsafe fn from_repr<T: AtomicType>(r: T::Repr) -> T { + // SAFETY: Per the safety requirement of the function, the transmute operation is sound. + unsafe { core::mem::transmute_copy(&r) } +} + +impl<T: AtomicType> Atomic<T> { + /// Creates a new atomic `T`. + pub const fn new(v: T) -> Self { + // INVARIANT: Per the safety requirement of `AtomicType`, `into_repr(v)` is a valid `T`. + Self(AtomicRepr::new(into_repr(v))) + } + + /// Creates a reference to an atomic `T` from a pointer of `T`. + /// + /// This usually is used when communicating with C side or manipulating a C struct, see + /// examples below. + /// + /// # Safety + /// + /// - `ptr` is aligned to `align_of::<T>()`. + /// - `ptr` is valid for reads and writes for `'a`. + /// - For the duration of `'a`, other accesses to `*ptr` must not cause data races (defined + /// by [`LKMM`]) against atomic operations on the returned reference. Note that if all other + /// accesses are atomic, then this safety requirement is trivially fulfilled. + /// + /// [`LKMM`]: srctree/tools/memory-model + /// + /// # Examples + /// + /// Using [`Atomic::from_ptr()`] combined with [`Atomic::load()`] or [`Atomic::store()`] can + /// achieve the same functionality as `READ_ONCE()`/`smp_load_acquire()` or + /// `WRITE_ONCE()`/`smp_store_release()` in C side: + /// + /// ``` + /// # use kernel::types::Opaque; + /// use kernel::sync::atomic::{Atomic, Relaxed, Release}; + /// + /// // Assume there is a C struct `foo`. + /// mod cbindings { + /// #[repr(C)] + /// pub(crate) struct foo { + /// pub(crate) a: i32, + /// pub(crate) b: i32 + /// } + /// } + /// + /// let tmp = Opaque::new(cbindings::foo { a: 1, b: 2 }); + /// + /// // struct foo *foo_ptr = ..; + /// let foo_ptr = tmp.get(); + /// + /// // SAFETY: `foo_ptr` is valid, and `.a` is in bounds. + /// let foo_a_ptr = unsafe { &raw mut (*foo_ptr).a }; + /// + /// // a = READ_ONCE(foo_ptr->a); + /// // + /// // SAFETY: `foo_a_ptr` is valid for read, and all other accesses on it is atomic, so no + /// // data race. + /// let a = unsafe { Atomic::from_ptr(foo_a_ptr) }.load(Relaxed); + /// # assert_eq!(a, 1); + /// + /// // smp_store_release(&foo_ptr->a, 2); + /// // + /// // SAFETY: `foo_a_ptr` is valid for writes, and all other accesses on it is atomic, so + /// // no data race. + /// unsafe { Atomic::from_ptr(foo_a_ptr) }.store(2, Release); + /// ``` + pub unsafe fn from_ptr<'a>(ptr: *mut T) -> &'a Self { + // CAST: `T` and `Atomic<T>` have the same size, alignment and bit validity. + // SAFETY: Per function safety requirement, `ptr` is a valid pointer and the object will + // live long enough. It's safe to return a `&Atomic<T>` because function safety requirement + // guarantees other accesses won't cause data races. + unsafe { &*ptr.cast::<Self>() } + } + + /// Returns a pointer to the underlying atomic `T`. + /// + /// Note that use of the return pointer must not cause data races defined by [`LKMM`]. + /// + /// # Guarantees + /// + /// The returned pointer is valid and properly aligned (i.e. aligned to [`align_of::<T>()`]). + /// + /// [`LKMM`]: srctree/tools/memory-model + /// [`align_of::<T>()`]: core::mem::align_of + pub const fn as_ptr(&self) -> *mut T { + // GUARANTEE: Per the function guarantee of `AtomicRepr::as_ptr()`, the `self.0.as_ptr()` + // must be a valid and properly aligned pointer for `T::Repr`, and per the safety guarantee + // of `AtomicType`, it's a valid and properly aligned pointer of `T`. + self.0.as_ptr().cast() + } + + /// Returns a mutable reference to the underlying atomic `T`. + /// + /// This is safe because the mutable reference of the atomic `T` guarantees exclusive access. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Relaxed}; + /// + /// let mut atomic_val = Atomic::new(0u32); + /// let val_mut = atomic_val.get_mut(); + /// *val_mut = 101; + /// assert_eq!(101, atomic_val.load(Relaxed)); + /// ``` + pub fn get_mut(&mut self) -> &mut T { + // CAST: `T` and `T::Repr` has the same size and alignment per the safety requirement of + // `AtomicType`, and per the type invariants `self.0` is a valid `T`, therefore the casting + // result is a valid pointer of `T`. + // SAFETY: The pointer is valid per the CAST comment above, and the mutable reference + // guarantees exclusive access. + unsafe { &mut *self.0.as_ptr().cast() } + } +} + +impl<T: AtomicType> Atomic<T> +where + T::Repr: AtomicBasicOps, +{ + /// Loads the value from the atomic `T`. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Relaxed}; + /// + /// let x = Atomic::new(42i32); + /// + /// assert_eq!(42, x.load(Relaxed)); + /// + /// let x = Atomic::new(42i64); + /// + /// assert_eq!(42, x.load(Relaxed)); + /// ``` + #[doc(alias("atomic_read", "atomic64_read"))] + #[inline(always)] + pub fn load<Ordering: ordering::AcquireOrRelaxed>(&self, _: Ordering) -> T { + let v = { + match Ordering::TYPE { + OrderingType::Relaxed => T::Repr::atomic_read(&self.0), + OrderingType::Acquire => T::Repr::atomic_read_acquire(&self.0), + _ => build_error!("Wrong ordering"), + } + }; + + // SAFETY: `v` comes from reading `self.0`, which is a valid `T` per the type invariants. + unsafe { from_repr(v) } + } + + /// Stores a value to the atomic `T`. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Relaxed}; + /// + /// let x = Atomic::new(42i32); + /// + /// assert_eq!(42, x.load(Relaxed)); + /// + /// x.store(43, Relaxed); + /// + /// assert_eq!(43, x.load(Relaxed)); + /// ``` + #[doc(alias("atomic_set", "atomic64_set"))] + #[inline(always)] + pub fn store<Ordering: ordering::ReleaseOrRelaxed>(&self, v: T, _: Ordering) { + let v = into_repr(v); + + // INVARIANT: `v` is a valid `T`, and is stored to `self.0` by `atomic_set*()`. + match Ordering::TYPE { + OrderingType::Relaxed => T::Repr::atomic_set(&self.0, v), + OrderingType::Release => T::Repr::atomic_set_release(&self.0, v), + _ => build_error!("Wrong ordering"), + } + } +} + +impl<T: AtomicType + core::fmt::Debug> core::fmt::Debug for Atomic<T> +where + T::Repr: AtomicBasicOps, +{ + fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result { + core::fmt::Debug::fmt(&self.load(Relaxed), f) + } +} + +impl<T: AtomicType> Atomic<T> +where + T::Repr: AtomicExchangeOps, +{ + /// Atomic exchange. + /// + /// Atomically updates `*self` to `v` and returns the old value of `*self`. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Acquire, Relaxed}; + /// + /// let x = Atomic::new(42); + /// + /// assert_eq!(42, x.xchg(52, Acquire)); + /// assert_eq!(52, x.load(Relaxed)); + /// ``` + #[doc(alias("atomic_xchg", "atomic64_xchg", "swap"))] + #[inline(always)] + pub fn xchg<Ordering: ordering::Ordering>(&self, v: T, _: Ordering) -> T { + let v = into_repr(v); + + // INVARIANT: `self.0` is a valid `T` after `atomic_xchg*()` because `v` is transmutable to + // `T`. + let ret = { + match Ordering::TYPE { + OrderingType::Full => T::Repr::atomic_xchg(&self.0, v), + OrderingType::Acquire => T::Repr::atomic_xchg_acquire(&self.0, v), + OrderingType::Release => T::Repr::atomic_xchg_release(&self.0, v), + OrderingType::Relaxed => T::Repr::atomic_xchg_relaxed(&self.0, v), + } + }; + + // SAFETY: `ret` comes from reading `*self`, which is a valid `T` per type invariants. + unsafe { from_repr(ret) } + } + + /// Atomic compare and exchange. + /// + /// If `*self` == `old`, atomically updates `*self` to `new`. Otherwise, `*self` is not + /// modified. + /// + /// Compare: The comparison is done via the byte level comparison between `*self` and `old`. + /// + /// Ordering: When succeeds, provides the corresponding ordering as the `Ordering` type + /// parameter indicates, and a failed one doesn't provide any ordering, the load part of a + /// failed cmpxchg is a [`Relaxed`] load. + /// + /// Returns `Ok(value)` if cmpxchg succeeds, and `value` is guaranteed to be equal to `old`, + /// otherwise returns `Err(value)`, and `value` is the current value of `*self`. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Full, Relaxed}; + /// + /// let x = Atomic::new(42); + /// + /// // Checks whether cmpxchg succeeded. + /// let success = x.cmpxchg(52, 64, Relaxed).is_ok(); + /// # assert!(!success); + /// + /// // Checks whether cmpxchg failed. + /// let failure = x.cmpxchg(52, 64, Relaxed).is_err(); + /// # assert!(failure); + /// + /// // Uses the old value if failed, probably re-try cmpxchg. + /// match x.cmpxchg(52, 64, Relaxed) { + /// Ok(_) => { }, + /// Err(old) => { + /// // do something with `old`. + /// # assert_eq!(old, 42); + /// } + /// } + /// + /// // Uses the latest value regardlessly, same as atomic_cmpxchg() in C. + /// let latest = x.cmpxchg(42, 64, Full).unwrap_or_else(|old| old); + /// # assert_eq!(42, latest); + /// assert_eq!(64, x.load(Relaxed)); + /// ``` + /// + /// [`Relaxed`]: ordering::Relaxed + #[doc(alias( + "atomic_cmpxchg", + "atomic64_cmpxchg", + "atomic_try_cmpxchg", + "atomic64_try_cmpxchg", + "compare_exchange" + ))] + #[inline(always)] + pub fn cmpxchg<Ordering: ordering::Ordering>( + &self, + mut old: T, + new: T, + o: Ordering, + ) -> Result<T, T> { + // Note on code generation: + // + // try_cmpxchg() is used to implement cmpxchg(), and if the helper functions are inlined, + // the compiler is able to figure out that branch is not needed if the users don't care + // about whether the operation succeeds or not. One exception is on x86, due to commit + // 44fe84459faf ("locking/atomic: Fix atomic_try_cmpxchg() semantics"), the + // atomic_try_cmpxchg() on x86 has a branch even if the caller doesn't care about the + // success of cmpxchg and only wants to use the old value. For example, for code like: + // + // let latest = x.cmpxchg(42, 64, Full).unwrap_or_else(|old| old); + // + // It will still generate code: + // + // movl $0x40, %ecx + // movl $0x34, %eax + // lock + // cmpxchgl %ecx, 0x4(%rsp) + // jne 1f + // 2: + // ... + // 1: movl %eax, %ecx + // jmp 2b + // + // This might be "fixed" by introducing a try_cmpxchg_exclusive() that knows the "*old" + // location in the C function is always safe to write. + if self.try_cmpxchg(&mut old, new, o) { + Ok(old) + } else { + Err(old) + } + } + + /// Atomic compare and exchange and returns whether the operation succeeds. + /// + /// If `*self` == `old`, atomically updates `*self` to `new`. Otherwise, `*self` is not + /// modified, `*old` is updated to the current value of `*self`. + /// + /// "Compare" and "Ordering" part are the same as [`Atomic::cmpxchg()`]. + /// + /// Returns `true` means the cmpxchg succeeds otherwise returns `false`. + #[inline(always)] + fn try_cmpxchg<Ordering: ordering::Ordering>(&self, old: &mut T, new: T, _: Ordering) -> bool { + let mut tmp = into_repr(*old); + let new = into_repr(new); + + // INVARIANT: `self.0` is a valid `T` after `atomic_try_cmpxchg*()` because `new` is + // transmutable to `T`. + let ret = { + match Ordering::TYPE { + OrderingType::Full => T::Repr::atomic_try_cmpxchg(&self.0, &mut tmp, new), + OrderingType::Acquire => { + T::Repr::atomic_try_cmpxchg_acquire(&self.0, &mut tmp, new) + } + OrderingType::Release => { + T::Repr::atomic_try_cmpxchg_release(&self.0, &mut tmp, new) + } + OrderingType::Relaxed => { + T::Repr::atomic_try_cmpxchg_relaxed(&self.0, &mut tmp, new) + } + } + }; + + // SAFETY: `tmp` comes from reading `*self`, which is a valid `T` per type invariants. + *old = unsafe { from_repr(tmp) }; + + ret + } +} + +impl<T: AtomicType> Atomic<T> +where + T::Repr: AtomicArithmeticOps, +{ + /// Atomic add. + /// + /// Atomically updates `*self` to `(*self).wrapping_add(v)`. