3 //! ## The threading model
5 //! An executing Rust program consists of a collection of native OS threads,
6 //! each with their own stack and local state. Threads can be named, and
7 //! provide some built-in support for low-level synchronization.
9 //! Communication between threads can be done through
10 //! [channels], Rust's message-passing types, along with [other forms of thread
11 //! synchronization](../../std/sync/index.html) and shared-memory data
12 //! structures. In particular, types that are guaranteed to be
13 //! threadsafe are easily shared between threads using the
14 //! atomically-reference-counted container, [`Arc`].
16 //! Fatal logic errors in Rust cause *thread panic*, during which
17 //! a thread will unwind the stack, running destructors and freeing
18 //! owned resources. While not meant as a 'try/catch' mechanism, panics
19 //! in Rust can nonetheless be caught (unless compiling with `panic=abort`) with
20 //! [`catch_unwind`](../../std/panic/fn.catch_unwind.html) and recovered
21 //! from, or alternatively be resumed with
22 //! [`resume_unwind`](../../std/panic/fn.resume_unwind.html). If the panic
23 //! is not caught the thread will exit, but the panic may optionally be
24 //! detected from a different thread with [`join`]. If the main thread panics
25 //! without the panic being caught, the application will exit with a
26 //! non-zero exit code.
28 //! When the main thread of a Rust program terminates, the entire program shuts
29 //! down, even if other threads are still running. However, this module provides
30 //! convenient facilities for automatically waiting for the termination of a
31 //! thread (i.e., join).
33 //! ## Spawning a thread
35 //! A new thread can be spawned using the [`thread::spawn`][`spawn`] function:
40 //! thread::spawn(move || {
45 //! In this example, the spawned thread is "detached," which means that there is
46 //! no way for the program to learn when the spawned thread completes or otherwise
49 //! To learn when a thread completes, it is necessary to capture the [`JoinHandle`]
50 //! object that is returned by the call to [`spawn`], which provides
51 //! a `join` method that allows the caller to wait for the completion of the
57 //! let thread_join_handle = thread::spawn(move || {
61 //! let res = thread_join_handle.join();
64 //! The [`join`] method returns a [`thread::Result`] containing [`Ok`] of the final
65 //! value produced by the spawned thread, or [`Err`] of the value given to
66 //! a call to [`panic!`] if the thread panicked.
68 //! Note that there is no parent/child relationship between a thread that spawns a
69 //! new thread and the thread being spawned. In particular, the spawned thread may or
70 //! may not outlive the spawning thread, unless the spawning thread is the main thread.
72 //! ## Configuring threads
74 //! A new thread can be configured before it is spawned via the [`Builder`] type,
75 //! which currently allows you to set the name and stack size for the thread:
78 //! # #![allow(unused_must_use)]
81 //! thread::Builder::new().name("thread1".to_string()).spawn(move || {
82 //! println!("Hello, world!");
86 //! ## The `Thread` type
88 //! Threads are represented via the [`Thread`] type, which you can get in one of
91 //! * By spawning a new thread, e.g., using the [`thread::spawn`][`spawn`]
92 //! function, and calling [`thread`][`JoinHandle::thread`] on the [`JoinHandle`].
93 //! * By requesting the current thread, using the [`thread::current`] function.
95 //! The [`thread::current`] function is available even for threads not spawned
96 //! by the APIs of this module.
98 //! ## Thread-local storage
100 //! This module also provides an implementation of thread-local storage for Rust
101 //! programs. Thread-local storage is a method of storing data into a global
102 //! variable that each thread in the program will have its own copy of.
103 //! Threads do not share this data, so accesses do not need to be synchronized.
105 //! A thread-local key owns the value it contains and will destroy the value when the
106 //! thread exits. It is created with the [`thread_local!`] macro and can contain any
107 //! value that is `'static` (no borrowed pointers). It provides an accessor function,
108 //! [`with`], that yields a shared reference to the value to the specified
109 //! closure. Thread-local keys allow only shared access to values, as there would be no
110 //! way to guarantee uniqueness if mutable borrows were allowed. Most values
111 //! will want to make use of some form of **interior mutability** through the
112 //! [`Cell`] or [`RefCell`] types.
114 //! ## Naming threads
116 //! Threads are able to have associated names for identification purposes. By default, spawned
117 //! threads are unnamed. To specify a name for a thread, build the thread with [`Builder`] and pass
118 //! the desired thread name to [`Builder::name`]. To retrieve the thread name from within the
119 //! thread, use [`Thread::name`]. A couple examples of where the name of a thread gets used:
121 //! * If a panic occurs in a named thread, the thread name will be printed in the panic message.
122 //! * The thread name is provided to the OS where applicable (e.g., `pthread_setname_np` in
123 //! unix-like platforms).
127 //! The default stack size for spawned threads is 2 MiB, though this particular stack size is
128 //! subject to change in the future. There are two ways to manually specify the stack size for
131 //! * Build the thread with [`Builder`] and pass the desired stack size to [`Builder::stack_size`].
132 //! * Set the `RUST_MIN_STACK` environment variable to an integer representing the desired stack
133 //! size (in bytes). Note that setting [`Builder::stack_size`] will override this.
135 //! Note that the stack size of the main thread is *not* determined by Rust.
