diff options
Diffstat (limited to '')
209 files changed, 10542 insertions, 8652 deletions
diff --git a/library/std/src/alloc.rs b/library/std/src/alloc.rs index a05e0db3a..61c1ff578 100644 --- a/library/std/src/alloc.rs +++ b/library/std/src/alloc.rs @@ -68,7 +68,10 @@ pub use alloc_crate::alloc::*; /// The default memory allocator provided by the operating system. /// /// This is based on `malloc` on Unix platforms and `HeapAlloc` on Windows, -/// plus related functions. +/// plus related functions. However, it is not valid to mix use of the backing +/// system allocator with `System`, as this implementation may include extra +/// work, such as to serve alignment requests greater than the alignment +/// provided directly by the backing system allocator. /// /// This type implements the `GlobalAlloc` trait and Rust programs by default /// work as if they had this definition: diff --git a/library/std/src/backtrace.rs b/library/std/src/backtrace.rs index 05e9b2eb6..9cb74f951 100644 --- a/library/std/src/backtrace.rs +++ b/library/std/src/backtrace.rs @@ -9,19 +9,13 @@ //! implementing `std::error::Error`) to get a causal chain of where an error //! was generated. //! -//! > **Note**: this module is unstable and is designed in [RFC 2504], and you -//! > can learn more about its status in the [tracking issue]. -//! -//! [RFC 2504]: https://github.com/rust-lang/rfcs/blob/master/text/2504-fix-error.md -//! [tracking issue]: https://github.com/rust-lang/rust/issues/53487 -//! //! ## Accuracy //! //! Backtraces are attempted to be as accurate as possible, but no guarantees //! are provided about the exact accuracy of a backtrace. Instruction pointers, //! symbol names, filenames, line numbers, etc, may all be incorrect when -//! reported. Accuracy is attempted on a best-effort basis, however, and bugs -//! are always welcome to indicate areas of improvement! +//! reported. Accuracy is attempted on a best-effort basis, however, any bug +//! reports are always welcome to indicate areas of improvement! //! //! For most platforms a backtrace with a filename/line number requires that //! programs be compiled with debug information. Without debug information @@ -45,7 +39,7 @@ //! default. Its behavior is governed by two environment variables: //! //! * `RUST_LIB_BACKTRACE` - if this is set to `0` then `Backtrace::capture` -//! will never capture a backtrace. Any other value this is set to will enable +//! will never capture a backtrace. Any other value set will enable //! `Backtrace::capture`. //! //! * `RUST_BACKTRACE` - if `RUST_LIB_BACKTRACE` is not set, then this variable @@ -64,7 +58,7 @@ //! `RUST_LIB_BACKTRACE` or `RUST_BACKTRACE` at runtime might not actually change //! how backtraces are captured. -#![unstable(feature = "backtrace", issue = "53487")] +#![stable(feature = "backtrace", since = "1.65.0")] #[cfg(test)] mod tests; @@ -110,6 +104,7 @@ use crate::vec::Vec; /// previous point in time. In some instances the `Backtrace` type may /// internally be empty due to configuration. For more information see /// `Backtrace::capture`. +#[stable(feature = "backtrace", since = "1.65.0")] #[must_use] pub struct Backtrace { inner: Inner, @@ -117,17 +112,21 @@ pub struct Backtrace { /// The current status of a backtrace, indicating whether it was captured or /// whether it is empty for some other reason. +#[stable(feature = "backtrace", since = "1.65.0")] #[non_exhaustive] #[derive(Debug, PartialEq, Eq)] pub enum BacktraceStatus { /// Capturing a backtrace is not supported, likely because it's not /// implemented for the current platform. + #[stable(feature = "backtrace", since = "1.65.0")] Unsupported, /// Capturing a backtrace has been disabled through either the /// `RUST_LIB_BACKTRACE` or `RUST_BACKTRACE` environment variables. + #[stable(feature = "backtrace", since = "1.65.0")] Disabled, /// A backtrace has been captured and the `Backtrace` should print /// reasonable information when rendered. + #[stable(feature = "backtrace", since = "1.65.0")] Captured, } @@ -174,6 +173,7 @@ enum BytesOrWide { Wide(Vec<u16>), } +#[stable(feature = "backtrace", since = "1.65.0")] impl fmt::Debug for Backtrace { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { let capture = match &self.inner { @@ -200,6 +200,7 @@ impl fmt::Debug for Backtrace { } } +#[unstable(feature = "backtrace_frames", issue = "79676")] impl fmt::Debug for BacktraceFrame { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { let mut dbg = fmt.debug_list(); @@ -288,6 +289,7 @@ impl Backtrace { /// /// To forcibly capture a backtrace regardless of environment variables, use /// the `Backtrace::force_capture` function. + #[stable(feature = "backtrace", since = "1.65.0")] #[inline(never)] // want to make sure there's a frame here to remove pub fn capture() -> Backtrace { if !Backtrace::enabled() { @@ -306,6 +308,7 @@ impl Backtrace { /// Note that capturing a backtrace can be an expensive operation on some /// platforms, so this should be used with caution in performance-sensitive /// parts of code. + #[stable(feature = "backtrace", since = "1.65.0")] #[inline(never)] // want to make sure there's a frame here to remove pub fn force_capture() -> Backtrace { Backtrace::create(Backtrace::force_capture as usize) @@ -313,6 +316,8 @@ impl Backtrace { /// Forcibly captures a disabled backtrace, regardless of environment /// variable configuration. + #[stable(feature = "backtrace", since = "1.65.0")] + #[rustc_const_stable(feature = "backtrace", since = "1.65.0")] pub const fn disabled() -> Backtrace { Backtrace { inner: Inner::Disabled } } @@ -320,8 +325,7 @@ impl Backtrace { // Capture a backtrace which start just before the function addressed by // `ip` fn create(ip: usize) -> Backtrace { - // SAFETY: We don't attempt to lock this reentrantly. - let _lock = unsafe { lock() }; + let _lock = lock(); let mut frames = Vec::new(); let mut actual_start = None; unsafe { @@ -356,6 +360,7 @@ impl Backtrace { /// Returns the status of this backtrace, indicating whether this backtrace /// request was unsupported, disabled, or a stack trace was actually /// captured. + #[stable(feature = "backtrace", since = "1.65.0")] #[must_use] pub fn status(&self) -> BacktraceStatus { match self.inner { @@ -375,6 +380,7 @@ impl<'a> Backtrace { } } +#[stable(feature = "backtrace", since = "1.65.0")] impl fmt::Display for Backtrace { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { let capture = match &self.inner { @@ -462,8 +468,7 @@ impl Capture { // Use the global backtrace lock to synchronize this as it's a // requirement of the `backtrace` crate, and then actually resolve // everything. - // SAFETY: We don't attempt to lock this reentrantly. - let _lock = unsafe { lock() }; + let _lock = lock(); for frame in self.frames.iter_mut() { let symbols = &mut frame.symbols; let frame = match &frame.frame { diff --git a/library/std/src/collections/hash/map.rs b/library/std/src/collections/hash/map.rs index db811343f..708edc5de 100644 --- a/library/std/src/collections/hash/map.rs +++ b/library/std/src/collections/hash/map.rs @@ -9,6 +9,7 @@ use crate::borrow::Borrow; use crate::cell::Cell; use crate::collections::TryReserveError; use crate::collections::TryReserveErrorKind; +use crate::error::Error; use crate::fmt::{self, Debug}; #[allow(deprecated)] use crate::hash::{BuildHasher, Hash, Hasher, SipHasher13}; @@ -279,7 +280,8 @@ impl<K, V, S> HashMap<K, V, S> { /// ``` #[inline] #[stable(feature = "hashmap_build_hasher", since = "1.7.0")] - pub fn with_hasher(hash_builder: S) -> HashMap<K, V, S> { + #[rustc_const_unstable(feature = "const_collections_with_hasher", issue = "102575")] + pub const fn with_hasher(hash_builder: S) -> HashMap<K, V, S> { HashMap { base: base::HashMap::with_hasher(hash_builder) } } @@ -757,7 +759,7 @@ where /// Tries to reserve capacity for at least `additional` more elements to be inserted /// in the `HashMap`. The collection may reserve more space to speculatively - /// avoid frequent reallocations. After calling `reserve`, + /// avoid frequent reallocations. After calling `try_reserve`, /// capacity will be greater than or equal to `self.len() + additional` if /// it returns `Ok(())`. /// Does nothing if capacity is already sufficient. @@ -2158,6 +2160,14 @@ impl<'a, K: Debug, V: Debug> fmt::Display for OccupiedError<'a, K, V> { } } +#[unstable(feature = "map_try_insert", issue = "82766")] +impl<'a, K: fmt::Debug, V: fmt::Debug> Error for OccupiedError<'a, K, V> { + #[allow(deprecated)] + fn description(&self) -> &str { + "key already exists" + } +} + #[stable(feature = "rust1", since = "1.0.0")] impl<'a, K, V, S> IntoIterator for &'a HashMap<K, V, S> { type Item = (&'a K, &'a V); diff --git a/library/std/src/collections/hash/map/tests.rs b/library/std/src/collections/hash/map/tests.rs index 7ebc41588..65634f206 100644 --- a/library/std/src/collections/hash/map/tests.rs +++ b/library/std/src/collections/hash/map/tests.rs @@ -268,10 +268,13 @@ fn test_lots_of_insertions() { // Try this a few times to make sure we never screw up the hashmap's // internal state. - for _ in 0..10 { + let loops = if cfg!(miri) { 2 } else { 10 }; + for _ in 0..loops { assert!(m.is_empty()); - for i in 1..1001 { + let count = if cfg!(miri) { 101 } else { 1001 }; + + for i in 1..count { assert!(m.insert(i, i).is_none()); for j in 1..=i { @@ -279,42 +282,42 @@ fn test_lots_of_insertions() { assert_eq!(r, Some(&j)); } - for j in i + 1..1001 { + for j in i + 1..count { let r = m.get(&j); assert_eq!(r, None); } } - for i in 1001..2001 { + for i in count..(2 * count) { assert!(!m.contains_key(&i)); } // remove forwards - for i in 1..1001 { + for i in 1..count { assert!(m.remove(&i).is_some()); for j in 1..=i { assert!(!m.contains_key(&j)); } - for j in i + 1..1001 { + for j in i + 1..count { assert!(m.contains_key(&j)); } } - for i in 1..1001 { + for i in 1..count { assert!(!m.contains_key(&i)); } - for i in 1..1001 { + for i in 1..count { assert!(m.insert(i, i).is_none()); } // remove backwards - for i in (1..1001).rev() { + for i in (1..count).rev() { assert!(m.remove(&i).is_some()); - for j in i..1001 { + for j in i..count { assert!(!m.contains_key(&j)); } @@ -817,6 +820,7 @@ fn test_retain() { } #[test] +#[cfg_attr(miri, ignore)] // Miri does not support signalling OOM #[cfg_attr(target_os = "android", ignore)] // Android used in CI has a broken dlmalloc fn test_try_reserve() { let mut empty_bytes: HashMap<u8, u8> = HashMap::new(); @@ -1111,3 +1115,9 @@ fn from_array() { // that's a problem! let _must_not_require_type_annotation = HashMap::from([(1, 2)]); } + +#[test] +fn const_with_hasher() { + const X: HashMap<(), (), ()> = HashMap::with_hasher(()); + assert_eq!(X.len(), 0); +} diff --git a/library/std/src/collections/hash/set.rs b/library/std/src/collections/hash/set.rs index abff82788..cee884145 100644 --- a/library/std/src/collections/hash/set.rs +++ b/library/std/src/collections/hash/set.rs @@ -239,7 +239,7 @@ impl<T, S> HashSet<T, S> { /// /// If the returned iterator is dropped before being fully consumed, it /// drops the remaining elements. The returned iterator keeps a mutable - /// borrow on the vector to optimize its implementation. + /// borrow on the set to optimize its implementation. /// /// # Examples /// @@ -376,7 +376,8 @@ impl<T, S> HashSet<T, S> { /// ``` #[inline] #[stable(feature = "hashmap_build_hasher", since = "1.7.0")] - pub fn with_hasher(hasher: S) -> HashSet<T, S> { + #[rustc_const_unstable(feature = "const_collections_with_hasher", issue = "102575")] + pub const fn with_hasher(hasher: S) -> HashSet<T, S> { HashSet { base: base::HashSet::with_hasher(hasher) } } @@ -461,7 +462,7 @@ where /// Tries to reserve capacity for at least `additional` more elements to be inserted /// in the `HashSet`. The collection may reserve more space to speculatively - /// avoid frequent reallocations. After calling `reserve`, + /// avoid frequent reallocations. After calling `try_reserve`, /// capacity will be greater than or equal to `self.len() + additional` if /// it returns `Ok(())`. /// Does nothing if capacity is already sufficient. diff --git a/library/std/src/collections/hash/set/tests.rs b/library/std/src/collections/hash/set/tests.rs index 233db276b..941a0450c 100644 --- a/library/std/src/collections/hash/set/tests.rs +++ b/library/std/src/collections/hash/set/tests.rs @@ -496,3 +496,9 @@ fn from_array() { // that's a problem! let _must_not_require_type_annotation = HashSet::from([1, 2]); } + +#[test] +fn const_with_hasher() { + const X: HashSet<(), ()> = HashSet::with_hasher(()); + assert_eq!(X.len(), 0); +} diff --git a/library/std/src/env.rs b/library/std/src/env.rs index 463f71406..6eb7cbea6 100644 --- a/library/std/src/env.rs +++ b/library/std/src/env.rs @@ -603,7 +603,7 @@ pub fn home_dir() -> Option<PathBuf> { /// # Platform-specific behavior /// /// On Unix, returns the value of the `TMPDIR` environment variable if it is -/// set, otherwise for non-Android it returns `/tmp`. If Android, since there +/// set, otherwise for non-Android it returns `/tmp`. On Android, since there /// is no global temporary folder (it is usually allocated per-app), it returns /// `/data/local/tmp`. /// On Windows, the behavior is equivalent to that of [`GetTempPath2`][GetTempPath2] / diff --git a/library/std/src/error.rs b/library/std/src/error.rs index 722df119d..05f8fd8de 100644 --- a/library/std/src/error.rs +++ b/library/std/src/error.rs @@ -1,366 +1,14 @@ -//! Interfaces for working with Errors. -//! -//! # Error Handling In Rust -//! -//! The Rust language provides two complementary systems for constructing / -//! representing, reporting, propagating, reacting to, and discarding errors. -//! These responsibilities are collectively known as "error handling." The -//! components of the first system, the panic runtime and interfaces, are most -//! commonly used to represent bugs that have been detected in your program. The -//! components of the second system, `Result`, the error traits, and user -//! defined types, are used to represent anticipated runtime failure modes of -//! your program. -//! -//! ## The Panic Interfaces -//! -//! The following are the primary interfaces of the panic system and the -//! responsibilities they cover: -//! -//! * [`panic!`] and [`panic_any`] (Constructing, Propagated automatically) -//! * [`PanicInfo`] (Reporting) -//! * [`set_hook`], [`take_hook`], and [`#[panic_handler]`][panic-handler] (Reporting) -//! * [`catch_unwind`] and [`resume_unwind`] (Discarding, Propagating) -//! -//! The following are the primary interfaces of the error system and the -//! responsibilities they cover: -//! -//! * [`Result`] (Propagating, Reacting) -//! * The [`Error`] trait (Reporting) -//! * User defined types (Constructing / Representing) -//! * [`match`] and [`downcast`] (Reacting) -//! * The question mark operator ([`?`]) (Propagating) -//! * The partially stable [`Try`] traits (Propagating, Constructing) -//! * [`Termination`] (Reporting) -//! -//! ## Converting Errors into Panics -//! -//! The panic and error systems are not entirely distinct. Often times errors -//! that are anticipated runtime failures in an API might instead represent bugs -//! to a caller. For these situations the standard library provides APIs for -//! constructing panics with an `Error` as it's source. -//! -//! * [`Result::unwrap`] -//! * [`Result::expect`] -//! -//! These functions are equivalent, they either return the inner value if the -//! `Result` is `Ok` or panic if the `Result` is `Err` printing the inner error -//! as the source. The only difference between them is that with `expect` you -//! provide a panic error message to be printed alongside the source, whereas -//! `unwrap` has a default message indicating only that you unwraped an `Err`. -//! -//! Of the two, `expect` is generally preferred since its `msg` field allows you -//! to convey your intent and assumptions which makes tracking down the source -//! of a panic easier. `unwrap` on the other hand can still be a good fit in -//! situations where you can trivially show that a piece of code will never -//! panic, such as `"127.0.0.1".parse::<std::net::IpAddr>().unwrap()` or early -//! prototyping. -//! -//! # Common Message Styles -//! -//! There are two common styles for how people word `expect` messages. Using -//! the message to present information to users encountering a panic -//! ("expect as error message") or using the message to present information -//! to developers debugging the panic ("expect as precondition"). -//! -//! In the former case the expect message is used to describe the error that -//! has occurred which is considered a bug. Consider the following example: -//! -//! ```should_panic -//! // Read environment variable, panic if it is not present -//! let path = std::env::var("IMPORTANT_PATH").unwrap(); -//! ``` -//! -//! In the "expect as error message" style we would use expect to describe -//! that the environment variable was not set when it should have been: -//! -//! ```should_panic -//! let path = std::env::var("IMPORTANT_PATH") -//! .expect("env variable `IMPORTANT_PATH` is not set"); -//! ``` -//! -//! In the "expect as precondition" style, we would instead describe the -//! reason we _expect_ the `Result` should be `Ok`. With this style we would -//! prefer to write: -//! -//! ```should_panic -//! let path = std::env::var("IMPORTANT_PATH") -//! .expect("env variable `IMPORTANT_PATH` should be set by `wrapper_script.sh`"); -//! ``` -//! -//! The "expect as error message" style does not work as well with the -//! default output of the std panic hooks, and often ends up repeating -//! information that is already communicated by the source error being -//! unwrapped: -//! -//! ```text -//! thread 'main' panicked at 'env variable `IMPORTANT_PATH` is not set: NotPresent', src/main.rs:4:6 -//! ``` -//! -//! In this example we end up mentioning that an env variable is not set, -//! followed by our source message that says the env is not present, the -//! only additional information we're communicating is the name of the -//! environment variable being checked. -//! -//! The "expect as precondition" style instead focuses on source code -//! readability, making it easier to understand what must have gone wrong in -//! situations where panics are being used to represent bugs exclusively. -//! Also, by framing our expect in terms of what "SHOULD" have happened to -//! prevent the source error, we end up introducing new information that is -//! independent from our source error. -//! -//! ```text -//! thread 'main' panicked at 'env variable `IMPORTANT_PATH` should be set by `wrapper_script.sh`: NotPresent', src/main.rs:4:6 -//! ``` -//! -//! In this example we are communicating not only the name of the -//! environment variable that should have been set, but also an explanation -//! for why it should have been set, and we let the source error display as -//! a clear contradiction to our expectation. -//! -//! **Hint**: If you're having trouble remembering how to phrase -//! expect-as-precondition style error messages remember to focus on the word -//! "should" as in "env variable should be set by blah" or "the given binary -//! should be available and executable by the current user". -//! -//! [`panic_any`]: crate::panic::panic_any -//! [`PanicInfo`]: crate::panic::PanicInfo -//! [`catch_unwind`]: crate::panic::catch_unwind -//! [`resume_unwind`]: crate::panic::resume_unwind -//! [`downcast`]: crate::error::Error -//! [`Termination`]: crate::process::Termination -//! [`Try`]: crate::ops::Try -//! [panic hook]: crate::panic::set_hook -//! [`set_hook`]: crate::panic::set_hook -//! [`take_hook`]: crate::panic::take_hook -//! [panic-handler]: <https://doc.rust-lang.org/nomicon/panic-handler.html> -//! [`match`]: ../../std/keyword.match.html -//! [`?`]: ../../std/result/index.html#the-question-mark-operator- - +#![doc = include_str!