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Relaxed}; + /// + /// let x = Atomic::new(42); + /// + /// assert_eq!(42, x.load(Relaxed)); + /// + /// x.add(12, Relaxed); + /// + /// assert_eq!(54, x.load(Relaxed)); + /// ``` + #[inline(always)] + pub fn add<Rhs>(&self, v: Rhs, _: ordering::Relaxed) + where + T: AtomicAdd<Rhs>, + { + let v = T::rhs_into_delta(v); + + // INVARIANT: `self.0` is a valid `T` after `atomic_add()` due to safety requirement of + // `AtomicAdd`. + T::Repr::atomic_add(&self.0, v); + } + + /// Atomic fetch and add. + /// + /// Atomically updates `*self` to `(*self).wrapping_add(v)`, and returns the value of `*self` + /// before the update. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Acquire, Full, Relaxed}; + /// + /// let x = Atomic::new(42); + /// assert_eq!(42, x.load(Relaxed)); + /// assert_eq!(42, x.fetch_add(12, Acquire)); + /// assert_eq!(54, x.load(Relaxed)); + /// + /// let x = Atomic::new(42); + /// assert_eq!(42, x.load(Relaxed)); + /// assert_eq!(42, x.fetch_add(12, Full)); + /// assert_eq!(54, x.load(Relaxed)); + /// ``` + #[inline(always)] + pub fn fetch_add<Rhs, Ordering: ordering::Ordering>(&self, v: Rhs, _: Ordering) -> T + where + T: AtomicAdd<Rhs>, + { + let v = T::rhs_into_delta(v); + + // INVARIANT: `self.0` is a valid `T` after `atomic_fetch_add*()` due to safety requirement + // of `AtomicAdd`. + let ret = { + match Ordering::TYPE { + OrderingType::Full => T::Repr::atomic_fetch_add(&self.0, v), + OrderingType::Acquire => T::Repr::atomic_fetch_add_acquire(&self.0, v), + OrderingType::Release => T::Repr::atomic_fetch_add_release(&self.0, v), + OrderingType::Relaxed => T::Repr::atomic_fetch_add_relaxed(&self.0, v), + } + }; + + // SAFETY: `ret` comes from reading `self.0`, which is a valid `T` per type invariants. + unsafe { from_repr(ret) } + } + + /// Atomic fetch and subtract. + /// + /// Atomically updates `*self` to `(*self).wrapping_sub(v)`, and returns the value of `*self` + /// before the update. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{Atomic, Acquire, Full, Relaxed}; + /// + /// let x = Atomic::new(42); + /// assert_eq!(42, x.load(Relaxed)); + /// assert_eq!(42, x.fetch_sub(12, Acquire)); + /// assert_eq!(30, x.load(Relaxed)); + /// + /// let x = Atomic::new(42); + /// assert_eq!(42, x.load(Relaxed)); + /// assert_eq!(42, x.fetch_sub(12, Full)); + /// assert_eq!(30, x.load(Relaxed)); + /// ``` + #[inline(always)] + pub fn fetch_sub<Rhs, Ordering: ordering::Ordering>(&self, v: Rhs, _: Ordering) -> T + where + // Types that support addition also support subtraction. + T: AtomicAdd<Rhs>, + { + let v = T::rhs_into_delta(v); + + // INVARIANT: `self.0` is a valid `T` after `atomic_fetch_sub*()` due to safety requirement + // of `AtomicAdd`. + let ret = { + match Ordering::TYPE { + OrderingType::Full => T::Repr::atomic_fetch_sub(&self.0, v), + OrderingType::Acquire => T::Repr::atomic_fetch_sub_acquire(&self.0, v), + OrderingType::Release => T::Repr::atomic_fetch_sub_release(&self.0, v), + OrderingType::Relaxed => T::Repr::atomic_fetch_sub_relaxed(&self.0, v), + } + }; + + // SAFETY: `ret` comes from reading `self.0`, which is a valid `T` per type invariants. + unsafe { from_repr(ret) } + } +} + +#[cfg(any(CONFIG_X86_64, CONFIG_UML, CONFIG_ARM, CONFIG_ARM64))] +#[repr(C)] +#[derive(Clone, Copy)] +struct Flag { + bool_field: bool, +} + +/// # Invariants +/// +/// `padding` must be all zeroes. +#[cfg(not(any(CONFIG_X86_64, CONFIG_UML, CONFIG_ARM, CONFIG_ARM64)))] +#[repr(C, align(4))] +#[derive(Clone, Copy)] +struct Flag { + #[cfg(target_endian = "big")] + padding: [u8; 3], + bool_field: bool, + #[cfg(target_endian = "little")] + padding: [u8; 3], +} + +impl Flag { + #[inline(always)] + const fn new(b: bool) -> Self { + // INVARIANT: `padding` is all zeroes. + Self { + bool_field: b, + #[cfg(not(any(CONFIG_X86_64, CONFIG_UML, CONFIG_ARM, CONFIG_ARM64)))] + padding: [0; 3], + } + } +} + +// SAFETY: `Flag` and `Repr` have the same size and alignment, and `Flag` is round-trip +// transmutable to the selected representation (`i8` or `i32`). +unsafe impl AtomicType for Flag { + #[cfg(any(CONFIG_X86_64, CONFIG_UML, CONFIG_ARM, CONFIG_ARM64))] + type Repr = i8; + #[cfg(not(any(CONFIG_X86_64, CONFIG_UML, CONFIG_ARM, CONFIG_ARM64)))] + type Repr = i32; +} + +/// An atomic flag type intended to be backed by performance-optimal integer type. +/// +/// The backing integer type is an implementation detail; it may vary by architecture and change +/// in the future. +/// +/// [`AtomicFlag`] is generally preferable to [`Atomic<bool>`] when you need read-modify-write +/// (RMW) operations (e.g. [`Atomic::xchg()`]/[`Atomic::cmpxchg()`]) or when [`Atomic<bool>`] does +/// not save memory due to padding. On some architectures that do not support byte-sized atomic +/// RMW operations, RMW operations on [`Atomic<bool>`] are slower. +/// +/// If you only use [`Atomic::load()`]/[`Atomic::store()`], [`Atomic<bool>`] is fine. +/// +/// # Examples +/// +/// ``` +/// use kernel::sync::atomic::{AtomicFlag, Relaxed}; +/// +/// let flag = AtomicFlag::new(false); +/// assert_eq!(false, flag.load(Relaxed)); +/// flag.store(true, Relaxed); +/// assert_eq!(true, flag.load(Relaxed)); +/// ``` +pub struct AtomicFlag(Atomic<Flag>); + +impl AtomicFlag { + /// Creates a new atomic flag. + #[inline(always)] + pub const fn new(b: bool) -> Self { + Self(Atomic::new(Flag::new(b))) + } + + /// Returns a mutable reference to the underlying flag as a [`bool`]. + /// + /// This is safe because the mutable reference of the atomic flag guarantees exclusive access. + /// + /// # Examples + /// + /// ``` + /// use kernel::sync::atomic::{AtomicFlag, Relaxed}; + /// + /// let mut atomic_flag = AtomicFlag::new(false); + /// assert_eq!(false, atomic_flag.load(Relaxed)); + /// *atomic_flag.get_mut() = true; + /// assert_eq!(true, atomic_flag.load(Relaxed)); + /// ``` + #[inline(always)] + pub fn get_mut(&mut self) -> &mut bool { + &mut self.0.get_mut().bool_field + } + + /// Loads the value from the atomic flag. + #[inline(always)] + pub fn load<Ordering: ordering::AcquireOrRelaxed>(&self, o: Ordering) -> bool { + self.0.load(o).bool_field + } + + /// Stores a value to the atomic flag. + #[inline(always)] + pub fn store<Ordering: ordering::ReleaseOrRelaxed>(&self, v: bool, o: Ordering) { + self.0.store(Flag::new(v), o); + } + + /// Stores a value to the atomic flag and returns the previous value. + #[inline(always)] + pub fn xchg<Ordering: ordering::Ordering>(&self, new: bool, o: Ordering) -> bool { + self.0.xchg(Flag::new(new), o).bool_field + } + + /// Store a value to the atomic flag if the current value is equal to `old`. + #[inline(always)] + pub fn cmpxchg<Ordering: ordering::Ordering>( + &self, + old: bool, + new: bool, + o: Ordering, + ) -> Result<bool, bool> { + match self.0.cmpxchg(Flag::new(old), Flag::new(new), o) { + Ok(_) => Ok(old), + Err(f) => Err(f.bool_field), + } + } +} + +/// Atomic load over raw pointers. +/// +/// This function provides a short-cut of `Atomic::from_ptr().load(..)`, and can be used to work +/// with C side on synchronizations: +/// +/// - `atomic_load(.., Relaxed)` maps to `READ_ONCE()` when used for inter-thread communication. +/// - `atomic_load(.., Acquire)` maps to `smp_load_acquire()`. +/// +/// # Safety +/// +/// - `ptr` is a valid pointer to `T` and aligned to `align_of::<T>()`. +/// - If there is a concurrent store from kernel (C or Rust), it has to be atomic. +#[doc(alias("READ_ONCE", "smp_load_acquire"))] +#[inline(always)] +pub unsafe fn atomic_load<T: AtomicType, Ordering: ordering::AcquireOrRelaxed>( + ptr: *mut T, + o: Ordering, +) -> T +where + T::Repr: AtomicBasicOps, +{ + // SAFETY: Per the function safety requirement, `ptr` is valid and aligned to + // `align_of::<T>()`, and all concurrent stores from kernel are atomic, hence no data race per + // LKMM. + unsafe { Atomic::from_ptr(ptr) }.load(o) +} + +/// Atomic store over raw pointers. +/// +/// This function provides a short-cut of `Atomic::from_ptr().load(..)`, and can be used to work +/// with C side on synchronizations: +/// +/// - `atomic_store(.., Relaxed)` maps to `WRITE_ONCE()` when used for inter-thread communication. +/// - `atomic_load(.., Release)` maps to `smp_store_release()`. +/// +/// # Safety +/// +/// - `ptr` is a valid pointer to `T` and aligned to `align_of::<T>()`. +/// - If there is a concurrent access from kernel (C or Rust), it has to be atomic. +#[doc(alias("WRITE_ONCE", "smp_store_release"))] +#[inline(always)] +pub unsafe fn atomic_store<T: AtomicType, Ordering: ordering::ReleaseOrRelaxed>( + ptr: *mut T, + v: T, + o: Ordering, +) where + T::Repr: AtomicBasicOps, +{ + // SAFETY: Per the function safety requirement, `ptr` is valid and aligned to + // `align_of::<T>()`, and all concurrent accesses from kernel are atomic, hence no data race + // per LKMM. + unsafe { Atomic::from_ptr(ptr) }.store(v, o); +} + +/// Atomic exchange over raw pointers. +/// +/// This function provides a short-cut of `Atomic::from_ptr().xchg(..)