137 //! [channels]: crate::sync::mpsc
138 //! [`join`]: JoinHandle::join
139 //! [`Result`]: crate::result::Result
140 //! [`Ok`]: crate::result::Result::Ok
141 //! [`Err`]: crate::result::Result::Err
142 //! [`thread::current`]: current
143 //! [`thread::Result`]: Result
144 //! [`unpark`]: Thread::unpark
145 //! [`thread::park_timeout`]: park_timeout
146 //! [`Cell`]: crate::cell::Cell
147 //! [`RefCell`]: crate::cell::RefCell
148 //! [`with`]: LocalKey::with
150 #![stable(feature = "rust1", since = "1.0.0")]
151 #![deny(unsafe_op_in_unsafe_fn)]
153 #[cfg(all(test, not(target_os = "emscripten")))]
157 use crate::cell::UnsafeCell;
158 use crate::ffi::{CStr, CString};
162 use crate::num::NonZeroU64;
163 use crate::num::NonZeroUsize;
165 use crate::panicking;
167 use crate::sync::Arc;
168 use crate::sys::thread as imp;
169 use crate::sys_common::mutex;
170 use crate::sys_common::thread;
171 use crate::sys_common::thread_info;
172 use crate::sys_common::thread_parker::Parker;
173 use crate::sys_common::{AsInner, IntoInner};
174 use crate::time::Duration;
176 ////////////////////////////////////////////////////////////////////////////////
177 // Thread-local storage
178 ////////////////////////////////////////////////////////////////////////////////
183 #[unstable(feature = "scoped_threads", issue = "93203")]
186 #[unstable(feature = "scoped_threads", issue = "93203")]
187 pub use scoped::{scope, Scope, ScopedJoinHandle};
189 #[stable(feature = "rust1", since = "1.0.0")]
190 pub use self::local::{AccessError, LocalKey};
192 // The types used by the thread_local! macro to access TLS keys. Note that there
193 // are two types, the "OS" type and the "fast" type. The OS thread local key
194 // type is accessed via platform-specific API calls and is slow, while the fast
195 // key type is accessed via code generated via LLVM, where TLS keys are set up
196 // by the elf linker. Note that the OS TLS type is always available: on macOS
197 // the standard library is compiled with support for older platform versions
198 // where fast TLS was not available; end-user code is compiled with fast TLS
199 // where available, but both are needed.
201 #[unstable(feature = "libstd_thread_internals", issue = "none")]
202 #[cfg(target_thread_local)]
204 pub use self::local::fast::Key as __FastLocalKeyInner;
205 #[unstable(feature = "libstd_thread_internals", issue = "none")]
207 pub use self::local::os::Key as __OsLocalKeyInner;
208 #[unstable(feature = "libstd_thread_internals", issue = "none")]
209 #[cfg(all(target_family = "wasm", not(target_feature = "atomics")))]
211 pub use self::local::statik::Key as __StaticLocalKeyInner;
213 ////////////////////////////////////////////////////////////////////////////////
215 ////////////////////////////////////////////////////////////////////////////////
217 /// Thread factory, which can be used in order to configure the properties of
220 /// Methods can be chained on it in order to configure it.
222 /// The two configurations available are:
224 /// - [`name`]: specifies an [associated name for the thread][naming-threads]
225 /// - [`stack_size`]: specifies the [desired stack size for the thread][stack-size]
227 /// The [`spawn`] method will take ownership of the builder and create an
228 /// [`io::Result`] to the thread handle with the given configuration.
230 /// The [`thread::spawn`] free function uses a `Builder` with default
231 /// configuration and [`unwrap`]s its return value.
233 /// You may want to use [`spawn`] instead of [`thread::spawn`], when you want
234 /// to recover from a failure to launch a thread, indeed the free function will
235 /// panic where the `Builder` method will return a [`io::Result`].
242 /// let builder = thread::Builder::new();
244 /// let handler = builder.spawn(|| {
248 /// handler.join().unwrap();
251 /// [`stack_size`]: Builder::stack_size
252 /// [`name`]: Builder::name
253 /// [`spawn`]: Builder::spawn
254 /// [`thread::spawn`]: spawn
255 /// [`io::Result`]: crate::io::Result
256 /// [`unwrap`]: crate::result::Result::unwrap
257 /// [naming-threads]: ./index.html#naming-threads
258 /// [stack-size]: ./index.html#stack-size
259 #[must_use = "must eventually spawn the thread"]
260 #[stable(feature = "rust1", since = "1.0.0")]
263 // A name for the thread-to-be, for identification in panic messages
264 name: Option<String>,
265 // The size of the stack for the spawned thread in bytes
266 stack_size: Option<usize>,
270 /// Generates the base configuration for spawning a thread, from which
271 /// configuration methods can be chained.
278 /// let builder = thread::Builder::new()
279 /// .name("foo".into())
280 /// .stack_size(32 * 1024);
282 /// let handler = builder.spawn(|| {
286 /// handler.join().unwrap();
288 #[stable(feature = "rust1", since = "1.0.0")]
289 pub fn new() -> Builder {
290 Builder { name: None, stack_size: None }
293 /// Names the thread-to-be. Currently the name is used for identification
294 /// only in panic messages.
296 /// The name must not contain null bytes (`\0`).
298 /// For more information about named threads, see
299 /// [this module-level documentation][naming-threads].
306 /// let builder = thread::Builder::new()
307 /// .name("foo".into());
309 /// let handler = builder.spawn(|| {
310 /// assert_eq!(thread::current().name(), Some("foo"))
313 /// handler.join().unwrap();
316 /// [naming-threads]: ./index.html#naming-threads
317 #[stable(feature = "rust1", since = "1.0.0")]
318 pub fn name(mut self, name: String) -> Builder {
319 self.name = Some(name);
323 /// Sets the size of the stack (in bytes) for the new thread.
325 /// The actual stack size may be greater than this value if
326 /// the platform specifies a minimal stack size.
328 /// For more information about the stack size for threads, see
329 /// [this module-level documentation][stack-size].
336 /// let builder = thread::Builder::new().stack_size(32 * 1024);
339 /// [stack-size]: ./index.html#stack-size
340 #[stable(feature = "rust1", since = "1.0.0")]
341 pub fn stack_size(mut self, size: usize) -> Builder {
342 self.stack_size = Some(size);
346 /// Spawns a new thread by taking ownership of the `Builder`, and returns an
347 /// [`io::Result`] to its [`JoinHandle`].
349 /// The spawned thread may outlive the caller (unless the caller thread
350 /// is the main thread; the whole process is terminated when the main
351 /// thread finishes). The join handle can be used to block on
352 /// termination of the spawned thread, including recovering its panics.
354 /// For a more complete documentation see [`thread::spawn`][`spawn`].
358 /// Unlike the [`spawn`] free function, this method yields an
359 /// [`io::Result`] to capture any failure to create the thread at
362 /// [`io::Result`]: crate::io::Result
366 /// Panics if a thread name was set and it contained null bytes.