("../../core/src/error.md")] #![stable(feature = "rust1", since = "1.0.0")] -// A note about crates and the facade: -// -// Originally, the `Error` trait was defined in libcore, and the impls -// were scattered about. However, coherence objected to this -// arrangement, because to create the blanket impls for `Box` required -// knowing that `&str: !Error`, and we have no means to deal with that -// sort of conflict just now. Therefore, for the time being, we have -// moved the `Error` trait into libstd. As we evolve a sol'n to the -// coherence challenge (e.g., specialization, neg impls, etc) we can -// reconsider what crate these items belong in. - #[cfg(test)] mod tests; -use core::array; -use core::convert::Infallible; - -use crate::alloc::{AllocError, LayoutError}; -use crate::any::{Demand, Provider, TypeId}; use crate::backtrace::Backtrace; -use crate::borrow::Cow; -use crate::cell; -use crate::char; -use crate::fmt::{self, Debug, Display, Write}; -use crate::io; -use crate::mem::transmute; -use crate::num; -use crate::str; -use crate::string; -use crate::sync::Arc; -use crate::time; +use crate::fmt::{self, Write}; -/// `Error` is a trait representing the basic expectations for error values, -/// i.e., values of type `E` in [`Result<T, E>`]. -/// -/// Errors must describe themselves through the [`Display`] and [`Debug`] -/// traits. Error messages are typically concise lowercase sentences without -/// trailing punctuation: -/// -/// ``` -/// let err = "NaN".parse::<u32>().unwrap_err(); -/// assert_eq!(err.to_string(), "invalid digit found in string"); -/// ``` -/// -/// Errors may provide cause chain information. [`Error::source()`] is generally -/// used when errors cross "abstraction boundaries". If one module must report -/// an error that is caused by an error from a lower-level module, it can allow -/// accessing that error via [`Error::source()`]. This makes it possible for the -/// high-level module to provide its own errors while also revealing some of the -/// implementation for debugging via `source` chains. #[stable(feature = "rust1", since = "1.0.0")] -#[cfg_attr(not(test), rustc_diagnostic_item = "Error")] -pub trait Error: Debug + Display { - /// The lower-level source of this error, if any. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::fmt; - /// - /// #[derive(Debug)] - /// struct SuperError { - /// source: SuperErrorSideKick, - /// } - /// - /// impl fmt::Display for SuperError { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "SuperError is here!") - /// } - /// } - /// - /// impl Error for SuperError { - /// fn source(&self) -> Option<&(dyn Error + 'static)> { - /// Some(&self.source) - /// } - /// } - /// - /// #[derive(Debug)] - /// struct SuperErrorSideKick; - /// - /// impl fmt::Display for SuperErrorSideKick { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "SuperErrorSideKick is here!") - /// } - /// } - /// - /// impl Error for SuperErrorSideKick {} - /// - /// fn get_super_error() -> Result<(), SuperError> { - /// Err(SuperError { source: SuperErrorSideKick }) - /// } - /// - /// fn main() { - /// match get_super_error() { - /// Err(e) => { - /// println!("Error: {e}"); - /// println!("Caused by: {}", e.source().unwrap()); - /// } - /// _ => println!("No error"), - /// } - /// } - /// ``` - #[stable(feature = "error_source", since = "1.30.0")] - fn source(&self) -> Option<&(dyn Error + 'static)> { - None - } - - /// Gets the `TypeId` of `self`. - #[doc(hidden)] - #[unstable( - feature = "error_type_id", - reason = "this is memory-unsafe to override in user code", - issue = "60784" - )] - fn type_id(&self, _: private::Internal) -> TypeId - where - Self: 'static, - { - TypeId::of::<Self>() - } - - /// ``` - /// if let Err(e) = "xc".parse::<u32>() { - /// // Print `e` itself, no need for description(). - /// eprintln!("Error: {e}"); - /// } - /// ``` - #[stable(feature = "rust1", since = "1.0.0")] - #[deprecated(since = "1.42.0", note = "use the Display impl or to_string()")] - fn description(&self) -> &str { - "description() is deprecated; use Display" - } - - #[stable(feature = "rust1", since = "1.0.0")] - #[deprecated( - since = "1.33.0", - note = "replaced by Error::source, which can support downcasting" - )] - #[allow(missing_docs)] - fn cause(&self) -> Option<&dyn Error> { - self.source() - } - - /// Provides type based access to context intended for error reports. - /// - /// Used in conjunction with [`Demand::provide_value`] and [`Demand::provide_ref`] to extract - /// references to member variables from `dyn Error` trait objects. - /// - /// # Example - /// - /// ```rust - /// #![feature(provide_any)] - /// #![feature(error_generic_member_access)] - /// use core::fmt; - /// use core::any::Demand; - /// - /// #[derive(Debug)] - /// struct MyBacktrace { - /// // ... - /// } - /// - /// impl MyBacktrace { - /// fn new() -> MyBacktrace { - /// // ... - /// # MyBacktrace {} - /// } - /// } - /// - /// #[derive(Debug)] - /// struct SourceError { - /// // ... - /// } - /// - /// impl fmt::Display for SourceError { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "Example Source Error") - /// } - /// } - /// - /// impl std::error::Error for SourceError {} - /// - /// #[derive(Debug)] - /// struct Error { - /// source: SourceError, - /// backtrace: MyBacktrace, - /// } - /// - /// impl fmt::Display for Error { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "Example Error") - /// } - /// } - /// - /// impl std::error::Error for Error { - /// fn provide<'a>(&'a self, req: &mut Demand<'a>) { - /// req - /// .provide_ref::<MyBacktrace>(&self.backtrace) - /// .provide_ref::<dyn std::error::Error + 'static>(&self.source); - /// } - /// } - /// - /// fn main() { - /// let backtrace = MyBacktrace::new(); - /// let source = SourceError {}; - /// let error = Error { source, backtrace }; - /// let dyn_error = &error as &dyn std::error::Error; - /// let backtrace_ref = dyn_error.request_ref::<MyBacktrace>().unwrap(); - /// - /// assert!(core::ptr::eq(&error.backtrace, backtrace_ref)); - /// } - /// ``` - #[unstable(feature = "error_generic_member_access", issue = "99301")] - #[allow(unused_variables)] - fn provide<'a>(&'a self, req: &mut Demand<'a>) {} -} - -#[unstable(feature = "error_generic_member_access", issue = "99301")] -impl<'b> Provider for dyn Error + 'b { - fn provide<'a>(&'a self, req: &mut Demand<'a>) { - self.provide(req) - } -} +pub use core::error::Error; mod private { // This is a hack to prevent `type_id` from being overridden by `Error` @@ -370,742 +18,10 @@ mod private { pub struct Internal; } -#[stable(feature = "rust1", since = "1.0.0")] -impl<'a, E: Error + 'a> From<E> for Box<dyn Error + 'a> { - /// Converts a type of [`Error`] into a box of dyn [`Error`]. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::fmt; - /// use std::mem; - /// - /// #[derive(Debug)] - /// struct AnError; - /// - /// impl fmt::Display for AnError { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "An error") - /// } - /// } - /// - /// impl Error for AnError {} - /// - /// let an_error = AnError; - /// assert!(0 == mem::size_of_val(&an_error)); - /// let a_boxed_error = Box::<dyn Error>::from(an_error); - /// assert!(mem::size_of::<Box<dyn Error>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - fn from(err: E) -> Box<dyn Error + 'a> { - Box::new(err) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl<'a, E: Error + Send + Sync + 'a> From<E> for Box<dyn Error + Send + Sync + 'a> { - /// Converts a type of [`Error`] + [`Send`] + [`Sync`] into a box of - /// dyn [`Error`] + [`Send`] + [`Sync`]. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::fmt; - /// use std::mem; - /// - /// #[derive(Debug)] - /// struct AnError; - /// - /// impl fmt::Display for AnError { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "An error") - /// } - /// } - /// - /// impl Error for AnError {} - /// - /// unsafe impl Send for AnError {} - /// - /// unsafe impl Sync for AnError {} - /// - /// let an_error = AnError; - /// assert!(0 == mem::size_of_val(&an_error)); - /// let a_boxed_error = Box::<dyn Error + Send + Sync>::from(an_error); - /// assert!( - /// mem::size_of::<Box<dyn Error + Send + Sync>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - fn from(err: E) -> Box<dyn Error + Send + Sync + 'a> { - Box::new(err) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl From<String> for Box<dyn Error + Send + Sync> { - /// Converts a [`String`] into a box of dyn [`Error`] + [`Send`] + [`Sync`]. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::mem; - /// - /// let a_string_error = "a string error".to_string(); - /// let a_boxed_error = Box::<dyn Error + Send + Sync>::from(a_string_error); - /// assert!( - /// mem::size_of::<Box<dyn Error + Send + Sync>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - #[inline] - fn from(err: String) -> Box<dyn Error + Send + Sync> { - struct StringError(String); - - impl Error for StringError { - #[allow(deprecated)] - fn description(&self) -> &str { - &self.0 - } - } - - impl Display for StringError { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - Display::fmt(&self.0, f) - } - } - - // Purposefully skip printing "StringError(..)" - impl Debug for StringError { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - Debug::fmt(&self.0, f) - } - } - - Box::new(StringError(err)) - } -} - -#[stable(feature = "string_box_error", since = "1.6.0")] -impl From<String> for Box<dyn Error> { - /// Converts a [`String`] into a box of dyn [`Error`]. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::mem; - /// - /// let a_string_error = "a string error".to_string(); - /// let a_boxed_error = Box::<dyn Error>::from(a_string_error); - /// assert!(mem::size_of::<Box<dyn Error>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - fn from(str_err: String) -> Box<dyn Error> { - let err1: Box<dyn Error + Send + Sync> = From::from(str_err); - let err2: Box<dyn Error> = err1; - err2 - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl<'a> From<&str> for Box<dyn Error + Send + Sync + 'a> { - /// Converts a [`str`] into a box of dyn [`Error`] + [`Send`] + [`Sync`]. - /// - /// [`str`]: prim@str - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::mem; - /// - /// let a_str_error = "a str error"; - /// let a_boxed_error = Box::<dyn Error + Send + Sync>::from(a_str_error); - /// assert!( - /// mem::size_of::<Box<dyn Error + Send + Sync>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - #[inline] - fn from(err: &str) -> Box<dyn Error + Send + Sync + 'a> { - From::from(String::from(err)) - } -} - -#[stable(feature = "string_box_error", since = "1.6.0")] -impl From<&str> for Box<dyn Error> { - /// Converts a [`str`] into a box of dyn [`Error`]. - /// - /// [`str`]: prim@str - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::mem; - /// - /// let a_str_error = "a str error"; - /// let a_boxed_error = Box::<dyn Error>::from(a_str_error); - /// assert!(mem::size_of::<Box<dyn Error>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - fn from(err: &str) -> Box<dyn Error> { - From::from(String::from(err)) - } -} - -#[stable(feature = "cow_box_error", since = "1.22.0")] -impl<'a, 'b> From<Cow<'b, str>> for Box<dyn Error + Send + Sync + 'a> { - /// Converts a [`Cow`] into a box of dyn [`Error`] + [`Send`] + [`Sync`]. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::mem; - /// use std::borrow::Cow; - /// - /// let a_cow_str_error = Cow::from("a str error"); - /// let a_boxed_error = Box::<dyn Error + Send + Sync>::from(a_cow_str_error); - /// assert!( - /// mem::size_of::<Box<dyn Error + Send + Sync>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - fn from(err: Cow<'b, str>) -> Box<dyn Error + Send + Sync + 'a> { - From::from(String::from(err)) - } -} - -#[stable(feature = "cow_box_error", since = "1.22.0")] -impl<'a> From<Cow<'a, str>> for Box<dyn Error> { - /// Converts a [`Cow`] into a box of dyn [`Error`]. - /// - /// # Examples - /// - /// ``` - /// use std::error::Error; - /// use std::mem; - /// use std::borrow::Cow; - /// - /// let a_cow_str_error = Cow::from("a str error"); - /// let a_boxed_error = Box::<dyn Error>::from(a_cow_str_error); - /// assert!(mem::size_of::<Box<dyn Error>>() == mem::size_of_val(&a_boxed_error)) - /// ``` - fn from(err: Cow<'a, str>) -> Box<dyn Error> { - From::from(String::from(err)) - } -} - -#[unstable(feature = "never_type", issue = "35121")] -impl Error for ! {} - -#[unstable( - feature = "allocator_api", - reason = "the precise API and guarantees it provides may be tweaked.", - issue = "32838" -)] -impl Error for AllocError {} - -#[stable(feature = "alloc_layout", since = "1.28.0")] -impl Error for LayoutError {} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for str::ParseBoolError { - #[allow(deprecated)] - fn description(&self) -> &str { - "failed to parse bool" - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for str::Utf8Error { - #[allow(deprecated)] - fn description(&self) -> &str { - "invalid utf-8: corrupt contents" - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for num::ParseIntError { - #[allow(deprecated)] - fn description(&self) -> &str { - self.__description() - } -} - -#[stable(feature = "try_from", since = "1.34.0")] -impl Error for num::TryFromIntError { - #[allow(deprecated)] - fn description(&self) -> &str { - self.__description() - } -} - -#[stable(feature = "try_from", since = "1.34.0")] -impl Error for array::TryFromSliceError { - #[allow(deprecated)] - fn description(&self) -> &str { - self.__description() - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for num::ParseFloatError { - #[allow(deprecated)] - fn description(&self) -> &str { - self.__description() - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for string::FromUtf8Error { - #[allow(deprecated)] - fn description(&self) -> &str { - "invalid utf-8" - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for string::FromUtf16Error { - #[allow(deprecated)] - fn description(&self) -> &str { - "invalid utf-16" - } -} - -#[stable(feature = "str_parse_error2", since = "1.8.0")] -impl Error for Infallible { - fn description(&self) -> &str { - match *self {} - } -} - -#[stable(feature = "decode_utf16", since = "1.9.0")] -impl Error for char::DecodeUtf16Error { - #[allow(deprecated)] - fn description(&self) -> &str { - "unpaired surrogate found" - } -} - -#[stable(feature = "u8_from_char", since = "1.59.0")] -impl Error for char::TryFromCharError {} - -#[unstable(feature = "map_try_insert", issue = "82766")] -impl<'a, K: Debug + Ord, V: Debug> Error - for crate::collections::btree_map::OccupiedError<'a, K, V> -{ - #[allow(deprecated)] - fn description(&self) -> &str { - "key already exists" - } -} - -#[unstable(feature = "map_try_insert", issue = "82766")] -impl<'a, K: Debug, V: Debug> Error for crate::collections::hash_map::OccupiedError<'a, K, V> { - #[allow(deprecated)] - fn description(&self) -> &str { - "key already exists" - } -} - -#[stable(feature = "box_error", since = "1.8.0")] -impl<T: Error> Error for Box<T> { - #[allow(deprecated, deprecated_in_future)] - fn description(&self) -> &str { - Error::description(&**self) - } - - #[allow(deprecated)] - fn cause(&self) -> Option<&dyn Error> { - Error::cause(&**self) - } - - fn source(&self) -> Option<&(dyn Error + 'static)> { - Error::source(&**self) - } -} - -#[unstable(feature = "thin_box", issue = "92791")] -impl<T: ?Sized + crate::error::Error> crate::error::Error for crate::boxed::ThinBox<T> { - fn source(&self) -> Option<&(dyn crate::error::Error + 'static)> { - use core::ops::Deref; - self.deref().source() - } -} - -#[stable(feature = "error_by_ref", since = "1.51.0")] -impl<'a, T: Error + ?Sized> Error for &'a T { - #[allow(deprecated, deprecated_in_future)] - fn description(&self) -> &str { - Error::description(&**self) - } - - #[allow(deprecated)] - fn cause(&self) -> Option<&dyn Error> { - Error::cause(&**self) - } - - fn source(&self) -> Option<&(dyn Error + 'static)> { - Error::source(&**self) - } - - fn provide<'b>(&'b self, req: &mut Demand<'b>) { - Error::provide(&**self, req); - } -} - -#[stable(feature = "arc_error", since = "1.52.0")] -impl<T: Error + ?Sized> Error for Arc<T> { - #[allow(deprecated, deprecated_in_future)] - fn description(&self) -> &str { - Error::description(&**self) - } - - #[allow(deprecated)] - fn cause(&self) -> Option<&dyn Error> { - Error::cause(&**self) - } - - fn source(&self) -> Option<&(dyn Error + 'static)> { - Error::source(&**self) - } - - fn provide<'a>(&'a self, req: &mut Demand<'a>) { - Error::provide(&**self, req); - } -} - -#[stable(feature = "fmt_error", since = "1.11.0")] -impl Error for fmt::Error { - #[allow(deprecated)] - fn description(&self) -> &str { - "an error occurred when formatting an argument" - } -} - -#[stable(feature = "try_borrow", since = "1.13.0")] -impl Error for cell::BorrowError { - #[allow(deprecated)] - fn description(&self) -> &str { - "already mutably borrowed" - } -} - -#[stable(feature = "try_borrow", since = "1.13.0")] -impl Error for cell::BorrowMutError { - #[allow(deprecated)] - fn description(&self) -> &str { - "already borrowed" - } -} - -#[stable(feature = "try_from", since = "1.34.0")] -impl Error for char::CharTryFromError { - #[allow(deprecated)] - fn description(&self) -> &str { - "converted integer out of range for `char`" - } -} - -#[stable(feature = "char_from_str", since = "1.20.0")] -impl Error for char::ParseCharError { - #[allow(deprecated)] - fn description(&self) -> &str { - self.__description() - } -} - -#[stable(feature = "try_reserve", since = "1.57.0")] -impl Error for alloc::collections::TryReserveError {} - -#[unstable(feature = "duration_checked_float", issue = "83400")] -impl Error for time::FromFloatSecsError {} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Error for alloc::ffi::NulError { - #[allow(deprecated)] - fn description(&self) -> &str { - "nul byte found in data" - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl From<alloc::ffi::NulError> for io::Error { - /// Converts a [`alloc::ffi::NulError`] into a [`io::Error`]. - fn from(_: alloc::ffi::NulError) -> io::Error { - io::const_io_error!(io::ErrorKind::InvalidInput, "data provided contains a nul byte") - } -} - -#[stable(feature = "frombyteswithnulerror_impls", since = "1.17.0")] -impl Error for core::ffi::FromBytesWithNulError { - #[allow(deprecated)] - fn description(&self) -> &str { - self.__description() - } -} - -#[unstable(feature = "cstr_from_bytes_until_nul", issue = "95027")] -impl Error for core::ffi::FromBytesUntilNulError {} - -#[stable(feature = "cstring_from_vec_with_nul", since = "1.58.0")] -impl Error for alloc::ffi::FromVecWithNulError {} - -#[stable(feature = "cstring_into", since = "1.7.0")] -impl Error for alloc::ffi::IntoStringError { - #[allow(deprecated)] - fn description(&self) -> &str { - "C string contained non-utf8 bytes" - } - - fn source(&self) -> Option<&(dyn Error + 'static)> { - Some(self.__source()) - } -} - -impl<'a> dyn Error + 'a { - /// Request a reference of type `T` as context about this error. - #[unstable(feature = "error_generic_member_access", issue = "99301")] - pub fn request_ref<T: ?Sized + 'static>(&'a self) -> Option<&'a T> { - core::any::request_ref(self) - } - - /// Request a value of type `T` as context about this error. - #[unstable(feature = "error_generic_member_access", issue = "99301")] - pub fn request_value<T: 'static>(&'a self) -> Option<T> { - core::any::request_value(self) - } -} - -// Copied from `any.rs`. -impl dyn Error + 'static { - /// Returns `true` if the inner type is the same as `T`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn is<T: Error + 'static>(&self) -> bool { - // Get `TypeId` of the type this function is instantiated with. - let t = TypeId::of::<T>(); - - // Get `TypeId` of the type in the trait object (`self`). - let concrete = self.type_id(private::Internal); - - // Compare both `TypeId`s on equality. - t == concrete - } - - /// Returns some reference to the inner value if it is of type `T`, or - /// `None` if it isn't. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn downcast_ref<T: Error + 'static>(&self) -> Option<&T> { - if self.is::<T>() { - unsafe { Some(&*(self as *const dyn Error as *const T)) } - } else { - None - } - } - - /// Returns some mutable reference to the inner value if it is of type `T`, or - /// `None` if it isn't. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn downcast_mut<T: Error + 'static>(&mut self) -> Option<&mut T> { - if self.is::<T>() { - unsafe { Some(&mut *(self as *mut dyn Error as *mut T)) } - } else { - None - } - } -} - -impl dyn Error + 'static + Send { - /// Forwards to the method defined on the type `dyn Error`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn is<T: Error + 'static>(&self) -> bool { - <dyn Error + 'static>::is::<T>(self) - } - - /// Forwards to the method defined on the type `dyn Error`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn downcast_ref<T: Error + 'static>(&self) -> Option<&T> { - <dyn Error + 'static>::downcast_ref::<T>(self) - } - - /// Forwards to the method defined on the type `dyn Error`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn downcast_mut<T: Error + 'static>(&mut self) -> Option<&mut T> { - <dyn Error + 'static>::downcast_mut::<T>(self) - } - - /// Request a reference of type `T` as context about this error. - #[unstable(feature = "error_generic_member_access", issue = "99301")] - pub fn request_ref<T: ?Sized + 'static>(&self) -> Option<&T> { - <dyn Error>::request_ref(self) - } - - /// Request a value of type `T` as context about this error. - #[unstable(feature = "error_generic_member_access", issue = "99301")] - pub fn request_value<T: 'static>(&self) -> Option<T> { - <dyn Error>::request_value(self) - } -} - -impl dyn Error + 'static + Send + Sync { - /// Forwards to the method defined on the type `dyn Error`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn is<T: Error + 'static>(&self) -> bool { - <dyn Error + 'static>::is::<T>(self) - } - - /// Forwards to the method defined on the type `dyn Error`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn downcast_ref<T: Error + 'static>(&self) -> Option<&T> { - <dyn Error + 'static>::downcast_ref::<T>(self) - } - - /// Forwards to the method defined on the type `dyn Error`. - #[stable(feature = "error_downcast", since = "1.3.0")] - #[inline] - pub fn downcast_mut<T: Error + 'static>(&mut self) -> Option<&mut T> { - <dyn Error + 'static>::downcast_mut::<T>(self) - } - - /// Request a reference of type `T` as context about this error. - #[unstable(feature = "error_generic_member_access", issue = "99301")] - pub fn request_ref<T: ?Sized + 'static>(&self) -> Option<&T> { - <dyn Error>::request_ref(self) - } - - /// Request a value of type `T` as context about this error. - #[unstable(feature = "error_generic_member_access", issue = "99301")] - pub fn request_value<T: 'static>(&self) -> Option<T> { - <dyn Error>::request_value(self) - } -} - -impl dyn Error { - #[inline] - #[stable(feature = "error_downcast", since = "1.3.0")] - /// Attempts to downcast the box to a concrete type. - pub fn downcast<T: Error + 'static>(self: Box<Self>) -> Result<Box<T>, Box<dyn Error>> { - if self.is::<T>() { - unsafe { - let raw: *mut dyn Error = Box::into_raw(self); - Ok(Box::from_raw(raw as *mut T)) - } - } else { - Err(self) - } - } - - /// Returns an iterator starting with the current error and continuing with - /// recursively calling [`Error::source`]. - /// - /// If you want to omit the current error and only use its sources, - /// use `skip(1)`. - /// - /// # Examples - /// - /// ``` - /// #![feature(error_iter)] - /// use std::error::Error; - /// use std::fmt; - /// - /// #[derive(Debug)] - /// struct A; - /// - /// #[derive(Debug)] - /// struct B(Option<Box<dyn Error + 'static>>); - /// - /// impl fmt::Display for A { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "A") - /// } - /// } - /// - /// impl fmt::Display for B { - /// fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - /// write!(f, "B") - /// } - /// } - /// - /// impl Error for A {} - /// - /// impl Error for B { - /// fn source(&self) -> Option<&(dyn Error + 'static)> { - /// self.0.as_ref().map(|e| e.as_ref()) - /// } - /// } - /// - /// let b = B(Some(Box::new(A))); - /// - /// // let err : Box<Error> = b.into(); // or - /// let err = &b as &(dyn Error); - /// - /// let mut iter = err.chain(); - /// - /// assert_eq!("B".to_string(), iter.next().unwrap().to_string()); - /// assert_eq!("A".to_string(), iter.next().unwrap().to_string()); - /// assert!(iter.next().is_none()); - /// assert!(iter.next().is_none()); - /// ``` - #[unstable(feature = "error_iter", issue = "58520")] - #[inline] - pub fn chain(&self) -> Chain<'_> { - Chain { current: Some(self) } - } -} - -/// An iterator over an [`Error`] and its sources. -/// -/// If you want to omit the initial error and only process -/// its sources, use `skip(1)`. -#[unstable(feature = "error_iter", issue = "58520")] -#[derive(Clone, Debug)] -pub struct Chain<'a> { - current: Option<&'a (dyn Error + 'static)>, -} - -#[unstable(feature = "error_iter", issue = "58520")] -impl<'a> Iterator for Chain<'a> { - type Item = &'a (dyn Error + 'static); - - fn next(&mut self) -> Option<Self::Item> { - let current = self.current; - self.current = self.current.and_then(Error::source); - current - } -} - -impl dyn Error + Send { - #[inline] - #[stable(feature = "error_downcast", since = "1.3.0")] - /// Attempts to downcast the box to a concrete type. - pub fn downcast<T: Error + 'static>(self: Box<Self>) -> Result<Box<T>, Box<dyn Error + Send>> { - let err: Box<dyn Error> = self; - <dyn Error>::downcast(err).map_err(|s| unsafe { - // Reapply the `Send` marker. - transmute::<Box<dyn Error>, Box<dyn Error + Send>>(s) - }) - } -} - -impl dyn Error + Send + Sync { - #[inline] - #[stable(feature = "error_downcast", since = "1.3.0")] - /// Attempts to downcast the box to a concrete type. - pub fn downcast<T: Error + 'static>(self: Box<Self>) -> Result<Box<T>, Box<Self>> { - let err: Box<dyn Error> = self; - <dyn Error>::downcast(err).map_err(|s| unsafe { - // Reapply the `Send + Sync` marker. - transmute::<Box<dyn Error>, Box<dyn Error + Send + Sync>>(s) - }) - } -} - /// An error reporter that prints an error and its sources. /// -/// Report also exposes configuration options for formatting the error chain, either entirely on a -/// single line, or in multi-line format with each cause in the error chain on a new line. +/// Report also exposes configuration options for formatting the error sources, either entirely on a +/// single line, or in multi-line format with each source on a new line. /// /// `Report` only requires that the wrapped error implement `Error`. It doesn't require that the /// wrapped error be `Send`, `Sync`, or `'static`. @@ -1246,7 +162,7 @@ impl dyn Error + Send + Sync { /// # Err(SuperError { source: SuperErrorSideKick }) /// # } /// -/// fn main() -> Result<(), Report> { +/// fn main() -> Result<(), Report<SuperError>> { /// get_super_error()?; /// Ok(()) /// } @@ -1293,7 +209,7 @@ impl dyn Error + Send + Sync { /// # Err(SuperError { source: SuperErrorSideKick }) /// # } /// -/// fn main() -> Result<(), Report> { +/// fn main() -> Result<(), Report<SuperError>> { /// get_super_error() /// .map_err(Report::from) /// .map_err(|r| r.pretty(true).show_backtrace(true))?; @@ -1450,11 +366,10 @@ impl<E> Report<E> { /// /// **Note**: Report will search for the first `Backtrace` it can find starting from the /// outermost error. In this example it will display the backtrace from the second error in the - /// chain, `SuperErrorSideKick`. + /// sources, `SuperErrorSideKick`. /// /// ```rust /// #![feature(error_reporter)] - /// #![feature(backtrace)] /// #![feature(provide_any)] /// #![feature(error_generic_member_access)] /// # use std::error::Error; @@ -1489,9 +404,8 @@ impl<E> Report<E> { /// } /// /// impl Error for SuperErrorSideKick { - /// fn provide<'a>(&'a self, req: &mut Demand<'a>) { - /// req - /// .provide_ref::<Backtrace>(&self.backtrace); + /// fn provide<'a>(&'a self, demand: &mut Demand<'a>) { + /// demand.provide_ref::<Backtrace>(&self.backtrace); /// } /// } /// @@ -1548,73 +462,7 @@ where let backtrace = backtrace.or_else(|| { self.error .source() - .map(|source| source.chain().find_map(|source| source.request_ref())) - .flatten() - }); - backtrace - } - - /// Format the report as a single line. - #[unstable(feature = "error_reporter", issue = "90172")] - fn fmt_singleline(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - write!(f, "{}", self.error)?; - - let sources = self.error.source().into_iter().flat_map(<dyn Error>::chain); - - for cause in sources { - write!(f, ": {cause}")?; - } - - Ok(()) - } - - /// Format the report as multiple lines, with each error cause on its own line. - #[unstable(feature = "error_reporter", issue = "90172")] - fn fmt_multiline(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - let error = &self.error; - - write!(f, "{error}")?; - - if let Some(cause) = error.source() { - write!(f, "\n\nCaused by:")?; - - let multiple = cause.source().is_some(); - - for (ind, error) in cause.chain().enumerate() { - writeln!(f)?; - let mut indented = Indented { inner: f }; - if multiple { - write!(indented, "{ind: >4}: {error}")?; - } else { - write!(indented, " {error}")?; - } - } - } - - if self.show_backtrace { - let backtrace = self.backtrace(); - - if let Some(backtrace) = backtrace { - let backtrace = backtrace.to_string(); - - f.write_str("\n\nStack backtrace:\n")?; - f.write_str(backtrace.trim_end())?; - } - } - - Ok(()) - } -} - -impl Report<Box<dyn Error>> { - fn backtrace(&self) -> Option<&Backtrace> { - // have to grab the backtrace on the first error directly since that error may not be - // 'static - let backtrace = self.error.request_ref(); - let backtrace = backtrace.or_else(|| { - self.error - .source() - .map(|source| source.chain().find_map(|source| source.request_ref())) + .map(|source| source.sources().find_map(|source| source.request_ref())) .flatten() }); backtrace @@ -1625,7 +473,7 @@ impl Report<Box<dyn Error>> { fn fmt_singleline(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { write!(f, "{}", self.error)?; - let sources = self.error.source().into_iter().flat_map(<dyn Error>::chain); + let sources = self.error.source().into_iter().flat_map(<dyn Error>::sources); for cause in sources { write!(f, ": {cause}")?; @@ -1646,7 +494,7 @@ impl Report<Box<dyn Error>> { let multiple = cause.source().is_some(); - for (ind, error) in cause.chain().enumerate() { + for (ind, error) in cause.sources().enumerate() { writeln!(f)?; let mut indented = Indented { inner: f }; if multiple { @@ -1683,17 +531,6 @@ where } #[unstable(feature = "error_reporter", issue = "90172")] -impl<'a, E> From<E> for Report<Box<dyn Error + 'a>> -where - E: Error + 'a, -{ - fn from(error: E) -> Self { - let error = box error; - Report { error, show_backtrace: false, pretty: false } - } -} - -#[unstable(feature = "error_reporter", issue = "90172")] impl<E> fmt::Display for Report<E> where E: Error, @@ -1703,13 +540,6 @@ where } } -#[unstable(feature = "error_reporter", issue = "90172")] -impl fmt::Display for Report<Box<dyn Error>> { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - if self.pretty { self.fmt_multiline(f) } else { self.fmt_singleline(f) } - } -} - // This type intentionally outputs the same format for `Display` and `Debug`for // situations where you unwrap a `Report` or return it from main. #[unstable(feature = "error_reporter", issue = "90172")] diff --git a/library/std/src/f32.rs b/library/std/src/f32.rs index 933b52b4d..3dd5b1250 100644 --- a/library/std/src/f32.rs +++ b/library/std/src/f32.rs @@ -1,4 +1,4 @@ -//! Constants specific to the `f32` single-precision floating point type. +//! Constants for the `f32` single-precision floating point type. //! //! *[See also the `f32` primitive type](primitive@f32).* //! diff --git a/library/std/src/f32/tests.rs b/library/std/src/f32/tests.rs index 69fa203ff..4ec16c84a 100644 --- a/library/std/src/f32/tests.rs +++ b/library/std/src/f32/tests.rs @@ -299,6 +299,84 @@ fn test_is_sign_negative() { assert!((-f32::NAN).is_sign_negative()); } +#[allow(unused_macros)] +macro_rules! assert_f32_biteq { + ($left : expr, $right : expr) => { + let l: &f32 = &$left; + let r: &f32 = &$right; + let lb = l.to_bits(); + let rb = r.to_bits(); + assert_eq!(lb, rb, "float {} ({:#x}) is not equal to {} ({:#x})", *l, lb, *r, rb); + }; +} + +// Ignore test on x87 floating point, these platforms do not guarantee NaN +// payloads are preserved and flush denormals to zero, failing the tests. +#[cfg(not(target_arch = "x86"))] +#[test] +fn test_next_up() { + let tiny = f32::from_bits(1); + let tiny_up = f32::from_bits(2); + let max_down = f32::from_bits(0x7f7f_fffe); + let largest_subnormal = f32::from_bits(0x007f_ffff); + let smallest_normal = f32::from_bits(0x0080_0000); + assert_f32_biteq!(f32::NEG_INFINITY.next_up(), f32::MIN); + assert_f32_biteq!(f32::MIN.next_up(), -max_down); + assert_f32_biteq!((-1.0 - f32::EPSILON).next_up(), -1.0); + assert_f32_biteq!((-smallest_normal).next_up(), -largest_subnormal); + assert_f32_biteq!((-tiny_up).next_up(), -tiny); + assert_f32_biteq!((-tiny).next_up(), -0.0f32); + assert_f32_biteq!((-0.0f32).next_up(), tiny); + assert_f32_biteq!(0.0f32.next_up(), tiny); + assert_f32_biteq!(tiny.next_up(), tiny_up); + assert_f32_biteq!(largest_subnormal.next_up(), smallest_normal); + assert_f32_biteq!(1.0f32.next_up(), 1.0 + f32::EPSILON); + assert_f32_biteq!(f32::MAX.next_up(), f32::INFINITY); + assert_f32_biteq!(f32::INFINITY.next_up(), f32::INFINITY); + + // Check that NaNs roundtrip. + let nan0 = f32::NAN; + let nan1 = f32::from_bits(f32::NAN.to_bits() ^ 0x002a_aaaa); + let nan2 = f32::from_bits(f32::NAN.to_bits() ^ 0x0055_5555); + assert_f32_biteq!(nan0.next_up(), nan0); + assert_f32_biteq!(nan1.next_up(), nan1); + assert_f32_biteq!(nan2.next_up(), nan2); +} + +// Ignore test on x87 floating point, these platforms do not guarantee NaN +// payloads are preserved and flush denormals to zero, failing the tests. +#[cfg(not(target_arch = "x86"))] +#[test] +fn test_next_down() { + let tiny = f32::from_bits(1); + let tiny_up = f32::from_bits(2); + let max_down = f32::from_bits(0x7f7f_fffe); + let largest_subnormal = f32::from_bits(0x007f_ffff); + let smallest_normal = f32::from_bits(0x0080_0000); + assert_f32_biteq!(f32::NEG_INFINITY.next_down(), f32::NEG_INFINITY); + assert_f32_biteq!(f32::MIN.next_down(), f32::NEG_INFINITY); + assert_f32_biteq!((-max_down).next_down(), f32::MIN); + assert_f32_biteq!((-1.0f32).next_down(), -1.0 - f32::EPSILON); + assert_f32_biteq!((-largest_subnormal).next_down(), -smallest_normal); + assert_f32_biteq!((-tiny).next_down(), -tiny_up); + assert_f32_biteq!((-0.0f32).next_down(), -tiny); + assert_f32_biteq!((0.0f32).next_down(), -tiny); + assert_f32_biteq!(tiny.next_down(), 0.0f32); + assert_f32_biteq!(tiny_up.next_down(), tiny); + assert_f32_biteq!(smallest_normal.next_down(), largest_subnormal); + assert_f32_biteq!((1.0 + f32::EPSILON).next_down(), 1.0f32); + assert_f32_biteq!(f32::MAX.next_down(), max_down); + assert_f32_biteq!(f32::INFINITY.next_down(), f32::MAX); + + // Check that NaNs roundtrip. + let nan0 = f32::NAN; + let nan1 = f32::from_bits(f32::NAN.to_bits() ^ 0x002a_aaaa); + let nan2 = f32::from_bits(f32::NAN.to_bits() ^ 0x0055_5555); + assert_f32_biteq!(nan0.next_down(), nan0); + assert_f32_biteq!(nan1.next_down(), nan1); + assert_f32_biteq!(nan2.next_down(), nan2); +} + #[test] fn test_mul_add() { let nan: f32 = f32::NAN; diff --git a/library/std/src/f64.rs b/library/std/src/f64.rs index a9aa84f70..31351a879 100644 --- a/library/std/src/f64.rs +++ b/library/std/src/f64.rs @@ -1,4 +1,4 @@ -//! Constants specific to the `f64` double-precision floating point type. +//! Constants for the `f64` double-precision floating point type. //! //! *[See also the `f64` primitive type](primitive@f64).* //! diff --git a/library/std/src/f64/tests.rs b/library/std/src/f64/tests.rs index 5c163cfe9..12baa68f4 100644 --- a/library/std/src/f64/tests.rs +++ b/library/std/src/f64/tests.rs @@ -289,6 +289,82 @@ fn test_is_sign_negative() { assert!((-f64::NAN).is_sign_negative()); } +#[allow(unused_macros)] +macro_rules! assert_f64_biteq { + ($left : expr, $right : expr) => { + let l: &f64 = &$left; + let r: &f64 = &$right; + let lb = l.to_bits(); + let rb = r.to_bits(); + assert_eq!(lb, rb, "float {} ({:#x}) is not equal to {} ({:#x})", *l, lb, *r, rb); + }; +} + +// Ignore test on x87 floating point, these platforms do not guarantee NaN +// payloads are preserved and flush denormals to zero, failing the tests. +#[cfg(not(target_arch = "x86"))] +#[test] +fn test_next_up() { + let tiny = f64::from_bits(1); + let tiny_up = f64::from_bits(2); + let max_down = f64::from_bits(0x7fef_ffff_ffff_fffe); + let largest_subnormal = f64::from_bits(0x000f_ffff_ffff_ffff); + let smallest_normal = f64::from_bits(0x0010_0000_0000_0000); + assert_f64_biteq!(f64::NEG_INFINITY.next_up(), f64::MIN); + assert_f64_biteq!(f64::MIN.next_up(), -max_down); + assert_f64_biteq!((-1.0 - f64::EPSILON).next_up(), -1.0); + assert_f64_biteq!((-smallest_normal).next_up(), -largest_subnormal); + assert_f64_biteq!((-tiny_up).next_up(), -tiny); + assert_f64_biteq!((-tiny).next_up(), -0.0f64); + assert_f64_biteq!((-0.0f64).next_up(), tiny); + assert_f64_biteq!(0.0f64.next_up(), tiny); + assert_f64_biteq!(tiny.next_up(), tiny_up); + assert_f64_biteq!(largest_subnormal.next_up(), smallest_normal); + assert_f64_biteq!(1.0f64.next_up(), 1.0 + f64::EPSILON); + assert_f64_biteq!(f64::MAX.next_up(), f64::INFINITY); + assert_f64_biteq!(f64::INFINITY.next_up(), f64::INFINITY); + + let nan0 = f64::NAN; + let nan1 = f64::from_bits(f64::NAN.to_bits() ^ 0x000a_aaaa_aaaa_aaaa); + let nan2 = f64::from_bits(f64::NAN.to_bits() ^ 0x0005_5555_5555_5555); + assert_f64_biteq!(nan0.next_up(), nan0); + assert_f64_biteq!(nan1.next_up(), nan1); + assert_f64_biteq!(nan2.next_up(), nan2); +} + +// Ignore test on x87 floating point, these platforms do not guarantee NaN +// payloads are preserved and flush denormals to zero, failing the tests. +#[cfg(not(target_arch = "x86"))] +#[test] +fn test_next_down() { + let tiny = f64::from_bits(1); + let tiny_up = f64::from_bits(2); + let max_down = f64::from_bits(0x7fef_ffff_ffff_fffe); + let largest_subnormal = f64::from_bits(0x000f_ffff_ffff_ffff); + let smallest_normal = f64::from_bits(0x0010_0000_0000_0000); + assert_f64_biteq!(f64::NEG_INFINITY.next_down(), f64::NEG_INFINITY); + assert_f64_biteq!(f64::MIN.next_down(), f64::NEG_INFINITY); + assert_f64_biteq!((-max_down).next_down(), f64::MIN); + assert_f64_biteq!((-1.0f64).next_down(), -1.0 - f64::EPSILON); + assert_f64_biteq!((-largest_subnormal).next_down(), -smallest_normal); + assert_f64_biteq!((-tiny).next_down(), -tiny_up); + assert_f64_biteq!((-0.0f64).next_down(), -tiny); + assert_f64_biteq!((0.0f64).next_down(), -tiny); + assert_f64_biteq!(tiny.next_down(), 0.0f64); + assert_f64_biteq!(tiny_up.next_down(), tiny); + assert_f64_biteq!(smallest_normal.next_down(), largest_subnormal); + assert_f64_biteq!((1.0 + f64::EPSILON).next_down(), 1.0f64); + assert_f64_biteq!(f64::MAX.next_down(), max_down); + assert_f64_biteq!(f64::INFINITY.next_down(), f64::MAX); + + let nan0 = f64::NAN; + let nan1 = f64::from_bits(f64::NAN.to_bits() ^ 0x000a_aaaa_aaaa_aaaa); + let nan2 = f64::from_bits(f64::NAN.to_bits() ^ 0x0005_5555_5555_5555); + assert_f64_biteq!(nan0.next_down(), nan0); + assert_f64_biteq!(nan1.next_down(), nan1); + assert_f64_biteq!