`, and can be used to work +/// with C side on synchronizations. +/// +/// # Safety +/// +/// - `ptr` is a valid pointer to `T` and aligned to `align_of::<T>()`. +/// - If there is a concurrent access from kernel (C or Rust), it has to be atomic. +#[inline(always)] +pub unsafe fn xchg<T: AtomicType, Ordering: ordering::Ordering>( + ptr: *mut T, + new: T, + o: Ordering, +) -> T +where + T::Repr: AtomicExchangeOps, +{ + // SAFETY: Per the function safety requirement, `ptr` is valid and aligned to + // `align_of::<T>()`, and all concurrent accesses from kernel are atomic, hence no data race + // per LKMM. + unsafe { Atomic::from_ptr(ptr) }.xchg(new, o) +} + +/// Atomic compare and exchange over raw pointers. +/// +/// This function provides a short-cut of `Atomic::from_ptr().cmpxchg(..)`, and can be used to work +/// with C side on synchronizations. +/// +/// # Safety +/// +/// - `ptr` is a valid pointer to `T` and aligned to `align_of::<T>()`. +/// - If there is a concurrent access from kernel (C or Rust), it has to be atomic. +#[doc(alias("try_cmpxchg"))] +#[inline(always)] +pub unsafe fn cmpxchg<T: AtomicType, Ordering: ordering::Ordering>( + ptr: *mut T, + old: T, + new: T, + o: Ordering, +) -> Result<T, T> +where + T::Repr: AtomicExchangeOps, +{ + // SAFETY: Per the function safety requirement, `ptr` is valid and aligned to + // `align_of::<T>()`, and all concurrent accesses from kernel are atomic, hence no data race + // per LKMM. + unsafe { Atomic::from_ptr(ptr) }.cmpxchg(old, new, o) +} diff --git a/rust/kernel/sync/atomic/internal.rs b/rust/kernel/sync/atomic/internal.rs new file mode 100644 index 000000000000..9c8a7a203abd --- /dev/null +++ b/rust/kernel/sync/atomic/internal.rs @@ -0,0 +1,352 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Atomic internal implementations. +//! +//! Provides 1:1 mapping to the C atomic operations. + +use crate::{ + bindings, + build_assert::static_assert, + macros::paste, // +}; +use core::cell::UnsafeCell; +use ffi::c_void; + +mod private { + /// Sealed trait marker to disable customized impls on atomic implementation traits. + pub trait Sealed {} +} + +// The C side supports atomic primitives only for `i32` and `i64` (`atomic_t` and `atomic64_t`), +// while the Rust side also provides atomic support for `i8`, `i16` and `*const c_void` on top of +// lower-level C primitives. +impl private::Sealed for i8 {} +impl private::Sealed for i16 {} +impl private::Sealed for *const c_void {} +impl private::Sealed for i32 {} +impl private::Sealed for i64 {} + +/// A marker trait for types that implement atomic operations with C side primitives. +/// +/// This trait is sealed, and only types that map directly to the C side atomics +/// or can be implemented with lower-level C primitives are allowed to implement this: +/// +/// - `i8`, `i16` and `*const c_void` are implemented with lower-level C primitives. +/// - `i32` map to `atomic_t` +/// - `i64` map to `atomic64_t` +pub trait AtomicImpl: Sized + Copy + private::Sealed { + /// The type of the delta in arithmetic or logical operations. + /// + /// For example, in `atomic_add(ptr, v)`, it's the type of `v`. Usually it's the same type of + /// [`Self`], but it may be different for the atomic pointer type. + type Delta; +} + +// The current helpers of load/store of atomic `i8`, `i16` and pointers use `{WRITE,READ}_ONCE()` +// hence the atomicity is only guaranteed against read-modify-write operations if the architecture +// supports native atomic RmW. +// +// In the future when a CONFIG_ARCH_SUPPORTS_ATOMIC_RMW=n architecture plans to support Rust, the +// load/store helpers that guarantee atomicity against RmW operations (usually via a lock) need to +// be added. +static_assert!( + cfg!(CONFIG_ARCH_SUPPORTS_ATOMIC_RMW), + "The current implementation of atomic i8/i16/ptr relies on the architecure being \ + ARCH_SUPPORTS_ATOMIC_RMW" +); + +impl AtomicImpl for i8 { + type Delta = Self; +} + +impl AtomicImpl for i16 { + type Delta = Self; +} + +impl AtomicImpl for *const c_void { + type Delta = isize; +} + +// `atomic_t` implements atomic operations on `i32`. +impl AtomicImpl for i32 { + type Delta = Self; +} + +// `atomic64_t` implements atomic operations on `i64`. +impl AtomicImpl for i64 { + type Delta = Self; +} + +/// Atomic representation. +#[repr(transparent)] +pub struct AtomicRepr<T: AtomicImpl>(UnsafeCell<T>); + +impl<T: AtomicImpl> AtomicRepr<T> { + /// Creates a new atomic representation `T`. + pub const fn new(v: T) -> Self { + Self(UnsafeCell::new(v)) + } + + /// Returns a pointer to the underlying `T`. + /// + /// # Guarantees + /// + /// The returned pointer is valid and properly aligned (i.e. aligned to [`align_of::<T>()`]). + pub const fn as_ptr(&self) -> *mut T { + // GUARANTEE: `self.0` is an `UnsafeCell<T>`, therefore the pointer returned by `.get()` + // must be valid and properly aligned. + self.0.get() + } +} + +// This macro generates the function signature with given argument list and return type. +macro_rules! declare_atomic_method { + ( + $(#[doc=$doc:expr])* + $func:ident($($arg:ident : $arg_type:ty),*) $(-> $ret:ty)? + ) => { + paste!( + $(#[doc = $doc])* + fn [< atomic_ $func >]($($arg: $arg_type,)*) $(-> $ret)?; + ); + }; + ( + $(#[doc=$doc:expr])* + $func:ident [$variant:ident $($rest:ident)*]($($arg_sig:tt)*) $(-> $ret:ty)? + ) => { + paste!( + declare_atomic_method!( + $(#[doc = $doc])* + [< $func _ $variant >]($($arg_sig)*) $(-> $ret)? + ); + ); + + declare_atomic_method!( + $(#[doc = $doc])* + $func [$($rest)*]($($arg_sig)*) $(-> $ret)? + ); + }; + ( + $(#[doc=$doc:expr])* + $func:ident []($($arg_sig:tt)*) $(-> $ret:ty)? + ) => { + declare_atomic_method!( + $(#[doc = $doc])* + $func($($arg_sig)*) $(-> $ret)? + ); + } +} + +// This macro generates the function implementation with given argument list and return type, and it +// will replace "call(...)" expression with "$ctype _ $func" to call the real C function. +macro_rules! impl_atomic_method { + ( + ($ctype:ident) $func:ident($($arg:ident: $arg_type:ty),*) $(-> $ret:ty)? { + $unsafe:tt { call($($c_arg:expr),*) } + } + ) => { + paste!( + #[inline(always)] + fn [< atomic_ $func >]($($arg: $arg_type,)*) $(-> $ret)? { + // TODO: Ideally we want to use the SAFETY comments written at the macro invocation + // (e.g. in `declare_and_impl_atomic_methods!()`, however, since SAFETY comments + // are just comments, and they are not passed to macros as tokens, therefore we + // cannot use them here. One potential improvement is that if we support using + // attributes as an alternative for SAFETY comments, then we can use that for macro + // generating code. + // + // SAFETY: specified on macro invocation. + $unsafe { bindings::[< $ctype _ $func >]($($c_arg,)*) } + } + ); + }; + ( + ($ctype:ident) $func:ident[$variant:ident $($rest:ident)*]($($arg_sig:tt)*) $(-> $ret:ty)? { + $unsafe:tt { call($($arg:tt)*) } + } + ) => { + paste!( + impl_atomic_method!( + ($ctype) [< $func _ $variant >]($($arg_sig)*) $( -> $ret)? { + $unsafe { call($($arg)*) } + } + ); + ); + impl_atomic_method!( + ($ctype) $func [$($rest)*]($($arg_sig)*) $( -> $ret)? { + $unsafe { call($($arg)*) } + } + ); + }; + ( + ($ctype:ident) $func:ident[]($($arg_sig:tt)*) $( -> $ret:ty)? { + $unsafe:tt { call($($arg:tt)*) } + } + ) => { + impl_atomic_method!( + ($ctype) $func($($arg_sig)*) $(-> $ret)? { + $unsafe { call($($arg)*) } + } + ); + } +} + +macro_rules! declare_atomic_ops_trait { + ( + $(#[$attr:meta])* $pub:vis trait $ops:ident { + $( + $(#[doc=$doc:expr])* + fn $func:ident [$($variant:ident),*]($($arg_sig:tt)*) $( -> $ret:ty)? { + $unsafe:tt { bindings::#call($($arg:tt)*) } + } + )* + } + ) => { + $(#[$attr])* + $pub trait $ops: AtomicImpl { + $( + declare_atomic_method!( + $(#[doc=$doc])* + $func[$($variant)*]($($arg_sig)*) $(-> $ret)? + ); + )* + } + } +} + +macro_rules! impl_atomic_ops_for_one { + ( + $ty:ty => $ctype:ident, + $(#[$attr:meta])* $pub:vis trait $ops:ident { + $( + $(#[doc=$doc:expr])* + fn $func:ident [$($variant:ident),*]($($arg_sig:tt)*) $( -> $ret:ty)? { + $unsafe:tt { bindings::#call($($arg:tt)*) } + } + )* + } + ) => { + impl $ops for $ty { + $( + impl_atomic_method!( + ($ctype) $func[$($variant)*]($($arg_sig)*) $(-> $ret)? { + $unsafe { call($($arg)*) } + } + ); + )* + } + } +} + +// Declares $ops trait with methods and implements the trait. +macro_rules! declare_and_impl_atomic_methods { + ( + [ $($map:tt)* ] + $(#[$attr:meta])* $pub:vis trait $ops:ident { $($body:tt)* } + ) => { + declare_and_impl_atomic_methods!( + @with_ops_def + [ $($map)* ] + ( $(#[$attr])* $pub trait $ops { $($body)* } ) + ); + }; + + (@with_ops_def [ $($map:tt)* ] ( $($ops_def:tt)* )) => { + declare_atomic_ops_trait!( $($ops_def)* ); + + declare_and_impl_atomic_methods!( + @munch + [ $($map)* ] + ( $($ops_def)* ) + ); + }; + + (@munch [] ( $($ops_def:tt)* )) => {}; + + (@munch [ $ty:ty => $ctype:ident $(, $($rest:tt)*)? ] ( $($ops_def:tt)* )) => { + impl_atomic_ops_for_one!( + $ty => $ctype, + $($ops_def)* + ); + + declare_and_impl_atomic_methods!( + @munch + [ $($($rest)*)? ] + ( $($ops_def)* ) + ); + }; +} + +declare_and_impl_atomic_methods!( + [ i8 => atomic_i8, i16 => atomic_i16, *const c_void => atomic_ptr, i32 => atomic, i64 => atomic64 ] + /// Basic atomic operations + pub trait AtomicBasicOps { + /// Atomic read (load). + fn read[acquire](a: &AtomicRepr<Self>) -> Self { + // SAFETY: `a.as_ptr()` is valid and properly aligned. + unsafe { bindings::#call(a.as_ptr().cast()) } + } + + /// Atomic set (store). + fn set[release](a: &AtomicRepr<Self>, v: Self) { + // SAFETY: `a.as_ptr()` is valid and properly aligned. + unsafe { bindings::#call(a.as_ptr().cast(), v) } + } + } +); + +declare_and_impl_atomic_methods!