373 /// let builder = thread::Builder::new();
375 /// let handler = builder.spawn(|| {
379 /// handler.join().unwrap();
381 #[stable(feature = "rust1", since = "1.0.0")]
382 pub fn spawn<F, T>(self, f: F) -> io::Result<JoinHandle<T>>
388 unsafe { self.spawn_unchecked(f) }
391 /// Spawns a new thread without any lifetime restrictions by taking ownership
392 /// of the `Builder`, and returns an [`io::Result`] to its [`JoinHandle`].
394 /// The spawned thread may outlive the caller (unless the caller thread
395 /// is the main thread; the whole process is terminated when the main
396 /// thread finishes). The join handle can be used to block on
397 /// termination of the spawned thread, including recovering its panics.
399 /// This method is identical to [`thread::Builder::spawn`][`Builder::spawn`],
400 /// except for the relaxed lifetime bounds, which render it unsafe.
401 /// For a more complete documentation see [`thread::spawn`][`spawn`].
405 /// Unlike the [`spawn`] free function, this method yields an
406 /// [`io::Result`] to capture any failure to create the thread at
411 /// Panics if a thread name was set and it contained null bytes.
415 /// The caller has to ensure that the spawned thread does not outlive any
416 /// references in the supplied thread closure and its return type.
417 /// This can be guaranteed in two ways:
419 /// - ensure that [`join`][`JoinHandle::join`] is called before any referenced
421 /// - use only types with `'static` lifetime bounds, i.e., those with no or only
422 /// `'static` references (both [`thread::Builder::spawn`][`Builder::spawn`]
423 /// and [`thread::spawn`][`spawn`] enforce this property statically)
428 /// #![feature(thread_spawn_unchecked)]
431 /// let builder = thread::Builder::new();
434 /// let thread_x = &x;
436 /// let handler = unsafe {
437 /// builder.spawn_unchecked(move || {
438 /// println!("x = {}", *thread_x);
442 /// // caller has to ensure `join()` is called, otherwise
443 /// // it is possible to access freed memory if `x` gets
444 /// // dropped before the thread closure is executed!
445 /// handler.join().unwrap();
448 /// [`io::Result`]: crate::io::Result
449 #[unstable(feature = "thread_spawn_unchecked", issue = "55132")]
450 pub unsafe fn spawn_unchecked<'a, F, T>(self, f: F) -> io::Result<JoinHandle<T>>
456 Ok(JoinHandle(unsafe { self.spawn_unchecked_(f, None) }?))
459 unsafe fn spawn_unchecked_<'a, 'scope, F, T>(
462 scope_data: Option<&'scope scoped::ScopeData>,
463 ) -> io::Result<JoinInner<'scope, T>>
470 let Builder { name, stack_size } = self;
472 let stack_size = stack_size.unwrap_or_else(thread::min_stack);
474 let my_thread = Thread::new(name.map(|name| {
475 CString::new(name).expect("thread name may not contain interior null bytes")
477 let their_thread = my_thread.clone();
479 let my_packet: Arc<Packet<'scope, T>> =
480 Arc::new(Packet { scope: scope_data, result: UnsafeCell::new(None) });
481 let their_packet = my_packet.clone();
483 let output_capture = crate::io::set_output_capture(None);
484 crate::io::set_output_capture(output_capture.clone());
487 if let Some(name) = their_thread.cname() {
488 imp::Thread::set_name(name);
491 crate::io::set_output_capture(output_capture);
493 // SAFETY: the stack guard passed is the one for the current thread.
494 // This means the current thread's stack and the new thread's stack
495 // are properly set and protected from each other.
496 thread_info::set(unsafe { imp::guard::current() }, their_thread);
497 let try_result = panic::catch_unwind(panic::AssertUnwindSafe(|| {
498 crate::sys_common::backtrace::__rust_begin_short_backtrace(f)
500 // SAFETY: `their_packet` as been built just above and moved by the
501 // closure (it is an Arc<...>) and `my_packet` will be stored in the
502 // same `JoinInner` as this closure meaning the mutation will be
503 // safe (not modify it and affect a value far away).
504 unsafe { *their_packet.result.get() = Some(try_result) };
507 if let Some(scope_data) = scope_data {
508 scope_data.increment_num_running_threads();
514 // `imp::Thread::new` takes a closure with a `'static` lifetime, since it's passed
515 // through FFI or otherwise used with low-level threading primitives that have no
516 // notion of or way to enforce lifetimes.
518 // As mentioned in the `Safety` section of this function's documentation, the caller of
519 // this function needs to guarantee that the passed-in lifetime is sufficiently long
520 // for the lifetime of the thread.
522 // Similarly, the `sys` implementation must guarantee that no references to the closure
523 // exist after the thread has terminated, which is signaled by `Thread::join`
528 mem::transmute::<Box<dyn FnOnce() + 'a>, Box<dyn FnOnce() + 'static>>(
539 ////////////////////////////////////////////////////////////////////////////////
541 ////////////////////////////////////////////////////////////////////////////////
543 /// Spawns a new thread, returning a [`JoinHandle`] for it.
545 /// The join handle provides a [`join`] method that can be used to join the spawned
546 /// thread. If the spawned thread panics, [`join`] will return an [`Err`] containing
547 /// the argument given to [`panic!`].
549 /// If the join handle is dropped, the spawned thread will implicitly be *detached*.
550 /// In this case, the spawned thread may no longer be joined.
551 /// (It is the responsibility of the program to either eventually join threads it
552 /// creates or detach them; otherwise, a resource leak will result.)
554 /// This call will create a thread using default parameters of [`Builder`], if you
555 /// want to specify the stack size or the name of the thread, use this API
558 /// As you can see in the signature of `spawn` there are two constraints on
559 /// both the closure given to `spawn` and its return value, let's explain them:
561 /// - The `'static` constraint means that the closure and its return value
562 /// must have a lifetime of the whole program execution. The reason for this
563 /// is that threads can outlive the lifetime they have been created in.
565 /// Indeed if the thread, and by extension its return value, can outlive their
566 /// caller, we need to make sure that they will be valid afterwards, and since
567 /// we *can't* know when it will return we need to have them valid as long as
568 /// possible, that is until the end of the program, hence the `'static`
570 /// - The [`Send`] constraint is because the closure will need to be passed
571 /// *by value* from the thread where it is spawned to the new thread. Its
572 /// return value will need to be passed from the new thread to the thread
573 /// where it is `join`ed.