(nan2.next_down(), nan2); +} + #[test] fn test_mul_add() { let nan: f64 = f64::NAN; diff --git a/library/std/src/ffi/os_str.rs b/library/std/src/ffi/os_str.rs index a0a5c003d..80ed34157 100644 --- a/library/std/src/ffi/os_str.rs +++ b/library/std/src/ffi/os_str.rs @@ -290,7 +290,8 @@ impl OsString { /// in the given `OsString`. The string may reserve more space to speculatively avoid /// frequent reallocations. After calling `try_reserve`, capacity will be /// greater than or equal to `self.len() + additional` if it returns `Ok(())`. - /// Does nothing if capacity is already sufficient. + /// Does nothing if capacity is already sufficient. This method preserves + /// the contents even if an error occurs. /// /// See the main `OsString` documentation information about encoding and capacity units. /// diff --git a/library/std/src/fs.rs b/library/std/src/fs.rs index c8e131b6e..188ff00e1 100644 --- a/library/std/src/fs.rs +++ b/library/std/src/fs.rs @@ -13,13 +13,13 @@ mod tests; use crate::ffi::OsString; use crate::fmt; -use crate::io::{self, IoSlice, IoSliceMut, Read, ReadBuf, Seek, SeekFrom, Write}; +use crate::io::{self, BorrowedCursor, IoSlice, IoSliceMut, Read, Seek, SeekFrom, Write}; use crate::path::{Path, PathBuf}; use crate::sys::fs as fs_imp; use crate::sys_common::{AsInner, AsInnerMut, FromInner, IntoInner}; use crate::time::SystemTime; -/// A reference to an open file on the filesystem. +/// An object providing access to an open file on the filesystem. /// /// An instance of a `File` can be read and/or written depending on what options /// it was opened with. Files also implement [`Seek`] to alter the logical cursor @@ -377,6 +377,35 @@ impl File { OpenOptions::new().write(true).create(true).truncate(true).open(path.as_ref()) } + /// Creates a new file in read-write mode; error if the file exists. + /// + /// This function will create a file if it does not exist, or return an error if it does. This + /// way, if the call succeeds, the file returned is guaranteed to be new. + /// + /// This option is useful because it is atomic. Otherwise between checking whether a file + /// exists and creating a new one, the file may have been created by another process (a TOCTOU + /// race condition / attack). + /// + /// This can also be written using + /// `File::options().read(true).write(true).create_new(true).open(...)`. + /// + /// # Examples + /// + /// ```no_run + /// #![feature(file_create_new)] + /// + /// use std::fs::File; + /// + /// fn main() -> std::io::Result<()> { + /// let mut f = File::create_new("foo.txt")?; + /// Ok(()) + /// } + /// ``` + #[unstable(feature = "file_create_new", issue = "none")] + pub fn create_new<P: AsRef<Path>>(path: P) -> io::Result<File> { + OpenOptions::new().read(true).write(true).create_new(true).open(path.as_ref()) + } + /// Returns a new OpenOptions object. /// /// This function returns a new OpenOptions object that you can use to @@ -703,8 +732,8 @@ impl Read for File { self.inner.read_vectored(bufs) } - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - self.inner.read_buf(buf) + fn read_buf(&mut self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + self.inner.read_buf(cursor) } #[inline] @@ -755,8 +784,8 @@ impl Read for &File { self.inner.read(buf) } - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - self.inner.read_buf(buf) + fn read_buf(&mut self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + self.inner.read_buf(cursor) } fn read_vectored(&mut self, bufs: &mut [IoSliceMut<'_>]) -> io::Result<usize> { @@ -1336,6 +1365,34 @@ impl FileTimes { impl Permissions { /// Returns `true` if these permissions describe a readonly (unwritable) file. /// + /// # Note + /// + /// This function does not take Access Control Lists (ACLs) or Unix group + /// membership into account. + /// + /// # Windows + /// + /// On Windows this returns [`FILE_ATTRIBUTE_READONLY`](https://docs.microsoft.com/en-us/windows/win32/fileio/file-attribute-constants). + /// If `FILE_ATTRIBUTE_READONLY` is set then writes to the file will fail + /// but the user may still have permission to change this flag. If + /// `FILE_ATTRIBUTE_READONLY` is *not* set then writes may still fail due + /// to lack of write permission. + /// The behavior of this attribute for directories depends on the Windows + /// version. + /// + /// # Unix (including macOS) + /// + /// On Unix-based platforms this checks if *any* of the owner, group or others + /// write permission bits are set. It does not check if the current + /// user is in the file's assigned group. It also does not check ACLs. + /// Therefore even if this returns true you may not be able to write to the + /// file, and vice versa. The [`PermissionsExt`] trait gives direct access + /// to the permission bits but also does not read ACLs. If you need to + /// accurately know whether or not a file is writable use the `access()` + /// function from libc. + /// + /// [`PermissionsExt`]: crate::os::unix::fs::PermissionsExt + /// /// # Examples /// /// ```no_run @@ -1361,8 +1418,40 @@ impl Permissions { /// using the resulting `Permission` will update file permissions to allow /// writing. /// - /// This operation does **not** modify the filesystem. To modify the - /// filesystem use the [`set_permissions`] function. + /// This operation does **not** modify the files attributes. This only + /// changes the in-memory value of these attributes for this `Permissions` + /// instance. To modify the files attributes use the [`set_permissions`] + /// function which commits these attribute changes to the file. + /// + /// # Note + /// + /// `set_readonly(false)` makes the file *world-writable* on Unix. + /// You can use the [`PermissionsExt`] trait on Unix to avoid this issue. + /// + /// It also does not take Access Control Lists (ACLs) or Unix group + /// membership into account. + /// + /// # Windows + /// + /// On Windows this sets or clears [`FILE_ATTRIBUTE_READONLY`](https://docs.microsoft.com/en-us/windows/win32/fileio/file-attribute-constants). + /// If `FILE_ATTRIBUTE_READONLY` is set then writes to the file will fail + /// but the user may still have permission to change this flag. If + /// `FILE_ATTRIBUTE_READONLY` is *not* set then the write may still fail if + /// the user does not have permission to write to the file. + /// + /// In Windows 7 and earlier this attribute prevents deleting empty + /// directories. It does not prevent modifying the directory contents. + /// On later versions of Windows this attribute is ignored for directories. + /// + /// # Unix (including macOS) + /// + /// On Unix-based platforms this sets or clears the write access bit for + /// the owner, group *and* others, equivalent to `chmod a+w <file>` + /// or `chmod a-w <file>` respectively. The latter will grant write access + /// to all users! You can use the [`PermissionsExt`] trait on Unix + /// to avoid this issue. + /// + /// [`PermissionsExt`]: crate::os::unix::fs::PermissionsExt /// /// # Examples /// @@ -1376,7 +1465,8 @@ impl Permissions { /// /// permissions.set_readonly(true); /// - /// // filesystem doesn't change + /// // filesystem doesn't change, only the in memory state of the + /// // readonly permission /// assert_eq!(false, metadata.permissions().readonly()); /// /// // just this particular `permissions`. diff --git a/library/std/src/io/buffered/bufreader.rs b/library/std/src/io/buffered/bufreader.rs index f7fbaa9c2..4f339a18a 100644 --- a/library/std/src/io/buffered/bufreader.rs +++ b/library/std/src/io/buffered/bufreader.rs @@ -2,7 +2,7 @@ mod buffer; use crate::fmt; use crate::io::{ - self, BufRead, IoSliceMut, Read, ReadBuf, Seek, SeekFrom, SizeHint, DEFAULT_BUF_SIZE, + self, BorrowedCursor, BufRead, IoSliceMut, Read, Seek, SeekFrom, SizeHint, DEFAULT_BUF_SIZE, }; use buffer::Buffer; @@ -224,6 +224,14 @@ impl<R> BufReader<R> { } } +// This is only used by a test which asserts that the initialization-tracking is correct. +#[cfg(test)] +impl<R> BufReader<R> { + pub fn initialized(&self) -> usize { + self.buf.initialized() + } +} + impl<R: Seek> BufReader<R> { /// Seeks relative to the current position. If the new position lies within the buffer, /// the buffer will not be flushed, allowing for more efficient seeks. @@ -266,21 +274,21 @@ impl<R: Read> Read for BufReader<R> { Ok(nread) } - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { + fn read_buf(&mut self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> { // If we don't have any buffered data and we're doing a massive read // (larger than our internal buffer), bypass our internal buffer // entirely. - if self.buf.pos() == self.buf.filled() && buf.remaining() >= self.capacity() { + if self.buf.pos() == self.buf.filled() && cursor.capacity() >= self.capacity() { self.discard_buffer(); - return self.inner.read_buf(buf); + return self.inner.read_buf(cursor); } - let prev = buf.filled_len(); + let prev = cursor.written(); let mut rem = self.fill_buf()?; - rem.read_buf(buf)?; + rem.read_buf(cursor.reborrow())?; - self.consume(buf.filled_len() - prev); //slice impl of read_buf known to never unfill buf + self.consume(cursor.written() - prev); //slice impl of read_buf known to never unfill buf Ok(()) } diff --git a/library/std/src/io/buffered/bufreader/buffer.rs b/library/std/src/io/buffered/bufreader/buffer.rs index 8ae01f3b0..e9e29d60c 100644 --- a/library/std/src/io/buffered/bufreader/buffer.rs +++ b/library/std/src/io/buffered/bufreader/buffer.rs @@ -9,7 +9,7 @@ /// that user code which wants to do reads from a `BufReader` via `buffer` + `consume` can do so /// without encountering any runtime bounds checks. use crate::cmp; -use crate::io::{self, Read, ReadBuf}; +use crate::io::{self, BorrowedBuf, Read}; use crate::mem::MaybeUninit; pub struct Buffer { @@ -20,13 +20,19 @@ pub struct Buffer { // Each call to `fill_buf` sets `filled` to indicate how many bytes at the start of `buf` are // initialized with bytes from a read. filled: usize, + // This is the max number of bytes returned across all `fill_buf` calls. We track this so that we + // can accurately tell `read_buf` how many bytes of buf are initialized, to bypass as much of its + // defensive initialization as possible. Note that while this often the same as `filled`, it + // doesn't need to be. Calls to `fill_buf` are not required to actually fill the buffer, and + // omitting this is a huge perf regression for `Read` impls that do not. + initialized: usize, } impl Buffer { #[inline] pub fn with_capacity(capacity: usize) -> Self { let buf = Box::new_uninit_slice(capacity); - Self { buf, pos: 0, filled: 0 } + Self { buf, pos: 0, filled: 0, initialized: 0 } } #[inline] @@ -51,6 +57,12 @@ impl Buffer { self.pos } + // This is only used by a test which asserts that the initialization-tracking is correct. + #[cfg(test)] + pub fn initialized(&self) -> usize { + self.initialized + } + #[inline] pub fn discard_buffer(&mut self) { self.pos = 0; @@ -93,12 +105,17 @@ impl Buffer { if self.pos >= self.filled { debug_assert!(self.pos == self.filled); - let mut readbuf = ReadBuf::uninit(&mut self.buf); + let mut buf = BorrowedBuf::from(&mut *self.buf); + // SAFETY: `self.filled` bytes will always have been initialized. + unsafe { + buf.set_init(self.initialized); + } - reader.read_buf(&mut readbuf)?; + reader.read_buf(buf.unfilled())?; - self.filled = readbuf.filled_len(); self.pos = 0; + self.filled = buf.len(); + self.initialized = buf.init_len(); } Ok(self.buffer()) } diff --git a/library/std/src/io/buffered/tests.rs b/library/std/src/io/buffered/tests.rs index fe45b1326..f4e688eb9 100644 --- a/library/std/src/io/buffered/tests.rs +++ b/library/std/src/io/buffered/tests.rs @@ -1,5 +1,7 @@ use crate::io::prelude::*; -use crate::io::{self, BufReader, BufWriter, ErrorKind, IoSlice, LineWriter, ReadBuf, SeekFrom}; +use crate::io::{ + self, BorrowedBuf, BufReader, BufWriter, ErrorKind, IoSlice, LineWriter, SeekFrom, +}; use crate::mem::MaybeUninit; use crate::panic; use crate::sync::atomic::{AtomicUsize, Ordering}; @@ -61,48 +63,48 @@ fn test_buffered_reader_read_buf() { let inner: &[u8] = &[5, 6, 7, 0, 1, 2, 3, 4]; let mut reader = BufReader::with_capacity(2, inner); - let mut buf = [MaybeUninit::uninit(); 3]; - let mut buf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 3]; + let mut buf: BorrowedBuf<'_> = buf.into(); - reader.read_buf(&mut buf).unwrap(); + reader.read_buf(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), [5, 6, 7]); assert_eq!(reader.buffer(), []); - let mut buf = [MaybeUninit::uninit(); 2]; - let mut buf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 2]; + let mut buf: BorrowedBuf<'_> = buf.into(); - reader.read_buf(&mut buf).unwrap(); + reader.read_buf(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), [0, 1]); assert_eq!(reader.buffer(), []); - let mut buf = [MaybeUninit::uninit(); 1]; - let mut buf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 1]; + let mut buf: BorrowedBuf<'_> = buf.into(); - reader.read_buf(&mut buf).unwrap(); + reader.read_buf(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), [2]); assert_eq!(reader.buffer(), [3]); - let mut buf = [MaybeUninit::uninit(); 3]; - let mut buf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 3]; + let mut buf: BorrowedBuf<'_> = buf.into(); - reader.read_buf(&mut buf).unwrap(); + reader.read_buf(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), [3]); assert_eq!(reader.buffer(), []); - reader.read_buf(&mut buf).unwrap(); + reader.read_buf(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), [3, 4]); assert_eq!(reader.buffer(), []); buf.clear(); - reader.read_buf(&mut buf).unwrap(); + reader.read_buf(buf.unfilled()).unwrap(); - assert_eq!(buf.filled_len(), 0); + assert!(buf.filled().is_empty()); } #[test] @@ -1037,3 +1039,27 @@ fn single_formatted_write() { writeln!(&mut writer, "{}, {}!", "hello", "world").unwrap(); assert_eq!(writer.get_ref().events, [RecordedEvent::Write("hello, world!\n".to_string())]); } + +#[test] +fn bufreader_full_initialize() { + struct OneByteReader; + impl Read for OneByteReader { + fn read(&mut self, buf: &mut [u8]) -> crate::io::Result<usize> { + if buf.len() > 0 { + buf[0] = 0; + Ok(1) + } else { + Ok(0) + } + } + } + let mut reader = BufReader::new(OneByteReader); + // Nothing is initialized yet. + assert_eq!(reader.initialized(), 0); + + let buf = reader.fill_buf().unwrap(); + // We read one byte... + assert_eq!(buf.len(), 1); + // But we initialized the whole buffer! + assert_eq!(reader.initialized(), reader.capacity()); +} diff --git a/library/std/src/io/copy.rs b/library/std/src/io/copy.rs index 1a10245e4..38b98afff 100644 --- a/library/std/src/io/copy.rs +++ b/library/std/src/io/copy.rs @@ -1,4 +1,4 @@ -use super::{BufWriter, ErrorKind, Read, ReadBuf, Result, Write, DEFAULT_BUF_SIZE}; +use super::{BorrowedBuf, BufWriter, ErrorKind, Read, Result, Write, DEFAULT_BUF_SIZE}; use crate::mem::MaybeUninit; /// Copies the entire contents of a reader into a writer. @@ -97,37 +97,39 @@ impl<I: Write> BufferedCopySpec for BufWriter<I> { loop { let buf = writer.buffer_mut(); - let mut read_buf = ReadBuf::uninit(buf.spare_capacity_mut()); + let mut read_buf: BorrowedBuf<'_> = buf.spare_capacity_mut().into(); - // SAFETY: init is either 0 or the initialized_len of the previous iteration unsafe { - read_buf.assume_init(init); + // SAFETY: init is either 0 or the init_len from the previous iteration. + read_buf.set_init(init); } if read_buf.capacity() >= DEFAULT_BUF_SIZE { - match reader.read_buf(&mut read_buf) { + let mut cursor = read_buf.unfilled(); + match reader.read_buf(cursor.reborrow()) { Ok(()) => { - let bytes_read = read_buf.filled_len(); + let bytes_read = cursor.written(); if bytes_read == 0 { return Ok(len); } - init = read_buf.initialized_len() - bytes_read; + init = read_buf.init_len() - bytes_read; + len += bytes_read as u64; - // SAFETY: ReadBuf guarantees all of its filled bytes are init + // SAFETY: BorrowedBuf guarantees all of its filled bytes are init unsafe { buf.set_len(buf.len() + bytes_read) }; - len += bytes_read as u64; + // Read again if the buffer still has enough capacity, as BufWriter itself would do // This will occur if the reader returns short reads - continue; } - Err(ref e) if e.kind() == ErrorKind::Interrupted => continue, + Err(ref e) if e.kind() == ErrorKind::Interrupted => {} Err(e) => return Err(e), } + } else { + writer.flush_buf()?; + init = 0; } - - writer.flush_buf()?; } } } @@ -136,13 +138,13 @@ fn stack_buffer_copy<R: Read + ?Sized, W: Write + ?Sized>( reader: &mut R, writer: &mut W, ) -> Result<u64> { - let mut buf = [MaybeUninit::uninit(); DEFAULT_BUF_SIZE]; - let mut buf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); DEFAULT_BUF_SIZE]; + let mut buf: BorrowedBuf<'_> = buf.into(); let mut len = 0; loop { - match reader.read_buf(&mut buf) { + match reader.read_buf(buf.unfilled()) { Ok(()) => {} Err(e) if e.kind() == ErrorKind::Interrupted => continue, Err(e) => return Err(e), diff --git a/library/std/src/io/cursor.rs b/library/std/src/io/cursor.rs index f3fbfc447..d98ab021c 100644 --- a/library/std/src/io/cursor.rs +++ b/library/std/src/io/cursor.rs @@ -5,7 +5,7 @@ use crate::io::prelude::*; use crate::alloc::Allocator; use crate::cmp; -use crate::io::{self, ErrorKind, IoSlice, IoSliceMut, ReadBuf, SeekFrom}; +use crate::io::{self, BorrowedCursor, ErrorKind, IoSlice, IoSliceMut, SeekFrom}; /// A `Cursor` wraps an in-memory buffer and provides it with a /// [`Seek`] implementation. @@ -323,12 +323,12 @@ where Ok(n) } - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - let prev_filled = buf.filled_len(); + fn read_buf(&mut self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> { + let prev_written = cursor.written(); - Read::read_buf(&mut self.fill_buf()?, buf)?; + Read::read_buf(&mut self.fill_buf()?, cursor.reborrow())?; - self.pos += (buf.filled_len() - prev_filled) as u64; + self.pos += (cursor.written() - prev_written) as u64; Ok(()) } diff --git a/library/std/src/io/error.rs b/library/std/src/io/error.rs index ff7fdcae1..3cabf2449 100644 --- a/library/std/src/io/error.rs +++ b/library/std/src/io/error.rs @@ -76,6 +76,14 @@ impl fmt::Debug for Error { } } +#[stable(feature = "rust1", since = "1.0.0")] +impl From<alloc::ffi::NulError> for Error { + /// Converts a [`alloc::ffi::NulError`] into a [`Error`]. + fn from(_: alloc::ffi::NulError) -> Error { + const_io_error!(ErrorKind::InvalidInput, "data provided contains a nul byte") + } +} + // Only derive debug in tests, to make sure it // doesn't accidentally get printed. #[cfg_attr(test, derive(Debug))] @@ -379,7 +387,7 @@ pub enum ErrorKind { impl ErrorKind { pub(crate) fn as_str(&self) -> &'static str { use ErrorKind::*; - // Strictly alphabetical, please. (Sadly rustfmt cannot do this yet.) + // tidy-alphabetical-start match *self { AddrInUse => "address in use", AddrNotAvailable => "address not available", @@ -423,6 +431,7 @@ impl ErrorKind { WouldBlock => "operation would block", WriteZero => "write zero", } + // tidy-alphabetical-end } } @@ -473,6 +482,7 @@ impl Error { /// originate from the OS itself. The `error` argument is an arbitrary /// payload which will be contained in this [`Error`]. /// + /// Note that this function allocates memory on the heap. /// If no extra payload is required, use the `From` conversion from /// `ErrorKind`. /// @@ -487,7 +497,7 @@ impl Error { /// // errors can also be created from other errors /// let custom_error2 = Error::new(ErrorKind::Interrupted, custom_error); /// - /// // creating an error without payload + /// // creating an error without payload (and without memory allocation) /// let eof_error = Error::from(ErrorKind::UnexpectedEof); /// ``` #[stable(feature = "rust1", since = "1.0.0")] @@ -564,6 +574,8 @@ impl Error { /// println!("last OS error: {os_error:?}"); /// ``` #[stable(feature = "rust1", since = "1.0.0")] + #[doc(alias = "GetLastError")] + #[doc(alias = "errno")] #[must_use] #[inline] pub fn last_os_error() -> Error { diff --git a/library/std/src/io/error/repr_bitpacked.rs b/library/std/src/io/error/repr_bitpacked.rs index 292bf4826..781ae03ad 100644 --- a/library/std/src/io/error/repr_bitpacked.rs +++ b/library/std/src/io/error/repr_bitpacked.rs @@ -269,10 +269,10 @@ where } TAG_SIMPLE_MESSAGE => ErrorData::SimpleMessage(&*ptr.cast::<SimpleMessage>().as_ptr()), TAG_CUSTOM => { - // It would be correct for us to use `ptr::sub` here (see the + // It would be correct for us to use `ptr::byte_sub` here (see the // comment above the `wrapping_add` call in `new_custom` for why), // but it isn't clear that it makes a difference, so we don't. - let custom = ptr.as_ptr().cast::<u8>().wrapping_sub(TAG_CUSTOM).cast::<Custom>(); + let custom = ptr.as_ptr().wrapping_byte_sub(TAG_CUSTOM).cast::<Custom>(); ErrorData::Custom(make_custom(custom)) } _ => { diff --git a/library/std/src/io/error/tests.rs b/library/std/src/io/error/tests.rs index c897a5e87..16c634e9a 100644 --- a/library/std/src/io/error/tests.rs +++ b/library/std/src/io/error/tests.rs @@ -86,7 +86,7 @@ fn test_errorkind_packing() { assert_eq!