( + [ i8 => atomic_i8, i16 => atomic_i16, *const c_void => atomic_ptr, i32 => atomic, i64 => atomic64 ] + /// Exchange and compare-and-exchange atomic operations + pub trait AtomicExchangeOps { + /// Atomic exchange. + /// + /// Atomically updates `*a` to `v` and returns the old value. + fn xchg[acquire, release, relaxed](a: &AtomicRepr<Self>, v: Self) -> Self { + // SAFETY: `a.as_ptr()` is valid and properly aligned. + unsafe { bindings::#call(a.as_ptr().cast(), v) } + } + + /// Atomic compare and exchange. + /// + /// If `*a` == `*old`, atomically updates `*a` to `new`. Otherwise, `*a` is not + /// modified, `*old` is updated to the current value of `*a`. + /// + /// Return `true` if the update of `*a` occurred, `false` otherwise. + fn try_cmpxchg[acquire, release, relaxed]( + a: &AtomicRepr<Self>, old: &mut Self, new: Self + ) -> bool { + // SAFETY: `a.as_ptr()` is valid and properly aligned. `core::ptr::from_mut(old)` + // is valid and properly aligned. + unsafe { bindings::#call(a.as_ptr().cast(), core::ptr::from_mut(old), new) } + } + } +); + +declare_and_impl_atomic_methods!( + [ i32 => atomic, i64 => atomic64 ] + /// Atomic arithmetic operations + pub trait AtomicArithmeticOps { + /// Atomic add (wrapping). + /// + /// Atomically updates `*a` to `(*a).wrapping_add(v)`. + fn add[](a: &AtomicRepr<Self>, v: Self::Delta) { + // SAFETY: `a.as_ptr()` is valid and properly aligned. + unsafe { bindings::#call(v, a.as_ptr().cast()) } + } + + /// Atomic fetch and add (wrapping). + /// + /// Atomically updates `*a` to `(*a).wrapping_add(v)`, and returns the value of `*a` + /// before the update. + fn fetch_add[acquire, release, relaxed](a: &AtomicRepr<Self>, v: Self::Delta) -> Self { + // SAFETY: `a.as_ptr()` guarantees the returned pointer is valid and properly aligned. + unsafe { bindings::#call(v, a.as_ptr().cast()) } + } + + fn fetch_sub[acquire, release, relaxed](a: &AtomicRepr<Self>, v: Self::Delta) -> Self { + // SAFETY: `a.as_ptr()` guarantees the returned pointer is valid and properly aligned. + unsafe { bindings::#call(v, a.as_ptr().cast()) } + } + } +); diff --git a/rust/kernel/sync/atomic/ordering.rs b/rust/kernel/sync/atomic/ordering.rs new file mode 100644 index 000000000000..3f103aa8db99 --- /dev/null +++ b/rust/kernel/sync/atomic/ordering.rs @@ -0,0 +1,104 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Memory orderings. +//! +//! The semantics of these orderings follows the [`LKMM`] definitions and rules. +//! +//! - [`Acquire`] provides ordering between the load part of the annotated operation and all the +//! following memory accesses, and if there is a store part, the store part has the [`Relaxed`] +//! ordering. +//! - [`Release`] provides ordering between all the preceding memory accesses and the store part of +//! the annotated operation, and if there is a load part, the load part has the [`Relaxed`] +//! ordering. +//! - [`Full`] means "fully-ordered", that is: +//! - It provides ordering between all the preceding memory accesses and the annotated operation. +//! - It provides ordering between the annotated operation and all the following memory accesses. +//! - It provides ordering between all the preceding memory accesses and all the following memory +//! accesses. +//! - All the orderings are the same strength as a full memory barrier (i.e. `smp_mb()`). +//! - [`Relaxed`] provides no ordering except the dependency orderings. Dependency orderings are +//! described in "DEPENDENCY RELATIONS" in [`LKMM`]'s [`explanation`]. +//! +//! [`LKMM`]: srctree/tools/memory-model/ +//! [`explanation`]: srctree/tools/memory-model/Documentation/explanation.txt + +/// The annotation type for relaxed memory ordering, for the description of relaxed memory +/// ordering, see [module-level documentation]. +/// +/// [module-level documentation]: crate::sync::atomic::ordering +pub struct Relaxed; + +/// The annotation type for acquire memory ordering, for the description of acquire memory +/// ordering, see [module-level documentation]. +/// +/// [module-level documentation]: crate::sync::atomic::ordering +pub struct Acquire; + +/// The annotation type for release memory ordering, for the description of release memory +/// ordering, see [module-level documentation]. +/// +/// [module-level documentation]: crate::sync::atomic::ordering +pub struct Release; + +/// The annotation type for fully-ordered memory ordering, for the description fully-ordered memory +/// ordering, see [module-level documentation]. +/// +/// [module-level documentation]: crate::sync::atomic::ordering +pub struct Full; + +/// Describes the exact memory ordering. +#[doc(hidden)] +pub enum OrderingType { + /// Relaxed ordering. + Relaxed, + /// Acquire ordering. + Acquire, + /// Release ordering. + Release, + /// Fully-ordered. + Full, +} + +mod internal { + /// Sealed trait, can be only implemented inside atomic mod. + pub trait Sealed {} + + impl Sealed for super::Relaxed {} + impl Sealed for super::Acquire {} + impl Sealed for super::Release {} + impl Sealed for super::Full {} +} + +/// The trait bound for annotating operations that support any ordering. +pub trait Ordering: internal::Sealed { + /// Describes the exact memory ordering. + const TYPE: OrderingType; +} + +impl Ordering for Relaxed { + const TYPE: OrderingType = OrderingType::Relaxed; +} + +impl Ordering for Acquire { + const TYPE: OrderingType = OrderingType::Acquire; +} + +impl Ordering for Release { + const TYPE: OrderingType = OrderingType::Release; +} + +impl Ordering for Full { + const TYPE: OrderingType = OrderingType::Full; +} + +/// The trait bound for operations that only support acquire or relaxed ordering. +pub trait AcquireOrRelaxed: Ordering {} + +impl AcquireOrRelaxed for Acquire {} +impl AcquireOrRelaxed for Relaxed {} + +/// The trait bound for operations that only support release or relaxed ordering. +pub trait ReleaseOrRelaxed: Ordering {} + +impl ReleaseOrRelaxed for Release {} +impl ReleaseOrRelaxed for Relaxed {} diff --git a/rust/kernel/sync/atomic/predefine.rs b/rust/kernel/sync/atomic/predefine.rs new file mode 100644 index 000000000000..3d63f40791fa --- /dev/null +++ b/rust/kernel/sync/atomic/predefine.rs @@ -0,0 +1,335 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Pre-defined atomic types + +use crate::prelude::*; + +// Ensure size and alignment requirements are checked. +static_assert!(size_of::<bool>() == size_of::<i8>()); +static_assert!(align_of::<bool>() == align_of::<i8>()); + +// SAFETY: `bool` has the same size and alignment as `i8`, and Rust guarantees that `bool` has +// only two valid bit patterns: 0 (false) and 1 (true). Those are valid `i8` values, so `bool` is +// round-trip transmutable to `i8`. +unsafe impl super::AtomicType for bool { + type Repr = i8; +} + +// SAFETY: `i8` has the same size and alignment with itself, and is round-trip transmutable to +// itself. +unsafe impl super::AtomicType for i8 { + type Repr = i8; +} + +// SAFETY: `i16` has the same size and alignment with itself, and is round-trip transmutable to +// itself. +unsafe impl super::AtomicType for i16 { + type Repr = i16; +} + +// SAFETY: +// +// - `*mut T` has the same size and alignment with `*const c_void`, and is round-trip +// transmutable to `*const c_void`. +// - `*mut T` is safe to transfer between execution contexts. See the safety requirement of +// [`AtomicType`]. +unsafe impl<T: Sized> super::AtomicType for *mut T { + type Repr = *const c_void; +} + +// SAFETY: +// +// - `*const T` has the same size and alignment with `*const c_void`, and is round-trip +// transmutable to `*const c_void`. +// - `*const T` is safe to transfer between execution contexts. See the safety requirement of +// [`AtomicType`]. +unsafe impl<T: Sized> super::AtomicType for *const T { + type Repr = *const c_void; +} + +// SAFETY: `i32` has the same size and alignment with itself, and is round-trip transmutable to +// itself. +unsafe impl super::AtomicType for i32 { + type Repr = i32; +} + +// SAFETY: The wrapping add result of two `i32`s is a valid `i32`. +unsafe impl super::AtomicAdd<i32> for i32 { + fn rhs_into_delta(rhs: i32) -> i32 { + rhs + } +} + +// SAFETY: `i64` has the same size and alignment with itself, and is round-trip transmutable to +// itself. +unsafe impl super::AtomicType for i64 { + type Repr = i64; +} + +// SAFETY: The wrapping add result of two `i64`s is a valid `i64`. +unsafe impl super::AtomicAdd<i64> for i64 { + fn rhs_into_delta(rhs: i64) -> i64 { + rhs + } +} + +// Defines an internal type that always maps to the integer type which has the same size alignment +// as `isize` and `usize`, and `isize` and `usize` are always bi-directional transmutable to +// `isize_atomic_repr`, which also always implements `AtomicImpl`. +#[allow(non_camel_case_types)] +#[cfg(not(testlib))] +#[cfg(not(CONFIG_64BIT))] +type isize_atomic_repr = i32; +#[allow(non_camel_case_types)] +#[cfg(not(testlib))] +#[cfg(CONFIG_64BIT)] +type isize_atomic_repr = i64; + +#[allow(non_camel_case_types)] +#[cfg(testlib)] +#[cfg(target_pointer_width = "32")] +type isize_atomic_repr = i32; +#[allow(non_camel_case_types)] +#[cfg(testlib)] +#[cfg(target_pointer_width = "64")] +type isize_atomic_repr = i64; + +// Ensure size and alignment requirements are checked. +static_assert!(size_of::<isize>() == size_of::<isize_atomic_repr>()); +static_assert!(align_of::<isize>() == align_of::<isize_atomic_repr>()); +static_assert!(size_of::<usize>() == size_of::<isize_atomic_repr>()); +static_assert!