574 /// As a reminder, the [`Send`] marker trait expresses that it is safe to be
575 /// passed from thread to thread. [`Sync`] expresses that it is safe to have a
576 /// reference be passed from thread to thread.
580 /// Panics if the OS fails to create a thread; use [`Builder::spawn`]
581 /// to recover from such errors.
585 /// Creating a thread.
590 /// let handler = thread::spawn(|| {
594 /// handler.join().unwrap();
597 /// As mentioned in the module documentation, threads are usually made to
598 /// communicate using [`channels`], here is how it usually looks.
600 /// This example also shows how to use `move`, in order to give ownership
601 /// of values to a thread.
605 /// use std::sync::mpsc::channel;
607 /// let (tx, rx) = channel();
609 /// let sender = thread::spawn(move || {
610 /// tx.send("Hello, thread".to_owned())
611 /// .expect("Unable to send on channel");
614 /// let receiver = thread::spawn(move || {
615 /// let value = rx.recv().expect("Unable to receive from channel");
616 /// println!("{}", value);
619 /// sender.join().expect("The sender thread has panicked");
620 /// receiver.join().expect("The receiver thread has panicked");
623 /// A thread can also return a value through its [`JoinHandle`], you can use
624 /// this to make asynchronous computations (futures might be more appropriate
630 /// let computation = thread::spawn(|| {
631 /// // Some expensive computation.
635 /// let result = computation.join().unwrap();
636 /// println!("{}", result);
639 /// [`channels`]: crate::sync::mpsc
640 /// [`join`]: JoinHandle::join
641 /// [`Err`]: crate::result::Result::Err
642 #[stable(feature = "rust1", since = "1.0.0")]
643 pub fn spawn<F, T>(f: F) -> JoinHandle<T>
649 Builder::new().spawn(f).expect("failed to spawn thread")
652 /// Gets a handle to the thread that invokes it.
656 /// Getting a handle to the current thread with `thread::current()`:
661 /// let handler = thread::Builder::new()
662 /// .name("named thread".into())
664 /// let handle = thread::current();
665 /// assert_eq!(handle.name(), Some("named thread"));
669 /// handler.join().unwrap();
672 #[stable(feature = "rust1", since = "1.0.0")]
673 pub fn current() -> Thread {
674 thread_info::current_thread().expect(
675 "use of std::thread::current() is not possible \
676 after the thread's local data has been destroyed",
680 /// Cooperatively gives up a timeslice to the OS scheduler.
682 /// This calls the underlying OS scheduler's yield primitive, signaling
683 /// that the calling thread is willing to give up its remaining timeslice
684 /// so that the OS may schedule other threads on the CPU.
686 /// A drawback of yielding in a loop is that if the OS does not have any
687 /// other ready threads to run on the current CPU, the thread will effectively
688 /// busy-wait, which wastes CPU time and energy.
690 /// Therefore, when waiting for events of interest, a programmer's first
691 /// choice should be to use synchronization devices such as [`channel`]s,
692 /// [`Condvar`]s, [`Mutex`]es or [`join`] since these primitives are
693 /// implemented in a blocking manner, giving up the CPU until the event
694 /// of interest has occurred which avoids repeated yielding.
696 /// `yield_now` should thus be used only rarely, mostly in situations where
697 /// repeated polling is required because there is no other suitable way to
698 /// learn when an event of interest has occurred.
705 /// thread::yield_now();
708 /// [`channel`]: crate::sync::mpsc
709 /// [`join`]: JoinHandle::join
710 /// [`Condvar`]: crate::sync::Condvar
711 /// [`Mutex`]: crate::sync::Mutex
712 #[stable(feature = "rust1", since = "1.0.0")]
714 imp::Thread::yield_now()
717 /// Determines whether the current thread is unwinding because of panic.
719 /// A common use of this feature is to poison shared resources when writing
720 /// unsafe code, by checking `panicking` when the `drop` is called.
722 /// This is usually not needed when writing safe code, as [`Mutex`es][Mutex]
723 /// already poison themselves when a thread panics while holding the lock.
725 /// This can also be used in multithreaded applications, in order to send a
726 /// message to other threads warning that a thread has panicked (e.g., for
727 /// monitoring purposes).
734 /// struct SomeStruct;
736 /// impl Drop for SomeStruct {
737 /// fn drop(&mut self) {
738 /// if thread::panicking() {
739 /// println!("dropped while unwinding");
741 /// println!("dropped while not unwinding");
748 /// let a = SomeStruct;
753 /// let b = SomeStruct;
758 /// [Mutex]: crate::sync::Mutex
761 #[stable(feature = "rust1", since = "1.0.0")]
762 pub fn panicking() -> bool {
763 panicking::panicking()
766 /// Puts the current thread to sleep for at least the specified amount of time.
768 /// The thread may sleep longer than the duration specified due to scheduling
769 /// specifics or platform-dependent functionality. It will never sleep less.
771 /// This function is blocking, and should not be used in `async` functions.
773 /// # Platform-specific behavior
775 /// On Unix platforms, the underlying syscall may be interrupted by a
776 /// spurious wakeup or signal handler. To ensure the sleep occurs for at least
777 /// the specified duration, this function may invoke that system call multiple
785 /// // Let's sleep for 2 seconds:
786 /// thread::sleep_ms(2000);
788 #[stable(feature = "rust1", since = "1.0.0")]
789 #[rustc_deprecated(since = "1.6.0", reason = "replaced by `std::thread::sleep`")]
790 pub fn sleep_ms(ms: u32) {
791 sleep(Duration::from_millis(ms as u64))
794 /// Puts the current thread to sleep for at least the specified amount of time.
796 /// The thread may sleep longer than the duration specified due to scheduling
797 /// specifics or platform-dependent functionality. It will never sleep less.
799 /// This function is blocking, and should not be used in `async` functions.
801 /// # Platform-specific behavior
803 /// On Unix platforms, the underlying syscall may be interrupted by a
804 /// spurious wakeup or signal handler. To ensure the sleep occurs for at least
805 /// the specified duration, this function may invoke that system call multiple
807 /// Platforms which do not support nanosecond precision for sleeping will
808 /// have `dur` rounded up to the nearest granularity of time they can sleep for.