(Error::from(ErrorKind::NotFound).kind(), ErrorKind::NotFound); assert_eq!(Error::from(ErrorKind::PermissionDenied).kind(), ErrorKind::PermissionDenied); assert_eq!(Error::from(ErrorKind::Uncategorized).kind(), ErrorKind::Uncategorized); - // Check that the innards look like like what we want. + // Check that the innards look like what we want. assert_matches!( Error::from(ErrorKind::OutOfMemory).repr.data(), ErrorData::Simple(ErrorKind::OutOfMemory), diff --git a/library/std/src/io/impls.rs b/library/std/src/io/impls.rs index 950725473..e5048dcc8 100644 --- a/library/std/src/io/impls.rs +++ b/library/std/src/io/impls.rs @@ -6,7 +6,7 @@ use crate::cmp; use crate::collections::VecDeque; use crate::fmt; use crate::io::{ - self, BufRead, ErrorKind, IoSlice, IoSliceMut, Read, ReadBuf, Seek, SeekFrom, Write, + self, BorrowedCursor, BufRead, ErrorKind, IoSlice, IoSliceMut, Read, Seek, SeekFrom, Write, }; use crate::mem; @@ -21,8 +21,8 @@ impl<R: Read + ?Sized> Read for &mut R { } #[inline] - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - (**self).read_buf(buf) + fn read_buf(&mut self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + (**self).read_buf(cursor) } #[inline] @@ -125,8 +125,8 @@ impl<R: Read + ?Sized> Read for Box<R> { } #[inline] - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - (**self).read_buf(buf) + fn read_buf(&mut self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + (**self).read_buf(cursor) } #[inline] @@ -249,11 +249,11 @@ impl Read for &[u8] { } #[inline] - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - let amt = cmp::min(buf.remaining(), self.len()); + fn read_buf(&mut self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> { + let amt = cmp::min(cursor.capacity(), self.len()); let (a, b) = self.split_at(amt); - buf.append(a); + cursor.append(a); *self = b; Ok(()) @@ -427,10 +427,10 @@ impl<A: Allocator> Read for VecDeque<u8, A> { } #[inline] - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { + fn read_buf(&mut self, cursor: BorrowedCursor<'_>) -> io::Result<()> { let (ref mut front, _) = self.as_slices(); - let n = cmp::min(buf.remaining(), front.len()); - Read::read_buf(front, buf)?; + let n = cmp::min(cursor.capacity(), front.len()); + Read::read_buf(front, cursor)?; self.drain(..n); Ok(()) } diff --git a/library/std/src/io/mod.rs b/library/std/src/io/mod.rs index 96addbd1a..23a13523f 100644 --- a/library/std/src/io/mod.rs +++ b/library/std/src/io/mod.rs @@ -262,9 +262,12 @@ use crate::sys_common::memchr; #[stable(feature = "bufwriter_into_parts", since = "1.56.0")] pub use self::buffered::WriterPanicked; +pub(crate) use self::stdio::attempt_print_to_stderr; #[unstable(feature = "internal_output_capture", issue = "none")] #[doc(no_inline, hidden)] pub use self::stdio::set_output_capture; +#[unstable(feature = "is_terminal", issue = "98070")] +pub use self::stdio::IsTerminal; #[unstable(feature = "print_internals", issue = "none")] pub use self::stdio::{_eprint, _print}; #[stable(feature = "rust1", since = "1.0.0")] @@ -278,7 +281,7 @@ pub use self::{ }; #[unstable(feature = "read_buf", issue = "78485")] -pub use self::readbuf::ReadBuf; +pub use self::readbuf::{BorrowedBuf, BorrowedCursor}; pub(crate) use error::const_io_error; mod buffered; @@ -362,29 +365,30 @@ pub(crate) fn default_read_to_end<R: Read + ?Sized>(r: &mut R, buf: &mut Vec<u8> buf.reserve(32); // buf is full, need more space } - let mut read_buf = ReadBuf::uninit(buf.spare_capacity_mut()); + let mut read_buf: BorrowedBuf<'_> = buf.spare_capacity_mut().into(); // SAFETY: These bytes were initialized but not filled in the previous loop unsafe { - read_buf.assume_init(initialized); + read_buf.set_init(initialized); } - match r.read_buf(&mut read_buf) { + let mut cursor = read_buf.unfilled(); + match r.read_buf(cursor.reborrow()) { Ok(()) => {} Err(e) if e.kind() == ErrorKind::Interrupted => continue, Err(e) => return Err(e), } - if read_buf.filled_len() == 0 { + if cursor.written() == 0 { return Ok(buf.len() - start_len); } // store how much was initialized but not filled - initialized = read_buf.initialized_len() - read_buf.filled_len(); - let new_len = read_buf.filled_len() + buf.len(); + initialized = cursor.init_ref().len(); - // SAFETY: ReadBuf's invariants mean this much memory is init + // SAFETY: BorrowedBuf's invariants mean this much memory is initialized. unsafe { + let new_len = read_buf.filled().len() + buf.len(); buf.set_len(new_len); } @@ -461,12 +465,15 @@ pub(crate) fn default_read_exact<R: Read + ?Sized>(this: &mut R, mut buf: &mut [ } } -pub(crate) fn default_read_buf<F>(read: F, buf: &mut ReadBuf<'_>) -> Result<()> +pub(crate) fn default_read_buf<F>(read: F, mut cursor: BorrowedCursor<'_>) -> Result<()> where F: FnOnce(&mut [u8]) -> Result<usize>, { - let n = read(buf.initialize_unfilled())?; - buf.add_filled(n); + let n = read(cursor.ensure_init().init_mut())?; + unsafe { + // SAFETY: we initialised using `ensure_init` so there is no uninit data to advance to. + cursor.advance(n); + } Ok(()) } @@ -576,7 +583,7 @@ pub trait Read { /// `n > buf.len()`. /// /// No guarantees are provided about the contents of `buf` when this - /// function is called, implementations cannot rely on any property of the + /// function is called, so implementations cannot rely on any property of the /// contents of `buf` being true. It is recommended that *implementations* /// only write data to `buf` instead of reading its contents. /// @@ -752,7 +759,7 @@ pub trait Read { /// specified buffer `buf`. /// /// No guarantees are provided about the contents of `buf` when this - /// function is called, implementations cannot rely on any property of the + /// function is called, so implementations cannot rely on any property of the /// contents of `buf` being true. It is recommended that implementations /// only write data to `buf` instead of reading its contents. The /// documentation on [`read`] has a more detailed explanation on this @@ -803,30 +810,30 @@ pub trait Read { /// Pull some bytes from this source into the specified buffer. /// - /// This is equivalent to the [`read`](Read::read) method, except that it is passed a [`ReadBuf`] rather than `[u8]` to allow use + /// This is equivalent to the [`read`](Read::read) method, except that it is passed a [`BorrowedCursor`] rather than `[u8]` to allow use /// with uninitialized buffers. The new data will be appended to any existing contents of `buf`. /// /// The default implementation delegates to `read`. #[unstable(feature = "read_buf", issue = "78485")] - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> Result<()> { + fn read_buf(&mut self, buf: BorrowedCursor<'_>) -> Result<()> { default_read_buf(|b| self.read(b), buf) } - /// Read the exact number of bytes required to fill `buf`. + /// Read the exact number of bytes required to fill `cursor`. /// - /// This is equivalent to the [`read_exact`](Read::read_exact) method, except that it is passed a [`ReadBuf`] rather than `[u8]` to + /// This is equivalent to the [`read_exact`](Read::read_exact) method, except that it is passed a [`BorrowedCursor`] rather than `[u8]` to /// allow use with uninitialized buffers. #[unstable(feature = "read_buf", issue = "78485")] - fn read_buf_exact(&mut self, buf: &mut ReadBuf<'_>) -> Result<()> { - while buf.remaining() > 0 { - let prev_filled = buf.filled().len(); - match self.read_buf(buf) { + fn read_buf_exact(&mut self, mut cursor: BorrowedCursor<'_>) -> Result<()> { + while cursor.capacity() > 0 { + let prev_written = cursor.written(); + match self.read_buf(cursor.reborrow()) { Ok(()) => {} Err(e) if e.kind() == ErrorKind::Interrupted => continue, Err(e) => return Err(e), } - if buf.filled().len() == prev_filled { + if cursor.written() == prev_written { return Err(Error::new(ErrorKind::UnexpectedEof, "failed to fill buffer")); } } @@ -883,6 +890,10 @@ pub trait Read { /// The yielded item is [`Ok`] if a byte was successfully read and [`Err`] /// otherwise. EOF is mapped to returning [`None`] from this iterator. /// + /// The default implementation calls `read` for each byte, + /// which can be very inefficient for data that's not in memory, + /// such as [`File`]. Consider using a [`BufReader`] in such cases. + /// /// # Examples /// /// [`File`]s implement `Read`: @@ -895,10 +906,11 @@ pub trait Read { /// ```no_run /// use std::io; /// use std::io::prelude::*; + /// use std::io::BufReader; /// use std::fs::File; /// /// fn main() -> io::Result<()> { - /// let f = File::open("foo.txt")?; + /// let f = BufReader::new(File::open("foo.txt")?); /// /// for byte in f.bytes() { /// println!("{}", byte.unwrap()); @@ -1028,8 +1040,6 @@ pub trait Read { /// # Examples /// /// ```no_run -/// #![feature(io_read_to_string)] -/// /// # use std::io; /// fn main() -> io::Result<()> { /// let stdin = io::read_to_string(io::stdin())?; @@ -1038,7 +1048,7 @@ pub trait Read { /// Ok(()) /// } /// ``` -#[unstable(feature = "io_read_to_string", issue = "80218")] +#[stable(feature = "io_read_to_string", since = "1.65.0")] pub fn read_to_string<R: Read>(mut reader: R) -> Result<String> { let mut buf = String::new(); reader.read_to_string(&mut buf)?; @@ -2582,50 +2592,48 @@ impl<T: Read> Read for Take<T> { Ok(n) } - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> Result<()> { + fn read_buf(&mut self, mut buf: BorrowedCursor<'_>) -> Result<()> { // Don't call into inner reader at all at EOF because it may still block if self.limit == 0 { return Ok(()); } - let prev_filled = buf.filled_len(); - - if self.limit <= buf.remaining() as u64 { + if self.limit <= buf.capacity() as u64 { // if we just use an as cast to convert, limit may wrap around on a 32 bit target let limit = cmp::min(self.limit, usize::MAX as u64) as usize; - let extra_init = cmp::min(limit as usize, buf.initialized_len() - buf.filled_len()); + let extra_init = cmp::min(limit as usize, buf.init_ref().len()); // SAFETY: no uninit data is written to ibuf - let ibuf = unsafe { &mut buf.unfilled_mut()[..limit] }; + let ibuf = unsafe { &mut buf.as_mut()[..limit] }; - let mut sliced_buf = ReadBuf::uninit(ibuf); + let mut sliced_buf: BorrowedBuf<'_> = ibuf.into(); // SAFETY: extra_init bytes of ibuf are known to be initialized unsafe { - sliced_buf.assume_init(extra_init); + sliced_buf.set_init(extra_init); } - self.inner.read_buf(&mut sliced_buf)?; + let mut cursor = sliced_buf.unfilled(); + self.inner.read_buf(cursor.reborrow())?; - let new_init = sliced_buf.initialized_len(); - let filled = sliced_buf.filled_len(); + let new_init = cursor.init_ref().len(); + let filled = sliced_buf.len(); - // sliced_buf / ibuf must drop here + // cursor / sliced_buf / ibuf must drop here - // SAFETY: new_init bytes of buf's unfilled buffer have been initialized unsafe { - buf.assume_init(new_init); + // SAFETY: filled bytes have been filled and therefore initialized + buf.advance(filled); + // SAFETY: new_init bytes of buf's unfilled buffer have been initialized + buf.set_init(new_init); } - buf.add_filled(filled); - self.limit -= filled as u64; } else { - self.inner.read_buf(buf)?; - - //inner may unfill - self.limit -= buf.filled_len().saturating_sub(prev_filled) as u64; + let written = buf.written(); + self.inner.read_buf(buf.reborrow())?; + self.limit -= (buf.written() - written) as u64; } Ok(()) diff --git a/library/std/src/io/readbuf.rs b/library/std/src/io/readbuf.rs index 78d1113f8..4800eeda0 100644 --- a/library/std/src/io/readbuf.rs +++ b/library/std/src/io/readbuf.rs @@ -3,11 +3,12 @@ #[cfg(test)] mod tests; -use crate::cmp; use crate::fmt::{self, Debug, Formatter}; -use crate::mem::MaybeUninit; +use crate::io::{Result, Write}; +use crate::mem::{self, MaybeUninit}; +use crate::{cmp, ptr}; -/// A wrapper around a byte buffer that is incrementally filled and initialized. +/// A borrowed byte buffer which is incrementally filled and initialized. /// /// This type is a sort of "double cursor". It tracks three regions in the buffer: a region at the beginning of the /// buffer that has been logically filled with data, a region that has been initialized at some point but not yet @@ -20,230 +21,289 @@ use crate::mem::MaybeUninit; /// [ filled | unfilled ] /// [ initialized | uninitialized ] /// ``` -pub struct ReadBuf<'a> { - buf: &'a mut [MaybeUninit<u8>], +/// +/// A `BorrowedBuf` is created around some existing data (or capacity for data) via a unique reference +/// (`&mut`). The `BorrowedBuf` can be configured (e.g., using `clear` or `set_init`), but cannot be +/// directly written. To write into the buffer, use `unfilled` to create a `BorrowedCursor`. The cursor +/// has write-only access to the unfilled portion of the buffer (you can think of it as a +/// write-only iterator). +/// +/// The lifetime `'data` is a bound on the lifetime of the underlying data. +pub struct BorrowedBuf<'data> { + /// The buffer's underlying data. + buf: &'data mut [MaybeUninit<u8>], + /// The length of `self.buf` which is known to be filled. filled: usize, - initialized: usize, + /// The length of `self.buf` which is known to be initialized. + init: usize, } -impl Debug for ReadBuf<'_> { +impl Debug for BorrowedBuf<'_> { fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result { - f.debug_struct("ReadBuf") - .field("init", &self.initialized()) + f.debug_struct("BorrowedBuf") + .field("init", &self.init) .field("filled", &self.filled) .field("capacity", &self.capacity()) .finish() } } -impl<'a> ReadBuf<'a> { - /// Creates a new `ReadBuf` from a fully initialized buffer. +/// Create a new `BorrowedBuf` from a fully initialized slice. +impl<'data> From<&'data mut [u8]> for BorrowedBuf<'data> { #[inline] - pub fn new(buf: &'a mut [u8]) -> ReadBuf<'a> { - let len = buf.len(); + fn from(slice: &'data mut [u8]) -> BorrowedBuf<'data> { + let len = slice.len(); - ReadBuf { - //SAFETY: initialized data never becoming uninitialized is an invariant of ReadBuf - buf: unsafe { (buf as *mut [u8]).as_uninit_slice_mut().unwrap() }, + BorrowedBuf { + // SAFETY: initialized data never becoming uninitialized is an invariant of BorrowedBuf + buf: unsafe { (slice as *mut [u8]).as_uninit_slice_mut().unwrap() }, filled: 0, - initialized: len, + init: len, } } +} - /// Creates a new `ReadBuf` from a fully uninitialized buffer. - /// - /// Use `assume_init` if part of the buffer is known to be already initialized. +/// Create a new `BorrowedBuf` from an uninitialized buffer. +/// +/// Use `set_init` if part of the buffer is known to be already initialized. +impl<'data> From<&'data mut [MaybeUninit<u8>]> for BorrowedBuf<'data> { #[inline] - pub fn uninit(buf: &'a mut [MaybeUninit<u8>]) -> ReadBuf<'a> { - ReadBuf { buf, filled: 0, initialized: 0 } + fn from(buf: &'data mut [MaybeUninit<u8>]) -> BorrowedBuf<'data> { + BorrowedBuf { buf, filled: 0, init: 0 } } +} +impl<'data> BorrowedBuf<'data> { /// Returns the total capacity of the buffer. #[inline] pub fn capacity(&self) -> usize { self.buf.len() } + /// Returns the length of the filled part of the buffer. + #[inline] + pub fn len(&self) -> usize { + self.filled + } + + /// Returns the length of the initialized part of the buffer. + #[inline] + pub fn init_len(&self) -> usize { + self.init + } + /// Returns a shared reference to the filled portion of the buffer. #[inline] pub fn filled(&self) -> &[u8] { - //SAFETY: We only slice the filled part of the buffer, which is always valid + // SAFETY: We only slice the filled part of the buffer, which is always valid unsafe { MaybeUninit::slice_assume_init_ref(&self.buf[0..self.filled]) } } - /// Returns a mutable reference to the filled portion of the buffer. + /// Returns a cursor over the unfilled part of the buffer. #[inline] - pub fn filled_mut(&mut self) -> &mut [u8] { - //SAFETY: We only slice the filled part of the buffer, which is always valid - unsafe { MaybeUninit::slice_assume_init_mut(&mut self.buf[0..self.filled]) } + pub fn unfilled<'this>(&'this mut self) -> BorrowedCursor<'this> { + BorrowedCursor { + start: self.filled, + // SAFETY: we never assign into `BorrowedCursor::buf`, so treating its + // lifetime covariantly is safe. + buf: unsafe { + mem::transmute::<&'this mut BorrowedBuf<'data>, &'this mut BorrowedBuf<'this>>(self) + }, + } } - /// Returns a shared reference to the initialized portion of the buffer. + /// Clears the buffer, resetting the filled region to empty. /// - /// This includes the filled portion. + /// The number of initialized bytes is not changed, and the contents of the buffer are not modified. #[inline] - pub fn initialized(&self) -> &[u8] { - //SAFETY: We only slice the initialized part of the buffer, which is always valid - unsafe { MaybeUninit::slice_assume_init_ref(&self.buf[0..self.initialized]) } + pub fn clear(&mut self) -> &mut Self { + self.filled = 0; + self } - /// Returns a mutable reference to the initialized portion of the buffer. + /// Asserts that the first `n` bytes of the buffer are initialized. /// - /// This includes the filled portion. - #[inline] - pub fn initialized_mut(&mut self) -> &mut [u8] { - //SAFETY: We only slice the initialized part of the buffer, which is always valid - unsafe { MaybeUninit::slice_assume_init_mut(&mut self.buf[0..self.initialized]) } - } - - /// Returns a mutable reference to the unfilled part of the buffer without ensuring that it has been fully - /// initialized. + /// `BorrowedBuf` assumes that bytes are never de-initialized, so this method does nothing when called with fewer + /// bytes than are already known to be initialized. /// /// # Safety /// - /// The caller must not de-initialize portions of the buffer that have already been initialized. + /// The caller must ensure that the first `n` unfilled bytes of the buffer have already been initialized. #[inline] - pub unsafe fn unfilled_mut(&mut self) -> &mut [MaybeUninit<u8>] { - &mut self.buf[self.filled..] + pub unsafe fn set_init(&mut self, n: usize) -> &mut Self { + self.init = cmp::max(self.init, n); + self } +} - /// Returns a mutable reference to the uninitialized part of the buffer. +/// A writeable view of the unfilled portion of a [`BorrowedBuf`](BorrowedBuf). +/// +/// Provides access to the initialized and uninitialized parts of the underlying `BorrowedBuf`. +/// Data can be written directly to the cursor by using [`append`](BorrowedCursor::append) or +/// indirectly by getting a slice of part or all of the cursor and writing into the slice. In the +/// indirect case, the caller must call [`advance`](BorrowedCursor::advance) after writing to inform +/// the cursor how many bytes have been written. +/// +/// Once data is written to the cursor, it becomes part of the filled portion of the underlying +/// `BorrowedBuf` and can no longer be accessed or re-written by the cursor. I.e., the cursor tracks +/// the unfilled part of the underlying `BorrowedBuf`. +/// +/// The lifetime `'a` is a bound on the lifetime of the underlying buffer (which means it is a bound +/// on the data in that buffer by transitivity). +#[derive(Debug)] +pub struct BorrowedCursor<'a> { + /// The underlying buffer. + // Safety invariant: we treat the type of buf as covariant in the lifetime of `BorrowedBuf` when + // we create a `BorrowedCursor`. This is only safe if we never replace `buf` by assigning into + // it, so don't do that! + buf: &'a mut BorrowedBuf<'a>, + /// The length of the filled portion of the underlying buffer at the time of the cursor's + /// creation. + start: usize, +} + +impl<'a> BorrowedCursor<'a> { + /// Reborrow this cursor by cloning it with a smaller lifetime. /// - /// It is safe to uninitialize any of these bytes. + /// Since a cursor maintains unique access to its underlying buffer, the borrowed cursor is + /// not accessible while the new cursor exists. #[inline] - pub fn uninitialized_mut(&mut self) -> &mut [MaybeUninit<u8>] { - &mut self.buf[self.initialized..] + pub fn reborrow<'this>(&'this mut self) -> BorrowedCursor<'this> { + BorrowedCursor { + // SAFETY: we never assign into `BorrowedCursor::buf`, so treating its + // lifetime covariantly is safe. + buf: unsafe { + mem::transmute::<&'this mut BorrowedBuf<'a>, &'this mut BorrowedBuf<'this>>( + self.buf, + ) + }, + start: self.start, + } } - /// Returns a mutable reference to the unfilled part of the buffer, ensuring it is fully initialized. - /// - /// Since `ReadBuf` tracks the region of the buffer that has been initialized, this is effectively "free" after - /// the first use. + /// Returns the available space in the cursor. #[inline] - pub fn initialize_unfilled(&mut self) -> &mut [u8] { - // should optimize out the assertion - self.initialize_unfilled_to(self.remaining()) + pub fn capacity(&self) -> usize { + self.buf.capacity() - self.buf.filled } - /// Returns a mutable reference to the first `n` bytes of the unfilled part of the buffer, ensuring it is - /// fully initialized. + /// Returns the number of bytes written to this cursor since it was created from a `BorrowedBuf`. /// - /// # Panics - /// - /// Panics if `self.remaining()` is less than `n`. + /// Note that if this cursor is a reborrowed clone of another, then the count returned is the + /// count written via either cursor, not the count since the cursor was reborrowed. #[inline] - pub fn initialize_unfilled_to(&mut self, n: usize) -> &mut [u8] { - assert!