(align_of::<usize>() == align_of::<isize_atomic_repr>()); + +// SAFETY: `isize` has the same size and alignment with `isize_atomic_repr`, and is round-trip +// transmutable to `isize_atomic_repr`. +unsafe impl super::AtomicType for isize { + type Repr = isize_atomic_repr; +} + +// SAFETY: The wrapping add result of two `isize_atomic_repr`s is a valid `usize`. +unsafe impl super::AtomicAdd<isize> for isize { + fn rhs_into_delta(rhs: isize) -> isize_atomic_repr { + rhs as isize_atomic_repr + } +} + +// SAFETY: `u32` and `i32` has the same size and alignment, and `u32` is round-trip transmutable to +// `i32`. +unsafe impl super::AtomicType for u32 { + type Repr = i32; +} + +// SAFETY: The wrapping add result of two `i32`s is a valid `u32`. +unsafe impl super::AtomicAdd<u32> for u32 { + fn rhs_into_delta(rhs: u32) -> i32 { + rhs as i32 + } +} + +// SAFETY: `u64` and `i64` has the same size and alignment, and `u64` is round-trip transmutable to +// `i64`. +unsafe impl super::AtomicType for u64 { + type Repr = i64; +} + +// SAFETY: The wrapping add result of two `i64`s is a valid `u64`. +unsafe impl super::AtomicAdd<u64> for u64 { + fn rhs_into_delta(rhs: u64) -> i64 { + rhs as i64 + } +} + +// SAFETY: `usize` has the same size and alignment with `isize_atomic_repr`, and is round-trip +// transmutable to `isize_atomic_repr`. +unsafe impl super::AtomicType for usize { + type Repr = isize_atomic_repr; +} + +// SAFETY: The wrapping add result of two `isize_atomic_repr`s is a valid `usize`. +unsafe impl super::AtomicAdd<usize> for usize { + fn rhs_into_delta(rhs: usize) -> isize_atomic_repr { + rhs as isize_atomic_repr + } +} + +#[cfg(CONFIG_RUST_ATOMICS_KUNIT_TEST)] +#[macros::kunit_tests(rust_atomics)] +mod tests { + use super::super::*; + + // Call $fn($val) with each $type of $val. + macro_rules! for_each_type { + ($val:literal in [$($type:ty),*] $fn:expr) => { + $({ + let v: $type = $val; + + $fn(v); + })* + } + } + + #[test] + fn atomic_basic_tests() { + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + + assert_eq!(v, x.load(Relaxed)); + }); + + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + let ptr = x.as_ptr(); + + // SAFETY: `ptr` is a valid pointer and no concurrent access. + assert_eq!(v, unsafe { atomic_load(ptr, Relaxed) }); + }); + } + + #[test] + fn atomic_acquire_release_tests() { + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(0); + + x.store(v, Release); + assert_eq!(v, x.load(Acquire)); + }); + + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(0); + let ptr = x.as_ptr(); + + // SAFETY: `ptr` is a valid pointer and no concurrent access. + unsafe { atomic_store(ptr, v, Release) }; + + // SAFETY: `ptr` is a valid pointer and no concurrent access. + assert_eq!(v, unsafe { atomic_load(ptr, Acquire) }); + }); + } + + #[test] + fn atomic_xchg_tests() { + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + + let old = v; + let new = v + 1; + + assert_eq!(old, x.xchg(new, Full)); + assert_eq!(new, x.load(Relaxed)); + }); + + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + let ptr = x.as_ptr(); + + let old = v; + let new = v + 1; + + // SAFETY: `ptr` is a valid pointer and no concurrent access. + assert_eq!(old, unsafe { xchg(ptr, new, Full) }); + assert_eq!(new, x.load(Relaxed)); + }); + } + + #[test] + fn atomic_cmpxchg_tests() { + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + + let old = v; + let new = v + 1; + + assert_eq!(Err(old), x.cmpxchg(new, new, Full)); + assert_eq!(old, x.load(Relaxed)); + assert_eq!(Ok(old), x.cmpxchg(old, new, Relaxed)); + assert_eq!(new, x.load(Relaxed)); + }); + + for_each_type!(42 in [i8, i16, i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + let ptr = x.as_ptr(); + + let old = v; + let new = v + 1; + + // SAFETY: `ptr` is a valid pointer and no concurrent access. + assert_eq!(Err(old), unsafe { cmpxchg(ptr, new, new, Full) }); + assert_eq!(old, x.load(Relaxed)); + // SAFETY: `ptr` is a valid pointer and no concurrent access. + assert_eq!(Ok(old), unsafe { cmpxchg(ptr, old, new, Relaxed) }); + assert_eq!(new, x.load(Relaxed)); + }); + } + + #[test] + fn atomic_arithmetic_tests() { + for_each_type!(42 in [i32, i64, u32, u64, isize, usize] |v| { + let x = Atomic::new(v); + + assert_eq!(v, x.fetch_add(12, Full)); + assert_eq!(v + 12, x.load(Relaxed)); + + x.add(13, Relaxed); + + assert_eq!(v + 25, x.load(Relaxed)); + }); + } + + #[test] + fn atomic_bool_tests() { + let x = Atomic::new(false); + + assert_eq!(false, x.load(Relaxed)); + x.store(true, Relaxed); + assert_eq!(true, x.load(Relaxed)); + + assert_eq!(true, x.xchg(false, Relaxed)); + assert_eq!(false, x.load(Relaxed)); + + assert_eq!(Err(false), x.cmpxchg(true, true, Relaxed)); + assert_eq!(false, x.load(Relaxed)); + assert_eq!(Ok(false), x.cmpxchg(false, true, Full)); + } + + #[test] + fn atomic_ptr_tests() { + let mut v = 42; + let mut u = 43; + let x = Atomic::new(&raw mut v); + + assert_eq!(x.load(Acquire), &raw mut v); + assert_eq!(x.cmpxchg(&raw mut u, &raw mut u, Relaxed), Err(&raw mut v)); + assert_eq!(x.cmpxchg(&raw mut v, &raw mut u, Relaxed), Ok(&raw mut v)); + assert_eq!(x.load(Relaxed), &raw mut u); + + let x = Atomic::new(&raw const v); + + assert_eq!(x.load(Acquire), &raw const v); + assert_eq!( + x.cmpxchg(&raw const u, &raw const u, Relaxed), + Err(&raw const v) + ); + assert_eq!( + x.cmpxchg(&raw const v, &raw const u, Relaxed), + Ok(&raw const v) + ); + assert_eq!(x.load(Relaxed), &raw const u); + } + + #[test] + fn atomic_flag_tests() { + let mut flag = AtomicFlag::new(false); + + assert_eq!(false, flag.load(Relaxed)); + + *flag.get_mut() = true; + assert_eq!(true, flag.load(Relaxed)); + + assert_eq!(true, flag.xchg(false, Relaxed)); + assert_eq!(false, flag.load(Relaxed)); + + *flag.get_mut() = true; + assert_eq!(Ok(true), flag.cmpxchg(true, false, Full)); + assert_eq!(false, flag.load(Relaxed)); + } +} diff --git a/rust/kernel/sync/barrier.rs b/rust/kernel/sync/barrier.rs new file mode 100644 index 000000000000..8f2d435fcd94 --- /dev/null +++ b/rust/kernel/sync/barrier.rs @@ -0,0 +1,61 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Memory barriers. +//! +//! These primitives have the same semantics as their C counterparts: and the precise definitions +//! of semantics can be found at [`LKMM`]. +//! +//! [`LKMM`]: srctree/tools/memory-model/ + +/// A compiler barrier. +/// +/// A barrier that prevents compiler from reordering memory accesses across the barrier. +#[inline(always)] +pub(crate) fn barrier() { + // By default, Rust inline asms are treated as being able to access any memory or flags, hence + // it suffices as a compiler barrier. + // + // SAFETY: An empty asm block. + unsafe { core::arch::asm!("") }; +} + +/// A full memory barrier. +/// +/// A barrier that prevents compiler and CPU from reordering memory accesses across the barrier. +#[inline(always)] +pub fn smp_mb() { + if cfg!(CONFIG_SMP) { + // SAFETY: `smp_mb()` is safe to call. + unsafe { bindings::smp_mb() }; + } else { + barrier(); + } +} + +/// A write-write memory barrier. +/// +/// A barrier that prevents compiler and CPU from reordering memory write accesses across the +/// barrier. +#[inline(always)] +pub fn smp_wmb() { + if cfg!(CONFIG_SMP) { + // SAFETY: `smp_wmb()` is safe to call. + unsafe { bindings::smp_wmb() }; + } else { + barrier(); + } +} + +/// A read-read memory barrier. +/// +/// A barrier that prevents compiler and CPU from reordering memory read accesses across the +/// barrier. +#[inline(always)] +pub fn smp_rmb() { + if cfg!(CONFIG_SMP) { + // SAFETY: `smp_rmb()` is safe to call. + unsafe { bindings::smp_rmb() }; + } else { + barrier(); + } +} diff --git a/rust/kernel/sync/completion.rs b/rust/kernel/sync/completion.rs new file mode 100644 index 000000000000..35ff049ff078 --- /dev/null +++ b/rust/kernel/sync/completion.rs @@ -0,0 +1,114 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Completion support. +//! +//! Reference: <https://docs.kernel.org/scheduler/completion.html> +//! +//! C header: [`include/linux/completion.h`](srctree/include/linux/completion.h) + +use crate::{bindings, prelude::*, types::Opaque}; + +/// Synchronization primitive to signal when a certain task has been completed. +/// +/// The [`Completion`] synchronization primitive signals when a certain task has been completed by +/// waking up other tasks that have been queued up to wait for the [`Completion`] to be completed. +/// +/// # Examples +/// +/// ``` +/// use kernel::sync::{Arc, Completion}; +/// use kernel::workqueue::{self, impl_has_work, new_work, Work, WorkItem}; +/// +/// #[pin_data] +/// struct MyTask { +/// #[pin] +/// work: Work<MyTask>, +/// #[pin] +/// done: Completion, +/// } +/// +/// impl_has_work! { +/// impl HasWork<Self> for MyTask { self.work } +/// } +/// +/// impl MyTask { +/// fn new() -> Result<Arc<Self>> { +/// let this = Arc::pin_init(pin_init!(MyTask { +/// work <- new_work!("MyTask::work"), +/// done <- Completion::new(), +/// }), GFP_KERNEL)?; +/// +/// let _ = workqueue::system().enqueue(this.clone()); +/// +/// Ok(this) +/// } +/// +/// fn wait_for_completion(&self) { +/// self.done.wait_for_completion(); +/// +/// pr_info!("Completion: task complete\n"); +/// } +/// } +/// +/// impl WorkItem for MyTask { +/// type Pointer = Arc<MyTask>; +/// +/// fn run(this: Arc<MyTask>) { +/// // process this task +/// this.done.complete_all(); +/// } +/// } +/// +/// let task = MyTask::new()?; +/// task.wait_for_completion(); +/// # Ok::<(), Error>(()) +/// ``` +#[pin_data] +pub struct Completion { + #[pin] + inner: Opaque<bindings::completion>, +} + +// SAFETY: `Completion` is safe to be send to any task. +unsafe impl Send for Completion {} + +// SAFETY: `Completion` is safe to be accessed concurrently. +unsafe impl Sync for Completion {} + +impl Completion { + /// Create an initializer for a new [`Completion`]. + pub fn new() -> impl PinInit<Self> { + pin_init!