810 /// Currently, specifying a zero duration on Unix platforms returns immediately
811 /// without invoking the underlying [`nanosleep`] syscall, whereas on Windows
812 /// platforms the underlying [`Sleep`] syscall is always invoked.
813 /// If the intention is to yield the current time-slice you may want to use
814 /// [`yield_now`] instead.
816 /// [`nanosleep`]: https://linux.die.net/man/2/nanosleep
817 /// [`Sleep`]: https://docs.microsoft.com/en-us/windows/win32/api/synchapi/nf-synchapi-sleep
822 /// use std::{thread, time};
824 /// let ten_millis = time::Duration::from_millis(10);
825 /// let now = time::Instant::now();
827 /// thread::sleep(ten_millis);
829 /// assert!(now.elapsed() >= ten_millis);
831 #[stable(feature = "thread_sleep", since = "1.4.0")]
832 pub fn sleep(dur: Duration) {
833 imp::Thread::sleep(dur)
836 /// Blocks unless or until the current thread's token is made available.
838 /// A call to `park` does not guarantee that the thread will remain parked
839 /// forever, and callers should be prepared for this possibility.
841 /// # park and unpark
843 /// Every thread is equipped with some basic low-level blocking support, via the
844 /// [`thread::park`][`park`] function and [`thread::Thread::unpark`][`unpark`]
845 /// method. [`park`] blocks the current thread, which can then be resumed from
846 /// another thread by calling the [`unpark`] method on the blocked thread's
849 /// Conceptually, each [`Thread`] handle has an associated token, which is
850 /// initially not present:
852 /// * The [`thread::park`][`park`] function blocks the current thread unless or
853 /// until the token is available for its thread handle, at which point it
854 /// atomically consumes the token. It may also return *spuriously*, without
855 /// consuming the token. [`thread::park_timeout`] does the same, but allows
856 /// specifying a maximum time to block the thread for.
858 /// * The [`unpark`] method on a [`Thread`] atomically makes the token available
859 /// if it wasn't already. Because the token is initially absent, [`unpark`]
860 /// followed by [`park`] will result in the second call returning immediately.
862 /// In other words, each [`Thread`] acts a bit like a spinlock that can be
863 /// locked and unlocked using `park` and `unpark`.
865 /// Notice that being unblocked does not imply any synchronization with someone
866 /// that unparked this thread, it could also be spurious.
867 /// For example, it would be a valid, but inefficient, implementation to make both [`park`] and
868 /// [`unpark`] return immediately without doing anything.
870 /// The API is typically used by acquiring a handle to the current thread,
871 /// placing that handle in a shared data structure so that other threads can
872 /// find it, and then `park`ing in a loop. When some desired condition is met, another
873 /// thread calls [`unpark`] on the handle.
875 /// The motivation for this design is twofold:
877 /// * It avoids the need to allocate mutexes and condvars when building new
878 /// synchronization primitives; the threads already provide basic
879 /// blocking/signaling.
881 /// * It can be implemented very efficiently on many platforms.
887 /// use std::sync::{Arc, atomic::{Ordering, AtomicBool}};
888 /// use std::time::Duration;
890 /// let flag = Arc::new(AtomicBool::new(false));
891 /// let flag2 = Arc::clone(&flag);
893 /// let parked_thread = thread::spawn(move || {
894 /// // We want to wait until the flag is set. We *could* just spin, but using
895 /// // park/unpark is more efficient.
896 /// while !flag2.load(Ordering::Acquire) {
897 /// println!("Parking thread");
899 /// // We *could* get here spuriously, i.e., way before the 10ms below are over!
900 /// // But that is no problem, we are in a loop until the flag is set anyway.
901 /// println!("Thread unparked");
903 /// println!("Flag received");
906 /// // Let some time pass for the thread to be spawned.
907 /// thread::sleep(Duration::from_millis(10));
909 /// // Set the flag, and let the thread wake up.
910 /// // There is no race condition here, if `unpark`
911 /// // happens first, `park` will return immediately.
912 /// // Hence there is no risk of a deadlock.
913 /// flag.store(true, Ordering::Release);
914 /// println!("Unpark the thread");
915 /// parked_thread.thread().unpark();
917 /// parked_thread.join().unwrap();
920 /// [`unpark`]: Thread::unpark
921 /// [`thread::park_timeout`]: park_timeout
922 #[stable(feature = "rust1", since = "1.0.0")]
924 // SAFETY: park_timeout is called on the parker owned by this thread.
926 current().inner.parker.park();
930 /// Use [`park_timeout`].
932 /// Blocks unless or until the current thread's token is made available or
933 /// the specified duration has been reached (may wake spuriously).
935 /// The semantics of this function are equivalent to [`park`] except
936 /// that the thread will be blocked for roughly no longer than `dur`. This
937 /// method should not be used for precise timing due to anomalies such as
938 /// preemption or platform differences that might not cause the maximum
939 /// amount of time waited to be precisely `ms` long.
941 /// See the [park documentation][`park`] for more detail.
942 #[stable(feature = "rust1", since = "1.0.0")]
943 #[rustc_deprecated(since = "1.6.0", reason = "replaced by `std::thread::park_timeout`")]
944 pub fn park_timeout_ms(ms: u32) {
945 park_timeout(Duration::from_millis(ms as u64))
948 /// Blocks unless or until the current thread's token is made available or
949 /// the specified duration has been reached (may wake spuriously).
951 /// The semantics of this function are equivalent to [`park`][park] except
952 /// that the thread will be blocked for roughly no longer than `dur`. This
953 /// method should not be used for precise timing due to anomalies such as
954 /// preemption or platform differences that might not cause the maximum
955 /// amount of time waited to be precisely `dur` long.
957 /// See the [park documentation][park] for more details.
959 /// # Platform-specific behavior
961 /// Platforms which do not support nanosecond precision for sleeping will have
962 /// `dur` rounded up to the nearest granularity of time they can sleep for.