(self.remaining() >= n); - - let extra_init = self.initialized - self.filled; - // If we don't have enough initialized, do zeroing - if n > extra_init { - let uninit = n - extra_init; - let unfilled = &mut self.uninitialized_mut()[0..uninit]; - - for byte in unfilled.iter_mut() { - byte.write(0); - } - - // SAFETY: we just initialized uninit bytes, and the previous bytes were already init - unsafe { - self.assume_init(n); - } - } - - let filled = self.filled; + pub fn written(&self) -> usize { + self.buf.filled - self.start + } - &mut self.initialized_mut()[filled..filled + n] + /// Returns a shared reference to the initialized portion of the cursor. + #[inline] + pub fn init_ref(&self) -> &[u8] { + // SAFETY: We only slice the initialized part of the buffer, which is always valid + unsafe { MaybeUninit::slice_assume_init_ref(&self.buf.buf[self.buf.filled..self.buf.init]) } } - /// Returns the number of bytes at the end of the slice that have not yet been filled. + /// Returns a mutable reference to the initialized portion of the cursor. #[inline] - pub fn remaining(&self) -> usize { - self.capacity() - self.filled + pub fn init_mut(&mut self) -> &mut [u8] { + // SAFETY: We only slice the initialized part of the buffer, which is always valid + unsafe { + MaybeUninit::slice_assume_init_mut(&mut self.buf.buf[self.buf.filled..self.buf.init]) + } } - /// Clears the buffer, resetting the filled region to empty. + /// Returns a mutable reference to the uninitialized part of the cursor. /// - /// The number of initialized bytes is not changed, and the contents of the buffer are not modified. + /// It is safe to uninitialize any of these bytes. #[inline] - pub fn clear(&mut self) -> &mut Self { - self.set_filled(0) // The assertion in `set_filled` is optimized out + pub fn uninit_mut(&mut self) -> &mut [MaybeUninit<u8>] { + &mut self.buf.buf[self.buf.init..] } - /// Increases the size of the filled region of the buffer. + /// Returns a mutable reference to the whole cursor. /// - /// The number of initialized bytes is not changed. - /// - /// # Panics + /// # Safety /// - /// Panics if the filled region of the buffer would become larger than the initialized region. + /// The caller must not uninitialize any bytes in the initialized portion of the cursor. #[inline] - pub fn add_filled(&mut self, n: usize) -> &mut Self { - self.set_filled(self.filled + n) + pub unsafe fn as_mut(&mut self) -> &mut [MaybeUninit<u8>] { + &mut self.buf.buf[self.buf.filled..] } - /// Sets the size of the filled region of the buffer. - /// - /// The number of initialized bytes is not changed. + /// Advance the cursor by asserting that `n` bytes have been filled. /// - /// Note that this can be used to *shrink* the filled region of the buffer in addition to growing it (for - /// example, by a `Read` implementation that compresses data in-place). + /// After advancing, the `n` bytes are no longer accessible via the cursor and can only be + /// accessed via the underlying buffer. I.e., the buffer's filled portion grows by `n` elements + /// and its unfilled portion (and the capacity of this cursor) shrinks by `n` elements. /// - /// # Panics + /// # Safety /// - /// Panics if the filled region of the buffer would become larger than the initialized region. + /// The caller must ensure that the first `n` bytes of the cursor have been properly + /// initialised. + #[inline] + pub unsafe fn advance(&mut self, n: usize) -> &mut Self { + self.buf.filled += n; + self.buf.init = cmp::max(self.buf.init, self.buf.filled); + self + } + + /// Initializes all bytes in the cursor. #[inline] - pub fn set_filled(&mut self, n: usize) -> &mut Self { - assert!(n <= self.initialized); + pub fn ensure_init(&mut self) -> &mut Self { + let uninit = self.uninit_mut(); + // SAFETY: 0 is a valid value for MaybeUninit<u8> and the length matches the allocation + // since it is comes from a slice reference. + unsafe { + ptr::write_bytes(uninit.as_mut_ptr(), 0, uninit.len()); + } + self.buf.init = self.buf.capacity(); - self.filled = n; self } - /// Asserts that the first `n` unfilled bytes of the buffer are initialized. + /// Asserts that the first `n` unfilled bytes of the cursor are initialized. /// - /// `ReadBuf` assumes that bytes are never de-initialized, so this method does nothing when called with fewer - /// bytes than are already known to be initialized. + /// `BorrowedBuf` assumes that bytes are never de-initialized, so this method does nothing when + /// called with fewer bytes than are already known to be initialized. /// /// # Safety /// - /// The caller must ensure that the first `n` unfilled bytes of the buffer have already been initialized. + /// The caller must ensure that the first `n` bytes of the buffer have already been initialized. #[inline] - pub unsafe fn assume_init(&mut self, n: usize) -> &mut Self { - self.initialized = cmp::max(self.initialized, self.filled + n); + pub unsafe fn set_init(&mut self, n: usize) -> &mut Self { + self.buf.init = cmp::max(self.buf.init, self.buf.filled + n); self } - /// Appends data to the buffer, advancing the written position and possibly also the initialized position. + /// Appends data to the cursor, advancing position within its buffer. /// /// # Panics /// - /// Panics if `self.remaining()` is less than `buf.len()`. + /// Panics if `self.capacity()` is less than `buf.len()`. #[inline] pub fn append(&mut self, buf: &[u8]) { - assert!(self.remaining() >= buf.len()); + assert!(self.capacity() >= buf.len()); // SAFETY: we do not de-initialize any of the elements of the slice unsafe { - MaybeUninit::write_slice(&mut self.unfilled_mut()[..buf.len()], buf); + MaybeUninit::write_slice(&mut self.as_mut()[..buf.len()], buf); } // SAFETY: We just added the entire contents of buf to the filled section. unsafe { - self.assume_init(buf.len()); + self.set_init(buf.len()); } - self.add_filled(buf.len()); + self.buf.filled += buf.len(); } +} - /// Returns the amount of bytes that have been filled. - #[inline] - pub fn filled_len(&self) -> usize { - self.filled +impl<'a> Write for BorrowedCursor<'a> { + fn write(&mut self, buf: &[u8]) -> Result<usize> { + self.append(buf); + Ok(buf.len()) } - /// Returns the amount of bytes that have been initialized. - #[inline] - pub fn initialized_len(&self) -> usize { - self.initialized + fn flush(&mut self) -> Result<()> { + Ok(()) } } diff --git a/library/std/src/io/readbuf/tests.rs b/library/std/src/io/readbuf/tests.rs index 3b7a5a56d..cc1b423f2 100644 --- a/library/std/src/io/readbuf/tests.rs +++ b/library/std/src/io/readbuf/tests.rs @@ -1,181 +1,175 @@ -use super::ReadBuf; +use super::BorrowedBuf; use crate::mem::MaybeUninit; -/// Test that ReadBuf has the correct numbers when created with new +/// Test that BorrowedBuf has the correct numbers when created with new #[test] fn new() { - let mut buf = [0; 16]; - let rbuf = ReadBuf::new(&mut buf); + let buf: &mut [_] = &mut [0; 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); - assert_eq!(rbuf.filled_len(), 0); - assert_eq!(rbuf.initialized_len(), 16); + assert_eq!(rbuf.filled().len(), 0); + assert_eq!(rbuf.init_len(), 16); assert_eq!(rbuf.capacity(), 16); - assert_eq!(rbuf.remaining(), 16); + assert_eq!(rbuf.unfilled().capacity(), 16); } -/// Test that ReadBuf has the correct numbers when created with uninit +/// Test that BorrowedBuf has the correct numbers when created with uninit #[test] fn uninit() { - let mut buf = [MaybeUninit::uninit(); 16]; - let rbuf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); - assert_eq!(rbuf.filled_len(), 0); - assert_eq!(rbuf.initialized_len(), 0); + assert_eq!(rbuf.filled().len(), 0); + assert_eq!(rbuf.init_len(), 0); assert_eq!(rbuf.capacity(), 16); - assert_eq!(rbuf.remaining(), 16); + assert_eq!(rbuf.unfilled().capacity(), 16); } #[test] fn initialize_unfilled() { - let mut buf = [MaybeUninit::uninit(); 16]; - let mut rbuf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); - rbuf.initialize_unfilled(); + rbuf.unfilled().ensure_init(); - assert_eq!(rbuf.initialized_len(), 16); + assert_eq!(rbuf.init_len(), 16); } #[test] -fn initialize_unfilled_to() { - let mut buf = [MaybeUninit::uninit(); 16]; - let mut rbuf = ReadBuf::uninit(&mut buf); +fn addvance_filled() { + let buf: &mut [_] = &mut [0; 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); - rbuf.initialize_unfilled_to(8); - - assert_eq!(rbuf.initialized_len(), 8); - - rbuf.initialize_unfilled_to(4); - - assert_eq!(rbuf.initialized_len(), 8); - - rbuf.set_filled(8); - - rbuf.initialize_unfilled_to(6); - - assert_eq!(rbuf.initialized_len(), 14); - - rbuf.initialize_unfilled_to(8); - - assert_eq!(rbuf.initialized_len(), 16); -} - -#[test] -fn add_filled() { - let mut buf = [0; 16]; - let mut rbuf = ReadBuf::new(&mut buf); - - rbuf.add_filled(1); - - assert_eq!(rbuf.filled_len(), 1); - assert_eq!(rbuf.remaining(), 15); -} - -#[test] -#[should_panic] -fn add_filled_panic() { - let mut buf = [MaybeUninit::uninit(); 16]; - let mut rbuf = ReadBuf::uninit(&mut buf); - - rbuf.add_filled(1); -} - -#[test] -fn set_filled() { - let mut buf = [0; 16]; - let mut rbuf = ReadBuf::new(&mut buf); - - rbuf.set_filled(16); - - assert_eq!(rbuf.filled_len(), 16); - assert_eq!(rbuf.remaining(), 0); - - rbuf.set_filled(6); - - assert_eq!(rbuf.filled_len(), 6); - assert_eq!(rbuf.remaining(), 10); -} - -#[test] -#[should_panic] -fn set_filled_panic() { - let mut buf = [MaybeUninit::uninit(); 16]; - let mut rbuf = ReadBuf::uninit(&mut buf); + unsafe { + rbuf.unfilled().advance(1); + } - rbuf.set_filled(16); + assert_eq!(rbuf.filled().len(), 1); + assert_eq!(rbuf.unfilled().capacity(), 15); } #[test] fn clear() { - let mut buf = [255; 16]; - let mut rbuf = ReadBuf::new(&mut buf); + let buf: &mut [_] = &mut [255; 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); - rbuf.set_filled(16); + unsafe { + rbuf.unfilled().advance(16); + } - assert_eq!(rbuf.filled_len(), 16); - assert_eq!(rbuf.remaining(), 0); + assert_eq!(rbuf.filled().len(), 16); + assert_eq!(rbuf.unfilled().capacity(), 0); rbuf.clear(); - assert_eq!(rbuf.filled_len(), 0); - assert_eq!(rbuf.remaining(), 16); + assert_eq!(rbuf.filled().len(), 0); + assert_eq!(rbuf.unfilled().capacity(), 16); - assert_eq!(rbuf.initialized(), [255; 16]); + assert_eq!(rbuf.unfilled().init_ref(), [255; 16]); } #[test] -fn assume_init() { - let mut buf = [MaybeUninit::uninit(); 16]; - let mut rbuf = ReadBuf::uninit(&mut buf); +fn set_init() { + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); unsafe { - rbuf.assume_init(8); + rbuf.set_init(8); } - assert_eq!(rbuf.initialized_len(), 8); + assert_eq!(rbuf.init_len(), 8); - rbuf.add_filled(4); + unsafe { + rbuf.unfilled().advance(4); + } unsafe { - rbuf.assume_init(2); + rbuf.set_init(2); } - assert_eq!(rbuf.initialized_len(), 8); + assert_eq!(rbuf.init_len(), 8); unsafe { - rbuf.assume_init(8); + rbuf.set_init(8); } - assert_eq!(rbuf.initialized_len(), 12); + assert_eq!(rbuf.init_len(), 8); } #[test] fn append() { - let mut buf = [MaybeUninit::new(255); 16]; - let mut rbuf = ReadBuf::uninit(&mut buf); + let buf: &mut [_] = &mut [MaybeUninit::new(255); 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); - rbuf.append(&[0; 8]); + rbuf.unfilled().append(&[0; 8]); - assert_eq!(rbuf.initialized_len(), 8); - assert_eq!(rbuf.filled_len(), 8); + assert_eq!(rbuf.init_len(), 8); + assert_eq!(rbuf.filled().len(), 8); assert_eq!(rbuf.filled(), [0; 8]); rbuf.clear(); - rbuf.append(&[1; 16]); + rbuf.unfilled().append(&[1; 16]); - assert_eq!(rbuf.initialized_len(), 16); - assert_eq!(rbuf.filled_len(), 16); + assert_eq!(rbuf.init_len(), 16); + assert_eq!(rbuf.filled().len(), 16); assert_eq!(rbuf.filled(), [1; 16]); } #[test] -fn filled_mut() { - let mut buf = [0; 16]; - let mut rbuf = ReadBuf::new(&mut buf); +fn reborrow_written() { + let buf: &mut [_] = &mut [MaybeUninit::new(0); 32]; + let mut buf: BorrowedBuf<'_> = buf.into(); + + let mut cursor = buf.unfilled(); + cursor.append(&[1; 16]); + + let mut cursor2 = cursor.reborrow(); + cursor2.append(&[2; 16]); + + assert_eq!(cursor2.written(), 32); + assert_eq!(cursor.written(), 32); + + assert_eq!(buf.unfilled().written(), 0); + assert_eq!(buf.init_len(), 32); + assert_eq!(buf.filled().len(), 32); + let filled = buf.filled(); + assert_eq!(&filled[..16], [1; 16]); + assert_eq!(&filled[16..], [2; 16]); +} + +#[test] +fn cursor_set_init() { + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 16]; + let mut rbuf: BorrowedBuf<'_> = buf.into(); + + unsafe { + rbuf.unfilled().set_init(8); + } - rbuf.add_filled(8); + assert_eq!(rbuf.init_len(), 8); + assert_eq!(rbuf.unfilled().init_ref().len(), 8); + assert_eq!(rbuf.unfilled().init_mut().len(), 8); + assert_eq!(rbuf.unfilled().uninit_mut().len(), 8); + assert_eq!(unsafe { rbuf.unfilled().as_mut() }.len(), 16); + + unsafe { + rbuf.unfilled().advance(4); + } - let filled = rbuf.filled().to_vec(); + unsafe { + rbuf.unfilled().set_init(2); + } + + assert_eq!(rbuf.init_len(), 8); + + unsafe { + rbuf.unfilled().set_init(8); + } - assert_eq!(&*filled, &*rbuf.filled_mut()); + assert_eq!(rbuf.init_len(), 12); + assert_eq!(rbuf.unfilled().init_ref().len(), 8); + assert_eq!(rbuf.unfilled().init_mut().len(), 8); + assert_eq!(rbuf.unfilled().uninit_mut().len(), 4); + assert_eq!(unsafe { rbuf.unfilled().as_mut() }.len(), 12); } diff --git a/library/std/src/io/stdio.rs b/library/std/src/io/stdio.rs index 4d3736f79..1141a957d 100644 --- a/library/std/src/io/stdio.rs +++ b/library/std/src/io/stdio.rs @@ -7,8 +7,8 @@ use crate::io::prelude::*; use crate::cell::{Cell, RefCell}; use crate::fmt; +use crate::fs::File; use crate::io::{self, BufReader, IoSlice, IoSliceMut, LineWriter, Lines}; -use crate::pin::Pin; use crate::sync::atomic::{AtomicBool, Ordering}; use crate::sync::{Arc, Mutex, MutexGuard, OnceLock}; use crate::sys::stdio; @@ -526,7 +526,7 @@ pub struct Stdout { // FIXME: this should be LineWriter or BufWriter depending on the state of // stdout (tty or not). Note that if this is not line buffered it // should also flush-on-panic or some form of flush-on-abort. - inner: Pin<&'static ReentrantMutex<RefCell<LineWriter<StdoutRaw>>>>, + inner: &'static ReentrantMutex<RefCell<LineWriter<StdoutRaw>>>, } /// A locked reference to the [`Stdout`] handle. @@ -603,22 +603,27 @@ static STDOUT: OnceLock<ReentrantMutex<RefCell<LineWriter<StdoutRaw>>>> = OnceLo #[stable(feature = "rust1", since = "1.0.0")] pub fn stdout() -> Stdout { Stdout { - inner: Pin::static_ref(&STDOUT).get_or_init_pin( - || unsafe { ReentrantMutex::new(RefCell::new(LineWriter::new(stdout_raw()))) }, - |mutex| unsafe { mutex.init() }, - ), + inner: STDOUT + .get_or_init(|| ReentrantMutex::new(RefCell::new(LineWriter::new(stdout_raw())))), } } +// Flush the data and disable buffering during shutdown +// by replacing the line writer by one with zero +// buffering capacity. pub fn cleanup() { - if let Some(instance) = STDOUT.get() { - // Flush the data and disable buffering during shutdown - // by replacing the line writer by one with zero - // buffering capacity. + let mut initialized = false; + let stdout = STDOUT.get_or_init(|| { + initialized = true; + ReentrantMutex::new(RefCell::new(LineWriter::with_capacity(0, stdout_raw()))) + }); + + if !initialized { + // The buffer was previously initialized, overwrite it here. // We use try_lock() instead of lock(), because someone // might have leaked a StdoutLock, which would // otherwise cause a deadlock here. - if let Some(lock) = Pin::static_ref(instance).try_lock() { + if let Some(lock) = stdout.try_lock() { *lock.borrow_mut() = LineWriter::with_capacity(0, stdout_raw()); } } @@ -761,7 +766,7 @@ impl fmt::Debug for StdoutLock<'_> { /// standard library or via raw Windows API calls, will fail. #[stable(feature = "rust1", since = "1.0.0")] pub struct Stderr { - inner: Pin<&'static ReentrantMutex<RefCell<StderrRaw>>>, + inner: &'static ReentrantMutex<RefCell<StderrRaw>>, } /// A locked reference to the [`Stderr`] handle. @@ -834,16 +839,12 @@ pub struct StderrLock<'a> { #[stable(feature = "rust1", since = "1.0.0")] pub fn stderr() -> Stderr { // Note that unlike `stdout()` we don't use `at_exit` here to register a - // destructor. Stderr is not buffered , so there's no need to run a + // destructor. Stderr is not buffered, so there's no need to run a // destructor for flushing the buffer - static INSTANCE: OnceLock<ReentrantMutex<RefCell<StderrRaw>>> = OnceLock::new(); + static INSTANCE: ReentrantMutex<RefCell<StderrRaw>> = + ReentrantMutex::new(RefCell::new(stderr_raw())); - Stderr { - inner: Pin::static_ref(&INSTANCE).get_or_init_pin( - || unsafe { ReentrantMutex::new(RefCell::new(stderr_raw())) }, - |mutex| unsafe { mutex.init() }, - ), - } + Stderr { inner: &INSTANCE } } impl Stderr { @@ -986,17 +987,31 @@ pub fn set_output_capture(sink: Option<LocalStream>) -> Option<LocalStream> { /// otherwise. `label` identifies the stream in a panic message. /// /// This function is used to print error messages, so it takes extra -/// care to avoid causing a panic when `local_s` is unusable. -/// For instance, if the TLS key for the local stream is -/// already destroyed, or if the local stream is locked by another -/// thread, it will just fall back to the global stream. +/// care to avoid causing a panic when `OUTPUT_CAPTURE` is unusable. +/// For instance, if the TLS key for output capturing is already destroyed, or +/// if the local stream is in use by another thread, it will just fall back to +/// the global stream. /// /// However, if the actual I/O causes an error, this function does panic. +/// +/// Writing to non-blocking stdout/stderr can cause an error, which will lead +/// this function to panic. fn print_to<T>(args: fmt::Arguments<'_>, global_s: fn() -> T, label: &str) where T: Write, { - if OUTPUT_CAPTURE_USED.load(Ordering::Relaxed) + if print_to_buffer_if_capture_used(args) { + // Successfully wrote to capture buffer. + return; + } + + if let Err(e) = global_s().write_fmt(args) { + panic!("failed printing to {label}: {e}"); + } +} + +fn print_to_buffer_if_capture_used(args: fmt::Arguments<'_>) -> bool { + OUTPUT_CAPTURE_USED.load(Ordering::Relaxed) && OUTPUT_CAPTURE.try_with(|s| { // Note that we completely remove a local sink to write to in case // our printing recursively panics/prints, so the recursive @@ -1006,16 +1021,49 @@ where s.set(Some(w)); }) }) == Ok(Some(())) - { - // Successfully wrote to capture buffer. +} + +/// Used by impl Termination for Result to print error after `main` or a test +/// has returned. Should avoid panicking, although we can't help it if one of +/// the Display impls inside args decides to. +pub(crate) fn attempt_print_to_stderr(args: fmt::Arguments<'_>) { + if print_to_buffer_if_capture_used(args) { return; } - if let Err(e) = global_s().write_fmt(args) { - panic!