(Self { + inner <- Opaque::ffi_init(|slot: *mut bindings::completion| { + // SAFETY: `slot` is a valid pointer to an uninitialized `struct completion`. + unsafe { bindings::init_completion(slot) }; + }), + }) + } + + fn as_raw(&self) -> *mut bindings::completion { + self.inner.get() + } + + /// Signal all tasks waiting on this completion. + /// + /// This method wakes up all tasks waiting on this completion; after this operation the + /// completion is permanently done, i.e. signals all current and future waiters. + #[inline] + pub fn complete_all(&self) { + // SAFETY: `self.as_raw()` is a pointer to a valid `struct completion`. + unsafe { bindings::complete_all(self.as_raw()) }; + } + + /// Wait for completion of a task. + /// + /// This method waits for the completion of a task; it is not interruptible and there is no + /// timeout. + /// + /// See also [`Completion::complete_all`]. + #[inline] + pub fn wait_for_completion(&self) { + // SAFETY: `self.as_raw()` is a pointer to a valid `struct completion`. + unsafe { bindings::wait_for_completion(self.as_raw()) }; + } +} diff --git a/rust/kernel/sync/condvar.rs b/rust/kernel/sync/condvar.rs index 7df565038d7d..69d58dfbad7b 100644 --- a/rust/kernel/sync/condvar.rs +++ b/rust/kernel/sync/condvar.rs @@ -8,16 +8,15 @@ use super::{lock::Backend, lock::Guard, LockClassKey}; use crate::{ ffi::{c_int, c_long}, - init::PinInit, - pin_init, - str::CStr, - task::{MAX_SCHEDULE_TIMEOUT, TASK_INTERRUPTIBLE, TASK_NORMAL, TASK_UNINTERRUPTIBLE}, + str::{CStr, CStrExt as _}, + task::{ + MAX_SCHEDULE_TIMEOUT, TASK_FREEZABLE, TASK_INTERRUPTIBLE, TASK_NORMAL, TASK_UNINTERRUPTIBLE, + }, time::Jiffies, types::Opaque, }; -use core::marker::PhantomPinned; -use core::ptr; -use macros::pin_data; +use core::{marker::PhantomPinned, pin::Pin, ptr}; +use pin_init::{pin_data, pin_init, PinInit}; /// Creates a [`CondVar`] initialiser with the given name and a newly-created lock class. #[macro_export] @@ -37,7 +36,7 @@ pub use new_condvar; /// spuriously. /// /// Instances of [`CondVar`] need a lock class and to be pinned. The recommended way to create such -/// instances is with the [`pin_init`](crate::pin_init) and [`new_condvar`] macros. +/// instances is with the [`pin_init`](pin_init::pin_init!) and [`new_condvar`] macros. /// /// # Examples /// @@ -101,7 +100,7 @@ unsafe impl Sync for CondVar {} impl CondVar { /// Constructs a new condvar initialiser. - pub fn new(name: &'static CStr, key: &'static LockClassKey) -> impl PinInit<Self> { + pub fn new(name: &'static CStr, key: Pin<&'static LockClassKey>) -> impl PinInit<Self> { pin_init!(Self { _pin: PhantomPinned, // SAFETY: `slot` is valid while the closure is called and both `name` and `key` have @@ -159,6 +158,25 @@ impl CondVar { crate::current!().signal_pending() } + /// Releases the lock and waits for a notification in interruptible and freezable mode. + /// + /// The process is allowed to be frozen during this sleep. No lock should be held when calling + /// this function, and there is a lockdep assertion for this. Freezing a task that holds a lock + /// can trivially deadlock vs another task that needs that lock to complete before it too can + /// hit freezable. + #[must_use = "wait_interruptible_freezable returns if a signal is pending, so the caller must check the return value"] + pub fn wait_interruptible_freezable<T: ?Sized, B: Backend>( + &self, + guard: &mut Guard<'_, T, B>, + ) -> bool { + self.wait_internal( + TASK_INTERRUPTIBLE | TASK_FREEZABLE, + guard, + MAX_SCHEDULE_TIMEOUT, + ); + crate::current!().signal_pending() + } + /// Releases the lock and waits for a notification in interruptible mode. /// /// Atomically releases the given lock (whose ownership is proven by the guard) and puts the @@ -198,6 +216,7 @@ impl CondVar { /// This method behaves like `notify_one`, except that it hints to the scheduler that the /// current thread is about to go to sleep, so it should schedule the target thread on the same /// CPU. + #[inline] pub fn notify_sync(&self) { // SAFETY: `wait_queue_head` points to valid memory. unsafe { bindings::__wake_up_sync(self.wait_queue_head.get(), TASK_NORMAL) }; @@ -207,6 +226,7 @@ impl CondVar { /// /// This is not 'sticky' in the sense that if no thread is waiting, the notification is lost /// completely (as opposed to automatically waking up the next waiter). + #[inline] pub fn notify_one(&self) { self.notify(1); } @@ -215,6 +235,7 @@ impl CondVar { /// /// This is not 'sticky' in the sense that if no thread is waiting, the notification is lost /// completely (as opposed to automatically waking up the next waiter). + #[inline] pub fn notify_all(&self) { self.notify(0); } diff --git a/rust/kernel/sync/lock.rs b/rust/kernel/sync/lock.rs index eb80048e0110..10b6b5e9b024 100644 --- a/rust/kernel/sync/lock.rs +++ b/rust/kernel/sync/lock.rs @@ -7,13 +7,11 @@ use super::LockClassKey; use crate::{ - init::PinInit, - pin_init, - str::CStr, + str::{CStr, CStrExt as _}, types::{NotThreadSafe, Opaque, ScopeGuard}, }; -use core::{cell::UnsafeCell, marker::PhantomPinned}; -use macros::pin_data; +use core::{cell::UnsafeCell, marker::PhantomPinned, pin::Pin}; +use pin_init::{pin_data, pin_init, PinInit, Wrapper}; pub mod mutex; pub mod spinlock; @@ -117,6 +115,7 @@ pub struct Lock<T: ?Sized, B: Backend> { _pin: PhantomPinned, /// The data protected by the lock. + #[pin] pub(crate) data: UnsafeCell<T>, } @@ -129,9 +128,13 @@ unsafe impl<T: ?Sized + Send, B: Backend> Sync for Lock<T, B> {} impl<T, B: Backend> Lock<T, B> { /// Constructs a new lock initialiser. - pub fn new(t: T, name: &'static CStr, key: &'static LockClassKey) -> impl PinInit<Self> { + pub fn new( + t: impl PinInit<T>, + name: &'static CStr, + key: Pin<&'static LockClassKey>, + ) -> impl PinInit<Self> { pin_init!(Self { - data: UnsafeCell::new(t), + data <- UnsafeCell::pin_init(t), _pin: PhantomPinned, // SAFETY: `slot` is valid while the closure is called and both `name` and `key` have // static lifetimes so they live indefinitely. @@ -153,6 +156,7 @@ impl<B: Backend> Lock<(), B> { /// the whole lifetime of `'a`. /// /// [`State`]: Backend::State + #[inline] pub unsafe fn from_raw<'a>(ptr: *mut B::State) -> &'a Self { // SAFETY: // - By the safety contract `ptr` must point to a valid initialised instance of `B::State` @@ -166,6 +170,7 @@ impl<B: Backend> Lock<(), B> { impl<T: ?Sized, B: Backend> Lock<T, B> { /// Acquires the lock and gives the caller access to the data protected by it. + #[inline] pub fn lock(&self) -> Guard<'_, T, B> { // SAFETY: The constructor of the type calls `init`, so the existence of the object proves // that `init` was called. @@ -177,6 +182,9 @@ impl<T: ?Sized, B: Backend> Lock<T, B> { /// Tries to acquire the lock. /// /// Returns a guard that can be used to access the data protected by the lock if successful. + // `Option<T>` is not `#[must_use]` even if `T` is, thus the attribute is needed here. + #[must_use = "if unused, the lock will be immediately unlocked"] + #[inline] pub fn try_lock(&self) -> Option<Guard<'_, T, B>> { // SAFETY: The constructor of the type calls `init`, so the existence of the object proves // that `init` was called. @@ -199,7 +207,37 @@ pub struct Guard<'a, T: ?Sized, B: Backend> { // SAFETY: `Guard` is sync when the data protected by the lock is also sync. unsafe impl<T: Sync + ?Sized, B: Backend> Sync for Guard<'_, T, B> {} -impl<T: ?Sized, B: Backend> Guard<'_, T, B> { +impl<'a, T: ?Sized, B: Backend> Guard<'a, T, B> { + /// Returns the lock that this guard originates from. + /// + /// # Examples + /// + /// The following example shows how to use [`Guard::lock_ref()`] to assert the corresponding + /// lock is held. + /// + /// ``` + /// # use kernel::{new_spinlock, sync::lock::{Backend, Guard, Lock}}; + /// # use pin_init::stack_pin_init; + /// + /// fn assert_held<T, B: Backend>(guard: &Guard<'_, T, B>, lock: &Lock<T, B>) { + /// // Address-equal means the same lock. + /// assert!(core::ptr::eq(guard.lock_ref(), lock)); + /// } + /// + /// // Creates a new lock on the stack. + /// stack_pin_init!{ + /// let l = new_spinlock!(42) + /// } + /// + /// let g = l.lock(); + /// + /// // `g` originates from `l`. + /// assert_held(&g, &l); + /// ``` + pub fn lock_ref(&self) -> &'a Lock<T, B> { + self.lock + } + pub(crate) fn do_unlocked<U>(&mut self, cb: impl FnOnce() -> U) -> U { // SAFETY: The caller owns the lock, so it is safe to unlock it. unsafe { B::unlock(self.lock.state.get(), &self.state) }; @@ -210,18 +248,48 @@ impl<T: ?Sized, B: Backend> Guard<'_, T, B> { cb() } + + /// Returns a pinned mutable reference to the protected data. + /// + /// The guard implements [`DerefMut`] when `T: Unpin`, so for [`Unpin`] + /// types [`DerefMut`] should be used instead of this function. + /// + /// [`DerefMut`]: core::ops::DerefMut + /// [`Unpin`]: core::marker::Unpin + /// + /// # Examples + /// + /// ``` + /// # use kernel::sync::{Mutex, MutexGuard}; + /// # use core::{pin::Pin, marker::PhantomPinned}; + /// struct Data(PhantomPinned); + /// + /// fn example(mutex: &Mutex<Data>) { + /// let mut data: MutexGuard<'_, Data> = mutex.lock(); + /// let mut data: Pin<&mut Data> = data.as_mut(); + /// } + /// ``` + pub fn as_mut(&mut self) -> Pin<&mut T> { + // SAFETY: `self.lock.data` is structurally pinned. + unsafe { Pin::new_unchecked(&mut *self.lock.data.get()) } + } } impl<T: ?Sized, B: Backend> core::ops::Deref for Guard<'_, T, B> { type Target = T; + #[inline] fn deref(&self) -> &Self::Target { // SAFETY: The caller owns the lock, so it is safe to deref the protected data. unsafe { &*self.lock.data.get() } } } -impl<T: ?Sized, B: Backend> core::ops::DerefMut for Guard<'_, T, B> { +impl<T: ?Sized, B: Backend> core::ops::DerefMut for Guard<'_, T, B> +where + T: Unpin, +{ + #[inline] fn deref_mut(&mut self) -> &mut Self::Target { // SAFETY: The caller owns the lock, so it is safe to deref the protected data. unsafe { &mut *self.lock.data.get() } @@ -229,6 +297,7 @@ impl<T: ?Sized, B: Backend> core::ops::DerefMut for Guard<'_, T, B> { } impl<T: ?Sized, B: Backend> Drop for Guard<'_, T, B> { + #[inline] fn drop(&mut self) { // SAFETY: The caller owns the lock, so it is safe to unlock it. unsafe { B::unlock(self.lock.state.get(), &self.state) }; @@ -241,6 +310,7 @@ impl<'a, T: ?Sized, B: Backend> Guard<'a, T, B> { /// # Safety /// /// The caller must ensure that it owns the lock. + #[inline] pub unsafe fn new(lock: &'a Lock<T, B>, state: B::GuardState) -> Self { // SAFETY: The caller can only hold the lock if `Backend::init` has already been called. unsafe { B::assert_is_held(lock.state.get()) }; diff --git a/rust/kernel/sync/lock/global.rs b/rust/kernel/sync/lock/global.rs index 480ee724e3cc..ec2dd84316fc 100644 --- a/rust/kernel/sync/lock/global.rs +++ b/rust/kernel/sync/lock/global.rs @@ -5,7 +5,7 @@ //! Support for defining statics containing locks. use crate::{ - str::CStr, + str::{CStr, CStrExt as _}, sync::lock::{Backend, Guard, Lock}, sync::{LockClassKey, LockedBy}, types::Opaque, @@ -13,6 +13,7 @@ use crate::{ use core::{ cell::UnsafeCell, marker::{PhantomData, PhantomPinned}, + pin::Pin, }; /// Trait implemented for marker types for global locks. @@ -26,7 +27,7 @@ pub trait GlobalLockBackend { /// The backend used for this global lock. type Backend: Backend + 'static; /// The class for this global lock. - fn get_lock_class() -> &'static LockClassKey; + fn get_lock_class() -> Pin<&'static LockClassKey>; } /// Type used for global locks. @@ -76,6 +77,7 @@ impl<B: GlobalLockBackend> GlobalLock<B> { } /// Lock this global lock. + #[inline] pub fn lock(&'static self) -> GlobalGuard<B> { GlobalGuard { inner: self.inner.lock(), @@ -83,6 +85,8 @@ impl<B: GlobalLockBackend> GlobalLock<B> { } /// Try to lock this global lock. + #[must_use = "if unused, the lock will be immediately unlocked"] + #[inline] pub fn try_lock(&'static self) -> Option<GlobalGuard<B>> { Some(GlobalGuard { inner: self.inner.try_lock()?, @@ -93,6 +97,7 @@ impl<B: GlobalLockBackend> GlobalLock<B> { /// A guard for a [`GlobalLock`]. /// /// See [`global_lock!