966 /// Waiting for the complete expiration of the timeout:
969 /// use std::thread::park_timeout;
970 /// use std::time::{Instant, Duration};
972 /// let timeout = Duration::from_secs(2);
973 /// let beginning_park = Instant::now();
975 /// let mut timeout_remaining = timeout;
977 /// park_timeout(timeout_remaining);
978 /// let elapsed = beginning_park.elapsed();
979 /// if elapsed >= timeout {
982 /// println!("restarting park_timeout after {:?}", elapsed);
983 /// timeout_remaining = timeout - elapsed;
986 #[stable(feature = "park_timeout", since = "1.4.0")]
987 pub fn park_timeout(dur: Duration) {
988 // SAFETY: park_timeout is called on the parker owned by this thread.
990 current().inner.parker.park_timeout(dur);
994 ////////////////////////////////////////////////////////////////////////////////
996 ////////////////////////////////////////////////////////////////////////////////
998 /// A unique identifier for a running thread.
1000 /// A `ThreadId` is an opaque object that uniquely identifies each thread
1001 /// created during the lifetime of a process. `ThreadId`s are guaranteed not to
1002 /// be reused, even when a thread terminates. `ThreadId`s are under the control
1003 /// of Rust's standard library and there may not be any relationship between
1004 /// `ThreadId` and the underlying platform's notion of a thread identifier --
1005 /// the two concepts cannot, therefore, be used interchangeably. A `ThreadId`
1006 /// can be retrieved from the [`id`] method on a [`Thread`].
1011 /// use std::thread;
1013 /// let other_thread = thread::spawn(|| {
1014 /// thread::current().id()
1017 /// let other_thread_id = other_thread.join().unwrap();
1018 /// assert!(thread::current().id() != other_thread_id);
1021 /// [`id`]: Thread::id
1022 #[stable(feature = "thread_id", since = "1.19.0")]
1023 #[derive(Eq, PartialEq, Clone, Copy, Hash, Debug)]
1024 pub struct ThreadId(NonZeroU64);
1027 // Generate a new unique thread ID.
1028 fn new() -> ThreadId {
1029 // It is UB to attempt to acquire this mutex reentrantly!
1030 static GUARD: mutex::StaticMutex = mutex::StaticMutex::new();
1031 static mut COUNTER: u64 = 1;
1034 let guard = GUARD.lock();
1036 // If we somehow use up all our bits, panic so that we're not
1037 // covering up subtle bugs of IDs being reused.
1038 if COUNTER == u64::MAX {
1039 drop(guard); // in case the panic handler ends up calling `ThreadId::new()`, avoid reentrant lock acquire.
1040 panic!("failed to generate unique thread ID: bitspace exhausted");
1046 ThreadId(NonZeroU64::new(id).unwrap())
1050 /// This returns a numeric identifier for the thread identified by this
1053 /// As noted in the documentation for the type itself, it is essentially an
1054 /// opaque ID, but is guaranteed to be unique for each thread. The returned
1055 /// value is entirely opaque -- only equality testing is stable. Note that
1056 /// it is not guaranteed which values new threads will return, and this may
1057 /// change across Rust versions.
1059 #[unstable(feature = "thread_id_value", issue = "67939")]
1060 pub fn as_u64(&self) -> NonZeroU64 {
1065 ////////////////////////////////////////////////////////////////////////////////
1067 ////////////////////////////////////////////////////////////////////////////////
1069 /// The internal representation of a `Thread` handle
1071 name: Option<CString>, // Guaranteed to be UTF-8
1077 #[stable(feature = "rust1", since = "1.0.0")]
1078 /// A handle to a thread.
1080 /// Threads are represented via the `Thread` type, which you can get in one of
1083 /// * By spawning a new thread, e.g., using the [`thread::spawn`][`spawn`]
1084 /// function, and calling [`thread`][`JoinHandle::thread`] on the
1086 /// * By requesting the current thread, using the [`thread::current`] function.
1088 /// The [`thread::current`] function is available even for threads not spawned
1089 /// by the APIs of this module.
1091 /// There is usually no need to create a `Thread` struct yourself, one
1092 /// should instead use a function like `spawn` to create new threads, see the
1093 /// docs of [`Builder`] and [`spawn`] for more details.
1095 /// [`thread::current`]: current
1101 // Used only internally to construct a thread object without spawning
1102 // Panics if the name contains nuls.
1103 pub(crate) fn new(name: Option<CString>) -> Thread {
1104 Thread { inner: Arc::new(Inner { name, id: ThreadId::new(), parker: Parker::new() }) }
1107 /// Atomically makes the handle's token available if it is not already.
1109 /// Every thread is equipped with some basic low-level blocking support, via
1110 /// the [`park`][park] function and the `unpark()` method. These can be
1111 /// used as a more CPU-efficient implementation of a spinlock.
1113 /// See the [park documentation][park] for more details.
1118 /// use std::thread;
1119 /// use std::time::Duration;
1121 /// let parked_thread = thread::Builder::new()
1123 /// println!("Parking thread");
1125 /// println!("Thread unparked");
1129 /// // Let some time pass for the thread to be spawned.
1130 /// thread::sleep(Duration::from_millis(10));
1132 /// println!("Unpark the thread");
1133 /// parked_thread.thread().unpark();
1135 /// parked_thread.join().unwrap();
1137 #[stable(feature = "rust1", since = "1.0.0")]
1139 pub fn unpark(&self) {
1140 self.inner.parker.unpark();
1143 /// Gets the thread's unique identifier.
1148 /// use std::thread;
1150 /// let other_thread = thread::spawn(|| {
1151 /// thread::current().id()
1154 /// let other_thread_id = other_thread.join().unwrap();
1155 /// assert!(thread::current().id() != other_thread_id);
1157 #[stable(feature = "thread_id", since = "1.19.0")]
1159 pub fn id(&self) -> ThreadId {
1163 /// Gets the thread's name.
1165 /// For more information about named threads, see
1166 /// [this module-level documentation][naming-threads].