("failed printing to {label}: {e}"); - } + // Ignore error if the write fails, for example because stderr is already + // closed. There is not much point panicking at this point. + let _ = stderr().write_fmt(args); +} + +/// Trait to determine if a descriptor/handle refers to a terminal/tty. +#[unstable(feature = "is_terminal", issue = "98070")] +pub trait IsTerminal: crate::sealed::Sealed { + /// Returns `true` if the descriptor/handle refers to a terminal/tty. + /// + /// On platforms where Rust does not know how to detect a terminal yet, this will return + /// `false`. This will also return `false` if an unexpected error occurred, such as from + /// passing an invalid file descriptor. + fn is_terminal(&self) -> bool; +} + +macro_rules! impl_is_terminal { + ($($t:ty),*$(,)?) => {$( + #[unstable(feature = "sealed", issue = "none")] + impl crate::sealed::Sealed for $t {} + + #[unstable(feature = "is_terminal", issue = "98070")] + impl IsTerminal for $t { + #[inline] + fn is_terminal(&self) -> bool { + crate::sys::io::is_terminal(self) + } + } + )*} } +impl_is_terminal!(File, Stdin, StdinLock<'_>, Stdout, StdoutLock<'_>, Stderr, StderrLock<'_>); + #[unstable( feature = "print_internals", reason = "implementation detail which may disappear or be replaced at any time", diff --git a/library/std/src/io/tests.rs b/library/std/src/io/tests.rs index f357f33ec..f4a886d88 100644 --- a/library/std/src/io/tests.rs +++ b/library/std/src/io/tests.rs @@ -1,4 +1,4 @@ -use super::{repeat, Cursor, ReadBuf, SeekFrom}; +use super::{repeat, BorrowedBuf, Cursor, SeekFrom}; use crate::cmp::{self, min}; use crate::io::{self, IoSlice, IoSliceMut}; use crate::io::{BufRead, BufReader, Read, Seek, Write}; @@ -94,7 +94,7 @@ fn read_to_end() { assert_eq!(c.read_to_end(&mut v).unwrap(), 1); assert_eq!(v, b"1"); - let cap = 1024 * 1024; + let cap = if cfg!(miri) { 1024 } else { 1024 * 1024 }; let data = (0..cap).map(|i| (i / 3) as u8).collect::<Vec<_>>(); let mut v = Vec::new(); let (a, b) = data.split_at(data.len() / 2); @@ -159,24 +159,24 @@ fn read_exact_slice() { #[test] fn read_buf_exact() { - let mut buf = [0; 4]; - let mut buf = ReadBuf::new(&mut buf); + let buf: &mut [_] = &mut [0; 4]; + let mut buf: BorrowedBuf<'_> = buf.into(); let mut c = Cursor::new(&b""[..]); - assert_eq!(c.read_buf_exact(&mut buf).unwrap_err().kind(), io::ErrorKind::UnexpectedEof); + assert_eq!(c.read_buf_exact(buf.unfilled()).unwrap_err().kind(), io::ErrorKind::UnexpectedEof); let mut c = Cursor::new(&b"123456789"[..]); - c.read_buf_exact(&mut buf).unwrap(); + c.read_buf_exact(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), b"1234"); buf.clear(); - c.read_buf_exact(&mut buf).unwrap(); + c.read_buf_exact(buf.unfilled()).unwrap(); assert_eq!(buf.filled(), b"5678"); buf.clear(); - assert_eq!(c.read_buf_exact(&mut buf).unwrap_err().kind(), io::ErrorKind::UnexpectedEof); + assert_eq!(c.read_buf_exact(buf.unfilled()).unwrap_err().kind(), io::ErrorKind::UnexpectedEof); } #[test] @@ -309,6 +309,7 @@ fn chain_zero_length_read_is_not_eof() { #[bench] #[cfg_attr(target_os = "emscripten", ignore)] +#[cfg_attr(miri, ignore)] // Miri isn't fast... fn bench_read_to_end(b: &mut test::Bencher) { b.iter(|| { let mut lr = repeat(1).take(10000000); @@ -614,10 +615,10 @@ fn bench_take_read(b: &mut test::Bencher) { #[bench] fn bench_take_read_buf(b: &mut test::Bencher) { b.iter(|| { - let mut buf = [MaybeUninit::uninit(); 64]; + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 64]; - let mut rbuf = ReadBuf::uninit(&mut buf); + let mut buf: BorrowedBuf<'_> = buf.into(); - [255; 128].take(64).read_buf(&mut rbuf).unwrap(); + [255; 128].take(64).read_buf(buf.unfilled()).unwrap(); }); } diff --git a/library/std/src/io/util.rs b/library/std/src/io/util.rs index c1300cd67..f076ee092 100644 --- a/library/std/src/io/util.rs +++ b/library/std/src/io/util.rs @@ -5,7 +5,7 @@ mod tests; use crate::fmt; use crate::io::{ - self, BufRead, IoSlice, IoSliceMut, Read, ReadBuf, Seek, SeekFrom, SizeHint, Write, + self, BorrowedCursor, BufRead, IoSlice, IoSliceMut, Read, Seek, SeekFrom, SizeHint, Write, }; /// A reader which is always at EOF. @@ -47,7 +47,7 @@ impl Read for Empty { } #[inline] - fn read_buf(&mut self, _buf: &mut ReadBuf<'_>) -> io::Result<()> { + fn read_buf(&mut self, _cursor: BorrowedCursor<'_>) -> io::Result<()> { Ok(()) } } @@ -130,21 +130,19 @@ impl Read for Repeat { Ok(buf.len()) } - fn read_buf(&mut self, buf: &mut ReadBuf<'_>) -> io::Result<()> { + fn read_buf(&mut self, mut buf: BorrowedCursor<'_>) -> io::Result<()> { // SAFETY: No uninit bytes are being written - for slot in unsafe { buf.unfilled_mut() } { + for slot in unsafe { buf.as_mut() } { slot.write(self.byte); } - let remaining = buf.remaining(); + let remaining = buf.capacity(); // SAFETY: the entire unfilled portion of buf has been initialized unsafe { - buf.assume_init(remaining); + buf.advance(remaining); } - buf.add_filled(remaining); - Ok(()) } diff --git a/library/std/src/io/util/tests.rs b/library/std/src/io/util/tests.rs index 08972a59a..ce5e2c9da 100644 --- a/library/std/src/io/util/tests.rs +++ b/library/std/src/io/util/tests.rs @@ -1,7 +1,7 @@ use crate::cmp::{max, min}; use crate::io::prelude::*; use crate::io::{ - copy, empty, repeat, sink, BufWriter, Empty, ReadBuf, Repeat, Result, SeekFrom, Sink, + copy, empty, repeat, sink, BorrowedBuf, BufWriter, Empty, Repeat, Result, SeekFrom, Sink, DEFAULT_BUF_SIZE, }; @@ -79,29 +79,29 @@ fn empty_reads() { assert_eq!(e.read(&mut [0; 1024]).unwrap(), 0); assert_eq!(e.by_ref().read(&mut [0; 1024]).unwrap(), 0); - let mut buf = []; - let mut buf = ReadBuf::uninit(&mut buf); - e.read_buf(&mut buf).unwrap(); - assert_eq!(buf.filled_len(), 0); - assert_eq!(buf.initialized_len(), 0); - - let mut buf = [MaybeUninit::uninit()]; - let mut buf = ReadBuf::uninit(&mut buf); - e.read_buf(&mut buf).unwrap(); - assert_eq!(buf.filled_len(), 0); - assert_eq!(buf.initialized_len(), 0); - - let mut buf = [MaybeUninit::uninit(); 1024]; - let mut buf = ReadBuf::uninit(&mut buf); - e.read_buf(&mut buf).unwrap(); - assert_eq!(buf.filled_len(), 0); - assert_eq!(buf.initialized_len(), 0); - - let mut buf = [MaybeUninit::uninit(); 1024]; - let mut buf = ReadBuf::uninit(&mut buf); - e.by_ref().read_buf(&mut buf).unwrap(); - assert_eq!(buf.filled_len(), 0); - assert_eq!(buf.initialized_len(), 0); + let buf: &mut [MaybeUninit<_>] = &mut []; + let mut buf: BorrowedBuf<'_> = buf.into(); + e.read_buf(buf.unfilled()).unwrap(); + assert_eq!(buf.len(), 0); + assert_eq!(buf.init_len(), 0); + + let buf: &mut [_] = &mut [MaybeUninit::uninit()]; + let mut buf: BorrowedBuf<'_> = buf.into(); + e.read_buf(buf.unfilled()).unwrap(); + assert_eq!(buf.len(), 0); + assert_eq!(buf.init_len(), 0); + + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 1024]; + let mut buf: BorrowedBuf<'_> = buf.into(); + e.read_buf(buf.unfilled()).unwrap(); + assert_eq!(buf.len(), 0); + assert_eq!(buf.init_len(), 0); + + let buf: &mut [_] = &mut [MaybeUninit::uninit(); 1024]; + let mut buf: BorrowedBuf<'_> = buf.into(); + e.by_ref().read_buf(buf.unfilled()).unwrap(); + assert_eq!(buf.len(), 0); + assert_eq!(buf.init_len(), 0); } #[test] diff --git a/library/std/src/keyword_docs.rs b/library/std/src/keyword_docs.rs index 7157b5af0..e35145c4a 100644 --- a/library/std/src/keyword_docs.rs +++ b/library/std/src/keyword_docs.rs @@ -1867,11 +1867,15 @@ mod type_keyword {} /// Code or interfaces whose [memory safety] cannot be verified by the type /// system. /// -/// The `unsafe` keyword has two uses: to declare the existence of contracts the -/// compiler can't check (`unsafe fn` and `unsafe trait`), and to declare that a -/// programmer has checked that these contracts have been upheld (`unsafe {}` -/// and `unsafe impl`, but also `unsafe fn` -- see below). They are not mutually -/// exclusive, as can be seen in `unsafe fn`. +/// The `unsafe` keyword has two uses: +/// - to declare the existence of contracts the compiler can't check (`unsafe fn` and `unsafe +/// trait`), +/// - and to declare that a programmer has checked that these contracts have been upheld (`unsafe +/// {}` and `unsafe impl`, but also `unsafe fn` -- see below). +/// +/// They are not mutually exclusive, as can be seen in `unsafe fn`: the body of an `unsafe fn` is, +/// by default, treated like an unsafe block. The `unsafe_op_in_unsafe_fn` lint can be enabled to +/// change that. /// /// # Unsafe abilities /// @@ -1914,14 +1918,14 @@ mod type_keyword {} /// - `unsafe impl`: the contract necessary to implement the trait has been /// checked by the programmer and is guaranteed to be respected. /// -/// `unsafe fn` also acts like an `unsafe {}` block +/// By default, `unsafe fn` also acts like an `unsafe {}` block /// around the code inside the function. This means it is not just a signal to /// the caller, but also promises that the preconditions for the operations -/// inside the function are upheld. Mixing these two meanings can be confusing -/// and [proposal]s exist to use `unsafe {}` blocks inside such functions when -/// making `unsafe` operations. +/// inside the function are upheld. Mixing these two meanings can be confusing, so the +/// `unsafe_op_in_unsafe_fn` lint can be enabled to warn against that and require explicit unsafe +/// blocks even inside `unsafe fn`. /// -/// See the [Rustnomicon] and the [Reference] for more informations. +/// See the [Rustnomicon] and the [Reference] for more information. /// /// # Examples /// @@ -1987,13 +1991,16 @@ mod type_keyword {} /// /// ```rust /// # #![allow(dead_code)] +/// #![deny(unsafe_op_in_unsafe_fn)] +/// /// /// Dereference the given pointer. /// /// /// /// # Safety /// /// /// /// `ptr` must be aligned and must not be dangling. /// unsafe fn deref_unchecked(ptr: *const i32) -> i32 { -/// *ptr +/// // SAFETY: the caller is required to ensure that `ptr` is aligned and dereferenceable. +/// unsafe { *ptr } /// } /// /// let a = 3; @@ -2003,35 +2010,118 @@ mod type_keyword {} /// unsafe { assert_eq!(*b, deref_unchecked(b)); }; /// ``` /// -/// Traits marked as `unsafe` must be [`impl`]emented using `unsafe impl`. This -/// makes a guarantee to other `unsafe` code that the implementation satisfies -/// the trait's safety contract. The [Send] and [Sync] traits are examples of -/// this behaviour in the standard library. +/// ## `unsafe` and traits +/// +/// The interactions of `unsafe` and traits can be surprising, so let us contrast the +/// two combinations of safe `fn` in `unsafe trait` and `unsafe fn` in safe trait using two +/// examples: +/// +/// ```rust +/// /// # Safety +/// /// +/// /// `make_even` must return an even number. +/// unsafe trait MakeEven { +/// fn make_even(&self) -> i32; +/// } +/// +/// // SAFETY: Our `make_even` always returns something even. +/// unsafe impl MakeEven for i32 { +/// fn make_even(&self) -> i32 { +/// self << 1 +/// } +/// } +/// +/// fn use_make_even(x: impl MakeEven) { +/// if x.make_even() % 2 == 1 { +/// // SAFETY: this can never happen, because all `MakeEven` implementations +/// // ensure that `make_even` returns something even. +/// unsafe { std::hint::unreachable_unchecked() }; +/// } +/// } +/// ``` +/// +/// Note how the safety contract of the trait is upheld by the implementation, and is itself used to +/// uphold the safety contract of the unsafe function `unreachable_unchecked` called by +/// `use_make_even`. `make_even` itself is a safe function because its *callers* do not have to +/// worry about any contract, only the *implementation* of `MakeEven` is required to uphold a +/// certain contract. `use_make_even` is safe because it can use the promise made by `MakeEven` +/// implementations to uphold the safety contract of the `unsafe fn unreachable_unchecked` it calls. +/// +/// It is also possible to have `unsafe fn` in a regular safe `trait`: /// /// ```rust -/// /// Implementors of this trait must guarantee an element is always -/// /// accessible with index 3. -/// unsafe trait ThreeIndexable<T> { -/// /// Returns a reference to the element with index 3 in `&self`. -/// fn three(&self) -> &T; +/// # #![feature(never_type)] +/// #![deny(unsafe_op_in_unsafe_fn)] +/// +/// trait Indexable { +/// const LEN: usize; +/// +/// /// # Safety +/// /// +/// /// The caller must ensure that `idx < LEN`. +/// unsafe fn idx_unchecked(&self, idx: usize) -> i32; /// } /// -/// // The implementation of `ThreeIndexable` for `[T; 4]` is `unsafe` -/// // because the implementor must abide by a contract the compiler cannot -/// // check but as a programmer we know there will always be a valid element -/// // at index 3 to access. -/// unsafe impl<T> ThreeIndexable<T> for [T; 4] { -/// fn three(&self) -> &T { -/// // SAFETY: implementing the trait means there always is an element -/// // with index 3 accessible. -/// unsafe { self.get_unchecked(3) } +/// // The implementation for `i32` doesn't need to do any contract reasoning. +/// impl Indexable for i32 { +/// const LEN: usize = 1; +/// +/// unsafe fn idx_unchecked(&self, idx: usize) -> i32 { +/// debug_assert_eq!(idx, 0); +/// *self /// } /// } /// -/// let a = [1, 2, 4, 8]; -/// assert_eq!(a.three(), &8); +/// // The implementation for arrays exploits the function contract to +/// // make use of `get_unchecked` on slices and avoid a run-time check. +/// impl Indexable for [i32; 42] { +/// const LEN: usize = 42; +/// +/// unsafe fn idx_unchecked(&self, idx: usize) -> i32 { +/// // SAFETY: As per this trait's documentation, the caller ensures +/// // that `idx < 42`. +/// unsafe { *self.get_unchecked(idx) } +/// } +/// } +/// +/// // The implementation for the never type declares a length of 0, +/// // which means `idx_unchecked` can never be called. +/// impl Indexable for ! { +/// const LEN: usize = 0; +/// +/// unsafe fn idx_unchecked(&self, idx: usize) -> i32 { +/// // SAFETY: As per this trait's documentation, the caller ensures +/// // that `idx < 0`, which is impossible, so this is dead code. +/// unsafe { std::hint::unreachable_unchecked() } +/// } +/// } +/// +/// fn use_indexable<I: Indexable>(x: I, idx: usize) -> i32 { +/// if idx < I::LEN { +/// // SAFETY: We have checked that `idx < I::LEN`. +/// unsafe { x.idx_unchecked(idx) } +/// } else { +/// panic!("index out-of-bounds") +/// } +/// } /// ``` /// +/// This time, `use_indexable` is safe because it uses a run-time check to discharge the safety +/// contract of `idx_unchecked`. Implementing `Indexable` is safe because when writing +/// `idx_unchecked`, we don't have to worry: our *callers* need to discharge a proof obligation +/// (like `use_indexable` does), but the *implementation* of `get_unchecked` has no proof obligation +/// to contend with. Of course, the implementation of `Indexable` may choose to call other unsafe +/// operations, and then it needs an `unsafe` *block* to indicate it discharged the proof +/// obligations of its callees. (We enabled `unsafe_op_in_unsafe_fn`, so the body of `idx_unchecked` +/// is not implicitly an unsafe block.) For that purpose it can make use of the contract that all +/// its callers must uphold -- the fact that `idx < LEN`. +/// +/// Formally speaking, an `unsafe fn` in a trait is a function with *preconditions* that go beyond +/// those encoded by the argument types (such as `idx < LEN`), whereas an `unsafe trait` can declare +/// that some of its functions have *postconditions* that go beyond those encoded in the return type +/// (such as returning an even integer). If a trait needs a function with both extra precondition +/// and extra postcondition, then it needs an `unsafe fn` in an `unsafe trait`. +/// /// [`extern`]: keyword.extern.html /// [`trait`]: keyword.trait.html /// [`static`]: keyword.static.html @@ -2043,7 +2133,6 @@ mod type_keyword {} /// [nomicon-soundness]: ../nomicon/safe-unsafe-meaning.html /// [soundness]: https://rust-lang.github.io/unsafe-code-guidelines/glossary.html#soundness-of-code--of-a-library /// [Reference]: ../reference/unsafety.html -/// [proposal]: https://github.com/rust-lang/rfcs/pull/2585 /// [discussion on Rust Internals]: https://internals.rust-lang.org/t/what-does-unsafe-mean/6696 mod unsafe_keyword {} @@ -2113,7 +2202,7 @@ mod use_keyword {} /// Add constraints that must be upheld to use an item. /// /// `where` allows specifying constraints on lifetime and generic parameters. -/// The [RFC] introducing `where` contains detailed informations about the +/// The [RFC] introducing `where` contains detailed information about the /// keyword. /// /// # Examples @@ -2355,7 +2444,7 @@ mod dyn_keyword {} /// println!("f = {f} and i = {i}"); /// ``` /// -/// See the [Reference][union] for more informations on `union`s. +/// See the [Reference][union] for more information on `union`s. /// /// [`struct`]: keyword.struct.html /// [union]: ../reference/items/unions.html diff --git a/library/std/src/lazy.rs b/library/std/src/lazy.rs deleted file mode 100644 index f8c06c3f9..000000000 --- a/library/std/src/lazy.rs +++ /dev/null @@ -1 +0,0 @@ -//! Lazy values and one-time initialization of static data. diff --git a/library/std/src/lib.rs b/library/std/src/lib.rs index 20d25a608..385585dad 100644 --- a/library/std/src/lib.rs +++ b/library/std/src/lib.rs @@ -145,8 +145,8 @@ //! abstracting over differences in common platforms, most notably Windows and //! Unix derivatives. //! -//! Common types of I/O, including [files], [TCP], [UDP], are defined in the -//! [`io`], [`fs`], and [`net`] modules. +//! Common types of I/O, including [files], [TCP], and [UDP], are defined in +//! the [`io`], [`fs`], and [`net`] modules. //! //! The [`thread`] module contains Rust's threading abstractions. [`sync`] //! contains further primitive shared memory types, including [`atomic`] and @@ -187,6 +187,7 @@ //! [rust-discord]: https://discord.gg/rust-lang //! [array]: prim@array //! [slice]: prim@slice + #![cfg_attr(not(feature = "restricted-std"), stable(feature = "rust1", since = "1.0.0"))] #![cfg_attr(feature = "restricted-std", unstable(feature = "restricted_std", issue = "none"))] #![doc( @@ -201,25 +202,35 @@ no_global_oom_handling, not(no_global_oom_handling) ))] +// To run libstd tests without x.py without ending up with two copies of libstd, Miri needs to be +// able to "empty" this crate. See <https://github.com/rust-lang/miri-test-libstd/issues/4>. +// rustc itself never sets the feature, so this line has no affect there. +#![cfg(any(not(feature = "miri-test-libstd"), test, doctest))] +// miri-test-libstd also prefers to make std use the sysroot versions of the dependencies. +#![cfg_attr(feature = "miri-test-libstd", feature(rustc_private))] // Don't link to std. We are std. #![no_std] +// Tell the compiler to link to either panic_abort or panic_unwind +#![needs_panic_runtime] +// +// Lints: #![warn(deprecated_in_future)] #![warn(missing_docs)] #![warn(missing_debug_implementations)] #![allow(explicit_outlives_requirements)] #![allow(unused_lifetimes)] -// Tell the compiler to link to either panic_abort or panic_unwind -#![needs_panic_runtime] +#![deny(rustc::existing_doc_keyword)] // Ensure that std can be linked against panic_abort despite compiled with `-C panic=unwind` -#![cfg_attr(not(bootstrap), deny(ffi_unwind_calls))] +#![deny(ffi_unwind_calls)] // std may use features in a platform-specific way #![allow(unused_features)] +// +// Features: #![cfg_attr(test, feature(internal_output_capture, print_internals, update_panic_count, rt))] #![cfg_attr( all(target_vendor = "fortanix", target_env = "sgx"), feature(slice_index_methods, coerce_unsized, sgx_platform) )] -#![deny(rustc::existing_doc_keyword)] // // Language features: #![feature(alloc_error_handler)] @@ -240,12 +251,13 @@ #![feature(doc_notable_trait)] #![feature(dropck_eyepatch)] #![feature(exhaustive_patterns)] +#![feature(if_let_guard)] #![feature(intra_doc_pointers)] -#![feature(label_break_value)] +#![feature(is_terminal)] #![feature(lang_items)] #![feature(let_chains)] -#![feature(let_else)] #![feature(linkage)] +#![feature(link_cfg)] #![feature(min_specialization)] #![feature(must_not_suspend)] #![feature(needs_panic_runtime)] @@ -258,6 +270,7 @@ #![feature(staged_api)] #![feature(thread_local)] #![feature(try_blocks)] +#![feature(utf8_chunks)] // // Library features (core): #![feature(array_error_internals)] @@ -267,23 +280,27 @@ #![feature(core_intrinsics)] #![feature(cstr_from_bytes_until_nul)] #![feature(cstr_internals)] -#![feature(duration_checked_float)] #![feature(duration_constants)] +#![feature(error_generic_member_access)] +#![feature(error_in_core)] +#![feature(error_iter)] #![feature(exact_size_is_empty)] #![feature(exclusive_wrapper)] #![feature(extend_one)] #![feature(float_minimum_maximum)] +#![feature(float_next_up_down)] #![feature(hasher_prefixfree_extras)] #![feature(hashmap_internals)] #![feature(int_error_internals)] -#![feature(is_some_with)] +#![feature(is_some_and)] #![feature(maybe_uninit_slice)] #![feature(maybe_uninit_write_slice)] -#![feature(mixed_integer_ops)] #![feature(nonnull_slice_from_raw_parts)] #![feature(panic_can_unwind)] #![feature(panic_info_message)] #![feature(panic_internals)] +#![feature(pointer_byte_offsets)] +#![feature(pointer_is_aligned)] #![feature(portable_simd)] #![feature(prelude_2024)] #![feature(provide_any)] @@ -294,6 +311,9 @@ #![feature(std_internals)] #![feature(str_internals)] #![feature(strict_provenance)] +#![feature(maybe_uninit_uninit_array)] +#![feature(const_maybe_uninit_uninit_array)] +#![feature(const_waker)] // // Library features (alloc): #![feature(alloc_layout_extra)] @@ -329,9 +349,9 @@ #![feature(trace_macros)] // // Only used in tests/benchmarks: -#![feature(bench_black_box)] // // Only for const-ness: +#![feature(const_collections_with_hasher)] #![feature(const_io_structs)] #![feature(const_ip)] #![feature(const_ipv4)] @@ -509,9 +529,6 @@ pub mod process; pub mod sync; pub mod time; -#[unstable(feature = "once_cell", issue = "74465")] -pub mod lazy; - // Pull in `std_float` crate into libstd. The contents of // `std_float` are in a different repository: rust-lang/portable-simd. #[path = "../../portable-simd/crates/std_float/src/lib.rs"] @@ -576,6 +593,7 @@ pub mod alloc; // Private support modules mod panicking; +mod personality; #[path = "../../backtrace/src/lib.rs"] #[allow(dead_code, unused_attributes)] diff --git a/library/std/src/macros.rs b/library/std/src/macros.rs index 0cb21ef53..6e4ba1404 100644 --- a/library/std/src/macros.rs +++ b/library/std/src/macros.rs @@ -27,17 +27,31 @@ macro_rules! panic { /// necessary to use [`io::stdout().flush()`][flush] to ensure the output is emitted /// immediately. /// +/// The `print!