`] for examples. +#[must_use = "the lock unlocks immediately when the guard is unused"] pub struct GlobalGuard<B: GlobalLockBackend> { inner: Guard<'static, B::Item, B::Backend>, } @@ -105,7 +110,10 @@ impl<B: GlobalLockBackend> core::ops::Deref for GlobalGuard<B> { } } -impl<B: GlobalLockBackend> core::ops::DerefMut for GlobalGuard<B> { +impl<B: GlobalLockBackend> core::ops::DerefMut for GlobalGuard<B> +where + B::Item: Unpin, +{ fn deref_mut(&mut self) -> &mut Self::Target { &mut self.inner } @@ -270,7 +278,7 @@ macro_rules! global_lock { type Item = $valuety; type Backend = $crate::global_lock_inner!(backend $kind); - fn get_lock_class() -> &'static $crate::sync::LockClassKey { + fn get_lock_class() -> Pin<&'static $crate::sync::LockClassKey> { $crate::static_lock_class!() } } diff --git a/rust/kernel/sync/lock/mutex.rs b/rust/kernel/sync/lock/mutex.rs index 70cadbc2e8e2..cda0203efefb 100644 --- a/rust/kernel/sync/lock/mutex.rs +++ b/rust/kernel/sync/lock/mutex.rs @@ -26,7 +26,7 @@ pub use new_mutex; /// Since it may block, [`Mutex`] needs to be used with care in atomic contexts. /// /// Instances of [`Mutex`] need a lock class and to be pinned. The recommended way to create such -/// instances is with the [`pin_init`](crate::pin_init) and [`new_mutex`] macros. +/// instances is with the [`pin_init`](pin_init::pin_init) and [`new_mutex`] macros. /// /// # Examples /// @@ -102,6 +102,7 @@ unsafe impl super::Backend for MutexBackend { type State = bindings::mutex; type GuardState = (); + #[inline] unsafe fn init( ptr: *mut Self::State, name: *const crate::ffi::c_char, @@ -112,18 +113,21 @@ unsafe impl super::Backend for MutexBackend { unsafe { bindings::__mutex_init(ptr, name, key) } } + #[inline] unsafe fn lock(ptr: *mut Self::State) -> Self::GuardState { // SAFETY: The safety requirements of this function ensure that `ptr` points to valid // memory, and that it has been initialised before. unsafe { bindings::mutex_lock(ptr) }; } + #[inline] unsafe fn unlock(ptr: *mut Self::State, _guard_state: &Self::GuardState) { // SAFETY: The safety requirements of this function ensure that `ptr` is valid and that the // caller is the owner of the mutex. unsafe { bindings::mutex_unlock(ptr) }; } + #[inline] unsafe fn try_lock(ptr: *mut Self::State) -> Option<Self::GuardState> { // SAFETY: The `ptr` pointer is guaranteed to be valid and initialized before use. let result = unsafe { bindings::mutex_trylock(ptr) }; @@ -135,6 +139,7 @@ unsafe impl super::Backend for MutexBackend { } } + #[inline] unsafe fn assert_is_held(ptr: *mut Self::State) { // SAFETY: The `ptr` pointer is guaranteed to be valid and initialized before use. unsafe { bindings::mutex_assert_is_held(ptr) } diff --git a/rust/kernel/sync/lock/spinlock.rs b/rust/kernel/sync/lock/spinlock.rs index ab2f8d075311..ef76fa07ca3a 100644 --- a/rust/kernel/sync/lock/spinlock.rs +++ b/rust/kernel/sync/lock/spinlock.rs @@ -24,7 +24,7 @@ pub use new_spinlock; /// unlocked, at which point another CPU will be allowed to make progress. /// /// Instances of [`SpinLock`] need a lock class and to be pinned. The recommended way to create such -/// instances is with the [`pin_init`](crate::pin_init) and [`new_spinlock`] macros. +/// instances is with the [`pin_init`](pin_init::pin_init) and [`new_spinlock`] macros. /// /// # Examples /// @@ -101,6 +101,7 @@ unsafe impl super::Backend for SpinLockBackend { type State = bindings::spinlock_t; type GuardState = (); + #[inline] unsafe fn init( ptr: *mut Self::State, name: *const crate::ffi::c_char, @@ -111,18 +112,21 @@ unsafe impl super::Backend for SpinLockBackend { unsafe { bindings::__spin_lock_init(ptr, name, key) } } + #[inline] unsafe fn lock(ptr: *mut Self::State) -> Self::GuardState { // SAFETY: The safety requirements of this function ensure that `ptr` points to valid // memory, and that it has been initialised before. unsafe { bindings::spin_lock(ptr) } } + #[inline] unsafe fn unlock(ptr: *mut Self::State, _guard_state: &Self::GuardState) { // SAFETY: The safety requirements of this function ensure that `ptr` is valid and that the // caller is the owner of the spinlock. unsafe { bindings::spin_unlock(ptr) } } + #[inline] unsafe fn try_lock(ptr: *mut Self::State) -> Option<Self::GuardState> { // SAFETY: The `ptr` pointer is guaranteed to be valid and initialized before use. let result = unsafe { bindings::spin_trylock(ptr) }; @@ -134,6 +138,7 @@ unsafe impl super::Backend for SpinLockBackend { } } + #[inline] unsafe fn assert_is_held(ptr: *mut Self::State) { // SAFETY: The `ptr` pointer is guaranteed to be valid and initialized before use. unsafe { bindings::spin_assert_is_held(ptr) } diff --git a/rust/kernel/sync/locked_by.rs b/rust/kernel/sync/locked_by.rs index 61f100a45b35..fb4a1430b3b4 100644 --- a/rust/kernel/sync/locked_by.rs +++ b/rust/kernel/sync/locked_by.rs @@ -3,7 +3,7 @@ //! A wrapper for data protected by a lock that does not wrap it. use super::{lock::Backend, lock::Lock}; -use crate::build_assert; +use crate::build_assert::build_assert; use core::{cell::UnsafeCell, mem::size_of, ptr}; /// Allows access to some data to be serialised by a lock that does not wrap it. diff --git a/rust/kernel/sync/poll.rs b/rust/kernel/sync/poll.rs index d5f17153b424..0ec985d560c8 100644 --- a/rust/kernel/sync/poll.rs +++ b/rust/kernel/sync/poll.rs @@ -9,9 +9,8 @@ use crate::{ fs::File, prelude::*, sync::{CondVar, LockClassKey}, - types::Opaque, }; -use core::ops::Deref; +use core::{marker::PhantomData, ops::Deref}; /// Creates a [`PollCondVar`] initialiser with the given name and a newly-created lock class. #[macro_export] @@ -23,58 +22,43 @@ macro_rules! new_poll_condvar { }; } -/// Wraps the kernel's `struct poll_table`. +/// Wraps the kernel's `poll_table`. /// /// # Invariants /// -/// This struct contains a valid `struct poll_table`. -/// -/// For a `struct poll_table` to be valid, its `_qproc` function must follow the safety -/// requirements of `_qproc` functions: -/// -/// * The `_qproc` function is given permission to enqueue a waiter to the provided `poll_table` -/// during the call. Once the waiter is removed and an rcu grace period has passed, it must no -/// longer access the `wait_queue_head`. +/// The pointer must be null or reference a valid `poll_table`. #[repr(transparent)] -pub struct PollTable(Opaque<bindings::poll_table>); +pub struct PollTable<'a> { + table: *mut bindings::poll_table, + _lifetime: PhantomData<&'a bindings::poll_table>, +} -impl PollTable { - /// Creates a reference to a [`PollTable`] from a valid pointer. +impl<'a> PollTable<'a> { + /// Creates a [`PollTable`] from a valid pointer. /// /// # Safety /// - /// The caller must ensure that for the duration of 'a, the pointer will point at a valid poll - /// table (as defined in the type invariants). - /// - /// The caller must also ensure that the `poll_table` is only accessed via the returned - /// reference for the duration of 'a. - pub unsafe fn from_ptr<'a>(ptr: *mut bindings::poll_table) -> &'a mut PollTable { - // SAFETY: The safety requirements guarantee the validity of the dereference, while the - // `PollTable` type being transparent makes the cast ok. - unsafe { &mut *ptr.cast() } - } - - fn get_qproc(&self) -> bindings::poll_queue_proc { - let ptr = self.0.get(); - // SAFETY: The `ptr` is valid because it originates from a reference, and the `_qproc` - // field is not modified concurrently with this call since we have an immutable reference. - unsafe { (*ptr)._qproc } + /// The pointer must be null or reference a valid `poll_table` for the duration of `'a`. + pub unsafe fn from_raw(table: *mut bindings::poll_table) -> Self { + // INVARIANTS: The safety requirements are the same as the struct invariants. + PollTable { + table, + _lifetime: PhantomData, + } } /// Register this [`PollTable`] with the provided [`PollCondVar`], so that it can be notified /// using the condition variable. - pub fn register_wait(&mut self, file: &File, cv: &PollCondVar) { - if let Some(qproc) = self.get_qproc() { - // SAFETY: The pointers to `file` and `self` need to be valid for the duration of this - // call to `qproc`, which they are because they are references. - // - // The `cv.wait_queue_head` pointer must be valid until an rcu grace period after the - // waiter is removed. The `PollCondVar` is pinned, so before `cv.wait_queue_head` can - // be destroyed, the destructor must run. That destructor first removes all waiters, - // and then waits for an rcu grace period. Therefore, `cv.wait_queue_head` is valid for - // long enough. - unsafe { qproc(file.as_ptr() as _, cv.wait_queue_head.get(), self.0.get()) }; - } + pub fn register_wait(&self, file: &File, cv: &PollCondVar) { + // SAFETY: + // * `file.as_ptr()` references a valid file for the duration of this call. + // * `self.table` is null or references a valid poll_table for the duration of this call. + // * Since `PollCondVar` is pinned, its destructor is guaranteed to run before the memory + // containing `cv.wait_queue_head` is invalidated. Since the destructor clears all + // waiters and then waits for an rcu grace period, it's guaranteed that + // `cv.wait_queue_head` remains valid for at least an rcu grace period after the removal + // of the last waiter. + unsafe { bindings::poll_wait(file.as_ptr(), cv.wait_queue_head.get(), self.table) } } } @@ -89,7 +73,7 @@ pub struct PollCondVar { impl PollCondVar { /// Constructs a new condvar initialiser. - pub fn new(name: &'static CStr, key: &'static LockClassKey) -> impl PinInit<Self> { + pub fn new(name: &'static CStr, key: Pin<&'static LockClassKey>) -> impl PinInit<Self> { pin_init!