1170 /// Threads by default have no name specified:
1173 /// use std::thread;
1175 /// let builder = thread::Builder::new();
1177 /// let handler = builder.spawn(|| {
1178 /// assert!(thread::current().name().is_none());
1181 /// handler.join().unwrap();
1184 /// Thread with a specified name:
1187 /// use std::thread;
1189 /// let builder = thread::Builder::new()
1190 /// .name("foo".into());
1192 /// let handler = builder.spawn(|| {
1193 /// assert_eq!(thread::current().name(), Some("foo"))
1196 /// handler.join().unwrap();
1199 /// [naming-threads]: ./index.html#naming-threads
1200 #[stable(feature = "rust1", since = "1.0.0")]
1202 pub fn name(&self) -> Option<&str> {
1203 self.cname().map(|s| unsafe { str::from_utf8_unchecked(s.to_bytes()) })
1206 fn cname(&self) -> Option<&CStr> {
1207 self.inner.name.as_deref()
1211 #[stable(feature = "rust1", since = "1.0.0")]
1212 impl fmt::Debug for Thread {
1213 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1214 f.debug_struct("Thread")
1215 .field("id", &self.id())
1216 .field("name", &self.name())
1217 .finish_non_exhaustive()
1221 ////////////////////////////////////////////////////////////////////////////////
1223 ////////////////////////////////////////////////////////////////////////////////
1225 /// A specialized [`Result`] type for threads.
1227 /// Indicates the manner in which a thread exited.
1229 /// The value contained in the `Result::Err` variant
1230 /// is the value the thread panicked with;
1231 /// that is, the argument the `panic!` macro was called with.
1232 /// Unlike with normal errors, this value doesn't implement
1233 /// the [`Error`](crate::error::Error) trait.
1235 /// Thus, a sensible way to handle a thread panic is to either:
1237 /// 1. propagate the panic with [`std::panic::resume_unwind`]
1238 /// 2. or in case the thread is intended to be a subsystem boundary
1239 /// that is supposed to isolate system-level failures,
1240 /// match on the `Err` variant and handle the panic in an appropriate way
1242 /// A thread that completes without panicking is considered to exit successfully.
1246 /// Matching on the result of a joined thread:
1249 /// use std::{fs, thread, panic};
1251 /// fn copy_in_thread() -> thread::Result<()> {
1252 /// thread::spawn(|| {
1253 /// fs::copy("foo.txt", "bar.txt").unwrap();
1258 /// match copy_in_thread() {
1259 /// Ok(_) => println!("copy succeeded"),
1260 /// Err(e) => panic::resume_unwind(e),
1265 /// [`Result`]: crate::result::Result
1266 /// [`std::panic::resume_unwind`]: crate::panic::resume_unwind
1267 #[stable(feature = "rust1", since = "1.0.0")]
1268 pub type Result<T> = crate::result::Result<T, Box<dyn Any + Send + 'static>>;
1270 // This packet is used to communicate the return value between the spawned
1271 // thread and the rest of the program. It is shared through an `Arc` and
1272 // there's no need for a mutex here because synchronization happens with `join()`
1273 // (the caller will never read this packet until the thread has exited).
1275 // An Arc to the packet is stored into a `JoinInner` which in turns is placed
1277 struct Packet<'scope, T> {
1278 scope: Option<&'scope scoped::ScopeData>,
1279 result: UnsafeCell<Option<Result<T>>>,
1282 // Due to the usage of `UnsafeCell` we need to manually implement Sync.
1283 // The type `T` should already always be Send (otherwise the thread could not
1284 // have been created) and the Packet is Sync because all access to the
1285 // `UnsafeCell` synchronized (by the `join()` boundary), and `ScopeData` is Sync.
1286 unsafe impl<'scope, T: Sync> Sync for Packet<'scope, T> {}
1288 impl<'scope, T> Drop for Packet<'scope, T> {
1289 fn drop(&mut self) {
1290 // Book-keeping so the scope knows when it's done.
1291 if let Some(scope) = self.scope {
1292 // If this packet was for a thread that ran in a scope, the thread
1293 // panicked, and nobody consumed the panic payload, we make sure
1294 // the scope function will panic.
1295 let unhandled_panic = matches!(self.result.get_mut(), Some(Err(_)));
1296 scope.decrement_num_running_threads(unhandled_panic);
1301 /// Inner representation for JoinHandle
1302 struct JoinInner<'scope, T> {
1303 native: imp::Thread,
1305 packet: Arc<Packet<'scope, T>>,
1308 impl<'scope, T> JoinInner<'scope, T> {
1309 fn join(mut self) -> Result<T> {
1311 Arc::get_mut(&mut self.packet).unwrap().result.get_mut().take().unwrap()
1315 /// An owned permission to join on a thread (block on its termination).
1317 /// A `JoinHandle` *detaches* the associated thread when it is dropped, which
1318 /// means that there is no longer any handle to the thread and no way to `join`
1321 /// Due to platform restrictions, it is not possible to [`Clone`] this
1322 /// handle: the ability to join a thread is a uniquely-owned permission.
1324 /// This `struct` is created by the [`thread::spawn`] function and the
1325 /// [`thread::Builder::spawn`] method.
1329 /// Creation from [`thread::spawn`]:
1332 /// use std::thread;
1334 /// let join_handle: thread::JoinHandle<_> = thread::spawn(|| {
1335 /// // some work here
1339 /// Creation from [`thread::Builder::spawn`]:
1342 /// use std::thread;
1344 /// let builder = thread::Builder::new();
1346 /// let join_handle: thread::JoinHandle<_> = builder.spawn(|| {
1347 /// // some work here
1351 /// A thread being detached and outliving the thread that spawned it:
1354 /// use std::thread;
1355 /// use std::time::Duration;
1357 /// let original_thread = thread::spawn(|| {
1358 /// let _detached_thread = thread::spawn(|| {
1359 /// // Here we sleep to make sure that the first thread returns before.
1360 /// thread::sleep(Duration::from_millis(10));
1361 /// // This will be called, even though the JoinHandle is dropped.
1362 /// println!("♫ Still alive ♫");
1366 /// original_thread.join().expect("The thread being joined has panicked");
1367 /// println!("Original thread is joined.");
1369 /// // We make sure that the new thread has time to run, before the main
1370 /// // thread returns.
1372 /// thread::sleep(Duration::from_millis(1000));
1375 /// [`thread::Builder::spawn`]: Builder::spawn
1376 /// [`thread::spawn`]: spawn
1377 #[stable(feature = "rust1", since = "1.0.0")]
1378 pub struct JoinHandle<T>(JoinInner<'static, T>);
1380 #[stable(feature = "joinhandle_impl_send_sync", since = "1.29.0")]
1381 unsafe impl<T> Send for JoinHandle<T> {}
1382 #[stable(feature = "joinhandle_impl_send_sync", since = "1.29.0")]
1383 unsafe impl<T> Sync for JoinHandle<T> {}
1385 impl<T> JoinHandle<T> {
1386 /// Extracts a handle to the underlying thread.
1391 /// use std::thread;
1393 /// let builder = thread::Builder::new();
1395 /// let join_handle: thread::JoinHandle<_> = builder.spawn(|| {
1396 /// // some work here
1399 /// let thread = join_handle.thread();
1400 /// println!("thread id: {:?}", thread.id());
1402 #[stable(feature = "rust1", since = "1.0.0")]
1404 pub fn thread(&self) -> &Thread {
1408 /// Waits for the associated thread to finish.
1410 /// This function will return immediately if the associated thread has already finished.
1412 /// In terms of [atomic memory orderings], the completion of the associated
1413 /// thread synchronizes with this function returning. In other words, all
1414 /// operations performed by that thread [happen
1415 /// before](https://doc.rust-lang.org/nomicon/atomics.html#data-accesses) all
1416 /// operations that happen after `join` returns.
1418 /// If the associated thread panics, [`Err`] is returned with the parameter given
1421 /// [`Err`]: crate::result::Result::Err
1422 /// [atomic memory orderings]: crate::sync::atomic
1426 /// This function may panic on some platforms if a thread attempts to join
1427 /// itself or otherwise may create a deadlock with joining threads.
1432 /// use std::thread;
1434 /// let builder = thread::Builder::new();
1436 /// let join_handle: thread::JoinHandle<_> = builder.spawn(|| {
1437 /// // some work here
1439 /// join_handle.join().expect("Couldn't join on the associated thread");
1441 #[stable(feature = "rust1", since = "1.0.0")]
1442 pub fn join(self) -> Result<T> {
1446 /// Checks if the associated thread is still running its main function.
1448 /// This might return `false` for a brief moment after the thread's main
1449 /// function has returned, but before the thread itself has stopped running.
1450 #[unstable(feature = "thread_is_running", issue = "90470")]
1451 pub fn is_running(&self) -> bool {
1452 Arc::strong_count(&self.0.packet) > 1
1456 impl<T> AsInner<imp::Thread> for JoinHandle<T> {
1457 fn as_inner(&self) -> &imp::Thread {
1462 impl<T> IntoInner<imp::Thread> for JoinHandle<T> {
1463 fn into_inner(self) -> imp::Thread {
1468 #[stable(feature = "std_debug", since = "1.16.0")]
1469 impl<T> fmt::Debug for JoinHandle<T> {
1470 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
1471 f.debug_struct("JoinHandle").finish_non_exhaustive()
1475 fn _assert_sync_and_send() {
1476 fn _assert_both<T: Send + Sync>() {}
1477 _assert_both::<JoinHandle<()>>();
1478 _assert_both::<Thread>();
1481 /// Returns an estimate of the default amount of parallelism a program should use.
1483 /// Parallelism is a resource. A given machine provides a certain capacity for
1484 /// parallelism, i.e., a bound on the number of computations it can perform
1485 /// simultaneously. This number often corresponds to the amount of CPUs a
1486 /// computer has, but it may diverge in various cases.
1488 /// Host environments such as VMs or container orchestrators may want to
1489 /// restrict the amount of parallelism made available to programs in them. This
1490 /// is often done to limit the potential impact of (unintentionally)
1491 /// resource-intensive programs on other programs running on the same machine.
1495 /// The purpose of this API is to provide an easy and portable way to query
1496 /// the default amount of parallelism the program should use. Among other things it
1497 /// does not expose information on NUMA regions, does not account for
1498 /// differences in (co)processor capabilities or current system load,
1499 /// and will not modify the program's global state in order to more accurately
1500 /// query the amount of available parallelism.
1502 /// Where both fixed steady-state and burst limits are available the steady-state
1503 /// capacity will be used to ensure more predictable latencies.
1505 /// Resource limits can be changed during the runtime of a program, therefore the value is
1506 /// not cached and instead recomputed every time this function is called. It should not be
1507 /// called from hot code.
1509 /// The value returned by this function should be considered a simplified
1510 /// approximation of the actual amount of parallelism available at any given
1511 /// time. To get a more detailed or precise overview of the amount of
1512 /// parallelism available to the program, you may wish to use
1513 /// platform-specific APIs as well. The following platform limitations currently
1514 /// apply to `available_parallelism`:
1517 /// - It may undercount the amount of parallelism available on systems with more
1518 /// than 64 logical CPUs. However, programs typically need specific support to
1519 /// take advantage of more than 64 logical CPUs, and in the absence of such
1520 /// support, the number returned by this function accurately reflects the
1521 /// number of logical CPUs the program can use by default.
1522 /// - It may overcount the amount of parallelism available on systems limited by
1523 /// process-wide affinity masks, or job object limitations.
1526 /// - It may overcount the amount of parallelism available when limited by a
1527 /// process-wide affinity mask, or when affected by cgroup limits.
1530 /// - It may overcount the amount of parallelism available when running in a VM
1531 /// with CPU usage limits (e.g. an overcommitted host).
1535 /// This function will, but is not limited to, return errors in the following
1538 /// - If the amount of parallelism is not known for the target platform.
1539 /// - If the program lacks permission to query the amount of parallelism made
1540 /// available to it.
1545 /// # #![allow(dead_code)]
1546 /// use std::{io, thread};
1548 /// fn main() -> io::Result<()> {
1549 /// let count = thread::available_parallelism()?.get();
1550 /// assert!(count >= 1_usize);
1554 #[doc(alias = "available_concurrency")] // Alias for a previous name we gave this API on unstable.
1555 #[doc(alias = "hardware_concurrency")] // Alias for C++ `std::thread::hardware_concurrency`.
1556 #[doc(alias = "num_cpus")] // Alias for a popular ecosystem crate which provides similar functionality.
1557 #[stable(feature = "available_parallelism", since = "1.59.0")]
1558 pub fn available_parallelism() -> io::Result<NonZeroUsize> {
1559 imp::available_parallelism()