(Self { inner <- CondVar::new(name, key), }) @@ -107,6 +91,7 @@ impl Deref for PollCondVar { #[pinned_drop] impl PinnedDrop for PollCondVar { + #[inline] fn drop(self: Pin<&mut Self>) { // Clear anything registered using `register_wait`. // diff --git a/rust/kernel/sync/rcu.rs b/rust/kernel/sync/rcu.rs index b51d9150ffe2..a32bef6e490b 100644 --- a/rust/kernel/sync/rcu.rs +++ b/rust/kernel/sync/rcu.rs @@ -17,6 +17,7 @@ pub struct Guard(NotThreadSafe); impl Guard { /// Acquires the RCU read side lock and returns a guard. + #[inline] pub fn new() -> Self { // SAFETY: An FFI call with no additional requirements. unsafe { bindings::rcu_read_lock() }; @@ -25,16 +26,19 @@ impl Guard { } /// Explicitly releases the RCU read side lock. + #[inline] pub fn unlock(self) {} } impl Default for Guard { + #[inline] fn default() -> Self { Self::new() } } impl Drop for Guard { + #[inline] fn drop(&mut self) { // SAFETY: By the type invariants, the RCU read side is locked, so it is ok to unlock it. unsafe { bindings::rcu_read_unlock() }; @@ -42,6 +46,7 @@ impl Drop for Guard { } /// Acquires the RCU read side lock. +#[inline] pub fn read_lock() -> Guard { Guard::new() } diff --git a/rust/kernel/sync/refcount.rs b/rust/kernel/sync/refcount.rs new file mode 100644 index 000000000000..23a5d201f343 --- /dev/null +++ b/rust/kernel/sync/refcount.rs @@ -0,0 +1,116 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! Atomic reference counting. +//! +//! C header: [`include/linux/refcount.h`](srctree/include/linux/refcount.h) + +use crate::{ + build_assert::build_assert, + sync::atomic::Atomic, + types::Opaque, // +}; + +/// Atomic reference counter. +/// +/// This type is conceptually an atomic integer, but provides saturation semantics compared to +/// normal atomic integers. Values in the negative range when viewed as a signed integer are +/// saturation (bad) values. For details about the saturation semantics, please refer to top of +/// [`include/linux/refcount.h`](srctree/include/linux/refcount.h). +/// +/// Wraps the kernel's C `refcount_t`. +#[repr(transparent)] +pub struct Refcount(Opaque<bindings::refcount_t>); + +impl Refcount { + /// Construct a new [`Refcount`] from an initial value. + /// + /// The initial value should be non-saturated. + // Always inline to optimize out error path of `build_assert`. + #[inline(always)] + pub fn new(value: i32) -> Self { + build_assert!(value >= 0, "initial value saturated"); + // SAFETY: There are no safety requirements for this FFI call. + Self(Opaque::new(unsafe { bindings::REFCOUNT_INIT(value) })) + } + + #[inline] + fn as_ptr(&self) -> *mut bindings::refcount_t { + self.0.get() + } + + /// Get the underlying atomic counter that backs the refcount. + /// + /// NOTE: Usage of this function is discouraged as it can circumvent the protections offered by + /// `refcount.h`. If there is no way to achieve the result using APIs in `refcount.h`, then + /// this function can be used. Otherwise consider adding a binding for the required API. + #[inline] + pub fn as_atomic(&self) -> &Atomic<i32> { + let ptr = self.0.get().cast(); + // SAFETY: `refcount_t` is a transparent wrapper of `atomic_t`, which is an atomic 32-bit + // integer that is layout-wise compatible with `Atomic<i32>`. All values are valid for + // `refcount_t`, despite some of the values being considered saturated and "bad". + unsafe { &*ptr } + } + + /// Set a refcount's value. + #[inline] + pub fn set(&self, value: i32) { + // SAFETY: `self.as_ptr()` is valid. + unsafe { bindings::refcount_set(self.as_ptr(), value) } + } + + /// Increment a refcount. + /// + /// It will saturate if overflows and `WARN`. It will also `WARN` if the refcount is 0, as this + /// represents a possible use-after-free condition. + /// + /// Provides no memory ordering, it is assumed that caller already has a reference on the + /// object. + #[inline] + pub fn inc(&self) { + // SAFETY: self is valid. + unsafe { bindings::refcount_inc(self.as_ptr()) } + } + + /// Decrement a refcount. + /// + /// It will `WARN` on underflow and fail to decrement when saturated. + /// + /// Provides release memory ordering, such that prior loads and stores are done + /// before. + #[inline] + pub fn dec(&self) { + // SAFETY: `self.as_ptr()` is valid. + unsafe { bindings::refcount_dec(self.as_ptr()) } + } + + /// Decrement a refcount and test if it is 0. + /// + /// It will `WARN` on underflow and fail to decrement when saturated. + /// + /// Provides release memory ordering, such that prior loads and stores are done + /// before, and provides an acquire ordering on success such that memory deallocation + /// must come after. + /// + /// Returns true if the resulting refcount is 0, false otherwise. + /// + /// # Notes + /// + /// A common pattern of using `Refcount` is to free memory when the reference count reaches + /// zero. This means that the reference to `Refcount` could become invalid after calling this + /// function. This is fine as long as the reference to `Refcount` is no longer used when this + /// function returns `false`. It is not necessary to use raw pointers in this scenario, see + /// <https://github.com/rust-lang/rust/issues/55005>. + #[inline] + #[must_use = "use `dec` instead if you do not need to test if it is 0"] + pub fn dec_and_test(&self) -> bool { + // SAFETY: `self.as_ptr()` is valid. + unsafe { bindings::refcount_dec_and_test(self.as_ptr()) } + } +} + +// SAFETY: `refcount_t` is thread-safe. +unsafe impl Send for Refcount {} + +// SAFETY: `refcount_t` is thread-safe. +unsafe impl Sync for Refcount {} diff --git a/rust/kernel/sync/set_once.rs b/rust/kernel/sync/set_once.rs new file mode 100644 index 000000000000..139cef05e935 --- /dev/null +++ b/rust/kernel/sync/set_once.rs @@ -0,0 +1,133 @@ +// SPDX-License-Identifier: GPL-2.0 + +//! A container that can be initialized at most once. + +use super::atomic::{ + ordering::{Acquire, Relaxed, Release}, + Atomic, +}; +use core::{cell::UnsafeCell, mem::MaybeUninit}; + +/// A container that can be populated at most once. Thread safe. +/// +/// Once the a [`SetOnce`] is populated, it remains populated by the same object for the +/// lifetime `Self`. +/// +/// # Invariants +/// +/// - `init` may only increase in value. +/// - `init` may only assume values in the range `0..=2`. +/// - `init == 0` if and only if `value` is uninitialized. +/// - `init == 1` if and only if there is exactly one thread with exclusive +/// access to `self.value`. +/// - `init == 2` if and only if `value` is initialized and valid for shared +/// access. +/// +/// # Example +/// +/// ``` +/// # use kernel::sync::SetOnce; +/// let value = SetOnce::new(); +/// assert_eq!(None, value.as_ref()); +/// +/// let status = value.populate(42u8); +/// assert_eq!(true, status); +/// assert_eq!(Some(&42u8), value.as_ref()); +/// assert_eq!(Some(42u8), value.copy()); +/// +/// let status = value.populate(101u8); +/// assert_eq!(false, status); +/// assert_eq!(Some(&42u8), value.as_ref()); +/// assert_eq!(Some(42u8), value.copy()); +/// ``` +pub struct SetOnce<T> { + init: Atomic<u32>, + value: UnsafeCell<MaybeUninit<T>>, +} + +impl<T> Default for SetOnce<T> { + fn default() -> Self { + Self::new() + } +} + +impl<T> SetOnce<T> { + /// Create a new [`SetOnce`]. + /// + /// The returned instance will be empty. + pub const fn new() -> Self { + // INVARIANT: The container is empty and we initialize `init` to `0`. + Self { + value: UnsafeCell::new(MaybeUninit::uninit()), + init: Atomic::new(0), + } + } + + /// Get a reference to the contained object. + /// + /// Returns [`None`] if this [`SetOnce`] is empty. + pub fn as_ref(&self) -> Option<&T> { + if self.init.load(Acquire) == 2 { + // SAFETY: By the type invariants of `Self`, `self.init == 2` means that `self.value` + // is initialized and valid for shared access. + Some(unsafe { &*self.value.get().cast() }) + } else { + None + } + } + + /// Populate the [`SetOnce`]. + /// + /// Returns `true` if the [`SetOnce`] was successfully populated. + pub fn populate(&self, value: T) -> bool { + // INVARIANT: If the swap succeeds: + // - We increase `init`. + // - We write the valid value `1` to `init`. + // - Only one thread can succeed in this write, so we have exclusive access after the + // write. + if let Ok(0) = self.init.cmpxchg(0, 1, Relaxed) { + // SAFETY: By the type invariants of `Self`, the fact that we succeeded in writing `1` + // to `self.init` means we obtained exclusive access to `self.value`. + unsafe { core::ptr::write(self.value.get().cast(), value) }; + // INVARIANT: + // - We increase `init`. + // - We write the valid value `2` to `init`. + // - We release our exclusive access to `self.value` and it is now valid for shared + // access. + self.init.store(2, Release); + true + } else { + false + } + } + + /// Get a copy of the contained object. + /// + /// Returns [`None`] if the [`SetOnce`] is empty. + pub fn copy(&self) -> Option<T> + where + T: Copy, + { + self.as_ref().copied() + } +} + +impl<T> Drop for SetOnce<T> { + fn drop(&mut self) { + if *self.init.get_mut() == 2 { + let value = self.value.get_mut(); + // SAFETY: By the type invariants of `Self`, `self.init == 2` means that `self.value` + // contains a valid value. We have exclusive access, as we hold a `mut` reference to + // `self`. + unsafe { value.assume_init_drop() }; + } + } +} + +// SAFETY: `SetOnce` can be transferred across thread boundaries iff the data it contains can. +unsafe impl<T: Send> Send for SetOnce<T> {} + +// SAFETY: `SetOnce` synchronises access to the inner value via atomic operations, +// so shared references are safe when `T: Sync`. Since the inner `T` may be dropped +// on any thread, we also require `T: Send`. +unsafe impl<T: Send + Sync> Sync for SetOnce<T> {} |