` macro will lock the standard output on each call. If you call +/// `print!` within a hot loop, this behavior may be the bottleneck of the loop. +/// To avoid this, lock stdout with [`io::stdout().lock()`][lock]: +/// ``` +/// use std::io::{stdout, Write}; +/// +/// let mut lock = stdout().lock(); +/// write!(lock, "hello world").unwrap(); +/// ``` +/// /// Use `print!` only for the primary output of your program. Use /// [`eprint!`] instead to print error and progress messages. /// /// [flush]: crate::io::Write::flush /// [`println!`]: crate::println /// [`eprint!`]: crate::eprint +/// [lock]: crate::io::Stdout /// /// # Panics /// /// Panics if writing to `io::stdout()` fails. /// +/// Writing to non-blocking stdout can cause an error, which will lead +/// this macro to panic. +/// /// # Examples /// /// ``` @@ -75,16 +89,30 @@ macro_rules! print { /// This macro uses the same syntax as [`format!`], but writes to the standard output instead. /// See [`std::fmt`] for more information. /// +/// The `println!` macro will lock the standard output on each call. If you call +/// `println!` within a hot loop, this behavior may be the bottleneck of the loop. +/// To avoid this, lock stdout with [`io::stdout().lock()`][lock]: +/// ``` +/// use std::io::{stdout, Write}; +/// +/// let mut lock = stdout().lock(); +/// writeln!(lock, "hello world").unwrap(); +/// ``` +/// /// Use `println!` only for the primary output of your program. Use /// [`eprintln!`] instead to print error and progress messages. /// /// [`std::fmt`]: crate::fmt /// [`eprintln!`]: crate::eprintln +/// [lock]: crate::io::Stdout /// /// # Panics /// /// Panics if writing to [`io::stdout`] fails. /// +/// Writing to non-blocking stdout can cause an error, which will lead +/// this macro to panic. +/// /// [`io::stdout`]: crate::io::stdout /// /// # Examples @@ -93,6 +121,8 @@ macro_rules! print { /// println!(); // prints just a newline /// println!("hello there!"); /// println!("format {} arguments", "some"); +/// let local_variable = "some"; +/// println!("format {local_variable} arguments"); /// ``` #[macro_export] #[stable(feature = "rust1", since = "1.0.0")] @@ -123,6 +153,9 @@ macro_rules! println { /// /// Panics if writing to `io::stderr` fails. /// +/// Writing to non-blocking stdout can cause an error, which will lead +/// this macro to panic. +/// /// # Examples /// /// ``` @@ -155,6 +188,9 @@ macro_rules! eprint { /// /// Panics if writing to `io::stderr` fails. /// +/// Writing to non-blocking stdout can cause an error, which will lead +/// this macro to panic. +/// /// # Examples /// /// ``` diff --git a/library/std/src/net/addr.rs b/library/std/src/net/addr.rs deleted file mode 100644 index 53fee952a..000000000 --- a/library/std/src/net/addr.rs +++ /dev/null @@ -1,988 +0,0 @@ -#[cfg(all(test, not(target_os = "emscripten")))] -mod tests; - -use crate::cmp::Ordering; -use crate::fmt; -use crate::hash; -use crate::io::{self, Write}; -use crate::iter; -use crate::mem; -use crate::net::{IpAddr, Ipv4Addr, Ipv6Addr}; -use crate::option; -use crate::slice; -use crate::sys::net::netc as c; -use crate::sys_common::net::LookupHost; -use crate::sys_common::{FromInner, IntoInner}; -use crate::vec; - -/// An internet socket address, either IPv4 or IPv6. -/// -/// Internet socket addresses consist of an [IP address], a 16-bit port number, as well -/// as possibly some version-dependent additional information. See [`SocketAddrV4`]'s and -/// [`SocketAddrV6`]'s respective documentation for more details. -/// -/// The size of a `SocketAddr` instance may vary depending on the target operating -/// system. -/// -/// [IP address]: IpAddr -/// -/// # Examples -/// -/// ``` -/// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; -/// -/// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); -/// -/// assert_eq!("127.0.0.1:8080".parse(), Ok(socket)); -/// assert_eq!(socket.port(), 8080); -/// assert_eq!(socket.is_ipv4(), true); -/// ``` -#[derive(Copy, Clone, PartialEq, Eq, Hash, PartialOrd, Ord)] -#[stable(feature = "rust1", since = "1.0.0")] -pub enum SocketAddr { - /// An IPv4 socket address. - #[stable(feature = "rust1", since = "1.0.0")] - V4(#[stable(feature = "rust1", since = "1.0.0")] SocketAddrV4), - /// An IPv6 socket address. - #[stable(feature = "rust1", since = "1.0.0")] - V6(#[stable(feature = "rust1", since = "1.0.0")] SocketAddrV6), -} - -/// An IPv4 socket address. -/// -/// IPv4 socket addresses consist of an [`IPv4` address] and a 16-bit port number, as -/// stated in [IETF RFC 793]. -/// -/// See [`SocketAddr`] for a type encompassing both IPv4 and IPv6 socket addresses. -/// -/// The size of a `SocketAddrV4` struct may vary depending on the target operating -/// system. Do not assume that this type has the same memory layout as the underlying -/// system representation. -/// -/// [IETF RFC 793]: https://tools.ietf.org/html/rfc793 -/// [`IPv4` address]: Ipv4Addr -/// -/// # Examples -/// -/// ``` -/// use std::net::{Ipv4Addr, SocketAddrV4}; -/// -/// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); -/// -/// assert_eq!("127.0.0.1:8080".parse(), Ok(socket)); -/// assert_eq!(socket.ip(), &Ipv4Addr::new(127, 0, 0, 1)); -/// assert_eq!(socket.port(), 8080); -/// ``` -#[derive(Copy, Clone, Eq, PartialEq)] -#[stable(feature = "rust1", since = "1.0.0")] -pub struct SocketAddrV4 { - ip: Ipv4Addr, - port: u16, -} - -/// An IPv6 socket address. -/// -/// IPv6 socket addresses consist of an [`IPv6` address], a 16-bit port number, as well -/// as fields containing the traffic class, the flow label, and a scope identifier -/// (see [IETF RFC 2553, Section 3.3] for more details). -/// -/// See [`SocketAddr`] for a type encompassing both IPv4 and IPv6 socket addresses. -/// -/// The size of a `SocketAddrV6` struct may vary depending on the target operating -/// system. Do not assume that this type has the same memory layout as the underlying -/// system representation. -/// -/// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 -/// [`IPv6` address]: Ipv6Addr -/// -/// # Examples -/// -/// ``` -/// use std::net::{Ipv6Addr, SocketAddrV6}; -/// -/// let socket = SocketAddrV6::new(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 1), 8080, 0, 0); -/// -/// assert_eq!("[2001:db8::1]:8080".parse(), Ok(socket)); -/// assert_eq!(socket.ip(), &Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 1)); -/// assert_eq!(socket.port(), 8080); -/// ``` -#[derive(Copy, Clone, Eq, PartialEq)] -#[stable(feature = "rust1", since = "1.0.0")] -pub struct SocketAddrV6 { - ip: Ipv6Addr, - port: u16, - flowinfo: u32, - scope_id: u32, -} - -impl SocketAddr { - /// Creates a new socket address from an [IP address] and a port number. - /// - /// [IP address]: IpAddr - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; - /// - /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); - /// assert_eq!(socket.ip(), IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1))); - /// assert_eq!(socket.port(), 8080); - /// ``` - #[stable(feature = "ip_addr", since = "1.7.0")] - #[must_use] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn new(ip: IpAddr, port: u16) -> SocketAddr { - match ip { - IpAddr::V4(a) => SocketAddr::V4(SocketAddrV4::new(a, port)), - IpAddr::V6(a) => SocketAddr::V6(SocketAddrV6::new(a, port, 0, 0)), - } - } - - /// Returns the IP address associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; - /// - /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); - /// assert_eq!(socket.ip(), IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1))); - /// ``` - #[must_use] - #[stable(feature = "ip_addr", since = "1.7.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn ip(&self) -> IpAddr { - match *self { - SocketAddr::V4(ref a) => IpAddr::V4(*a.ip()), - SocketAddr::V6(ref a) => IpAddr::V6(*a.ip()), - } - } - - /// Changes the IP address associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; - /// - /// let mut socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); - /// socket.set_ip(IpAddr::V4(Ipv4Addr::new(10, 10, 0, 1))); - /// assert_eq!(socket.ip(), IpAddr::V4(Ipv4Addr::new(10, 10, 0, 1))); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_ip(&mut self, new_ip: IpAddr) { - // `match (*self, new_ip)` would have us mutate a copy of self only to throw it away. - match (self, new_ip) { - (&mut SocketAddr::V4(ref mut a), IpAddr::V4(new_ip)) => a.set_ip(new_ip), - (&mut SocketAddr::V6(ref mut a), IpAddr::V6(new_ip)) => a.set_ip(new_ip), - (self_, new_ip) => *self_ = Self::new(new_ip, self_.port()), - } - } - - /// Returns the port number associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; - /// - /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); - /// assert_eq!(socket.port(), 8080); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn port(&self) -> u16 { - match *self { - SocketAddr::V4(ref a) => a.port(), - SocketAddr::V6(ref a) => a.port(), - } - } - - /// Changes the port number associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; - /// - /// let mut socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); - /// socket.set_port(1025); - /// assert_eq!(socket.port(), 1025); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_port(&mut self, new_port: u16) { - match *self { - SocketAddr::V4(ref mut a) => a.set_port(new_port), - SocketAddr::V6(ref mut a) => a.set_port(new_port), - } - } - - /// Returns [`true`] if the [IP address] in this `SocketAddr` is an - /// [`IPv4` address], and [`false`] otherwise. - /// - /// [IP address]: IpAddr - /// [`IPv4` address]: IpAddr::V4 - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; - /// - /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); - /// assert_eq!(socket.is_ipv4(), true); - /// assert_eq!(socket.is_ipv6(), false); - /// ``` - #[must_use] - #[stable(feature = "sockaddr_checker", since = "1.16.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn is_ipv4(&self) -> bool { - matches!(*self, SocketAddr::V4(_)) - } - - /// Returns [`true`] if the [IP address] in this `SocketAddr` is an - /// [`IPv6` address], and [`false`] otherwise. - /// - /// [IP address]: IpAddr - /// [`IPv6` address]: IpAddr::V6 - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv6Addr, SocketAddr}; - /// - /// let socket = SocketAddr::new(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 65535, 0, 1)), 8080); - /// assert_eq!(socket.is_ipv4(), false); - /// assert_eq!(socket.is_ipv6(), true); - /// ``` - #[must_use] - #[stable(feature = "sockaddr_checker", since = "1.16.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn is_ipv6(&self) -> bool { - matches!(*self, SocketAddr::V6(_)) - } -} - -impl SocketAddrV4 { - /// Creates a new socket address from an [`IPv4` address] and a port number. - /// - /// [`IPv4` address]: Ipv4Addr - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV4, Ipv4Addr}; - /// - /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); - /// ``` - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn new(ip: Ipv4Addr, port: u16) -> SocketAddrV4 { - SocketAddrV4 { ip, port } - } - - /// Returns the IP address associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV4, Ipv4Addr}; - /// - /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); - /// assert_eq!(socket.ip(), &Ipv4Addr::new(127, 0, 0, 1)); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn ip(&self) -> &Ipv4Addr { - &self.ip - } - - /// Changes the IP address associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV4, Ipv4Addr}; - /// - /// let mut socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); - /// socket.set_ip(Ipv4Addr::new(192, 168, 0, 1)); - /// assert_eq!(socket.ip(), &Ipv4Addr::new(192, 168, 0, 1)); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_ip(&mut self, new_ip: Ipv4Addr) { - self.ip = new_ip; - } - - /// Returns the port number associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV4, Ipv4Addr}; - /// - /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); - /// assert_eq!(socket.port(), 8080); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn port(&self) -> u16 { - self.port - } - - /// Changes the port number associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV4, Ipv4Addr}; - /// - /// let mut socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); - /// socket.set_port(4242); - /// assert_eq!(socket.port(), 4242); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_port(&mut self, new_port: u16) { - self.port = new_port; - } -} - -impl SocketAddrV6 { - /// Creates a new socket address from an [`IPv6` address], a 16-bit port number, - /// and the `flowinfo` and `scope_id` fields. - /// - /// For more information on the meaning and layout of the `flowinfo` and `scope_id` - /// parameters, see [IETF RFC 2553, Section 3.3]. - /// - /// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 - /// [`IPv6` address]: Ipv6Addr - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); - /// ``` - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn new(ip: Ipv6Addr, port: u16, flowinfo: u32, scope_id: u32) -> SocketAddrV6 { - SocketAddrV6 { ip, port, flowinfo, scope_id } - } - - /// Returns the IP address associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); - /// assert_eq!(socket.ip(), &Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn ip(&self) -> &Ipv6Addr { - &self.ip - } - - /// Changes the IP address associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); - /// socket.set_ip(Ipv6Addr::new(76, 45, 0, 0, 0, 0, 0, 0)); - /// assert_eq!(socket.ip(), &Ipv6Addr::new(76, 45, 0, 0, 0, 0, 0, 0)); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_ip(&mut self, new_ip: Ipv6Addr) { - self.ip = new_ip; - } - - /// Returns the port number associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); - /// assert_eq!(socket.port(), 8080); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn port(&self) -> u16 { - self.port - } - - /// Changes the port number associated with this socket address. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); - /// socket.set_port(4242); - /// assert_eq!(socket.port(), 4242); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_port(&mut self, new_port: u16) { - self.port = new_port; - } - - /// Returns the flow information associated with this address. - /// - /// This information corresponds to the `sin6_flowinfo` field in C's `netinet/in.h`, - /// as specified in [IETF RFC 2553, Section 3.3]. - /// It combines information about the flow label and the traffic class as specified - /// in [IETF RFC 2460], respectively [Section 6] and [Section 7]. - /// - /// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 - /// [IETF RFC 2460]: https://tools.ietf.org/html/rfc2460 - /// [Section 6]: https://tools.ietf.org/html/rfc2460#section-6 - /// [Section 7]: https://tools.ietf.org/html/rfc2460#section-7 - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 10, 0); - /// assert_eq!(socket.flowinfo(), 10); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn flowinfo(&self) -> u32 { - self.flowinfo - } - - /// Changes the flow information associated with this socket address. - /// - /// See [`SocketAddrV6::flowinfo`]'s documentation for more details. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 10, 0); - /// socket.set_flowinfo(56); - /// assert_eq!(socket.flowinfo(), 56); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_flowinfo(&mut self, new_flowinfo: u32) { - self.flowinfo = new_flowinfo; - } - - /// Returns the scope ID associated with this address. - /// - /// This information corresponds to the `sin6_scope_id` field in C's `netinet/in.h`, - /// as specified in [IETF RFC 2553, Section 3.3]. - /// - /// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 78); - /// assert_eq!(socket.scope_id(), 78); - /// ``` - #[must_use] - #[stable(feature = "rust1", since = "1.0.0")] - #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] - pub const fn scope_id(&self) -> u32 { - self.scope_id - } - - /// Changes the scope ID associated with this socket address. - /// - /// See [`SocketAddrV6::scope_id`]'s documentation for more details. - /// - /// # Examples - /// - /// ``` - /// use std::net::{SocketAddrV6, Ipv6Addr}; - /// - /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 78); - /// socket.set_scope_id(42); - /// assert_eq!(socket.scope_id(), 42); - /// ``` - #[stable(feature = "sockaddr_setters", since = "1.9.0")] - pub fn set_scope_id(&mut self, new_scope_id: u32) { - self.scope_id = new_scope_id; - } -} - -impl FromInner<c::sockaddr_in> for SocketAddrV4 { - fn from_inner(addr: c::sockaddr_in) -> SocketAddrV4 { - SocketAddrV4 { ip: Ipv4Addr::from_inner(addr.sin_addr), port: u16::from_be(addr.sin_port) } - } -} - -impl FromInner<c::sockaddr_in6> for SocketAddrV6 { - fn from_inner(addr: c::sockaddr_in6) -> SocketAddrV6 { - SocketAddrV6 { - ip: Ipv6Addr::from_inner(addr.sin6_addr), - port: u16::from_be(addr.sin6_port), - flowinfo: addr.sin6_flowinfo, - scope_id: addr.sin6_scope_id, - } - } -} - -impl IntoInner<c::sockaddr_in> for SocketAddrV4 { - fn into_inner(self) -> c::sockaddr_in { - c::sockaddr_in { - sin_family: c::AF_INET as c::sa_family_t, - sin_port: self.port.to_be(), - sin_addr: self.ip.into_inner(), - ..unsafe { mem::zeroed() } - } - } -} - -impl IntoInner<c::sockaddr_in6> for SocketAddrV6 { - fn into_inner(self) -> c::sockaddr_in6 { - c::sockaddr_in6 { - sin6_family: c::AF_INET6 as c::sa_family_t, - sin6_port: self.port.to_be(), - sin6_addr: self.ip.into_inner(), - sin6_flowinfo: self.flowinfo, - sin6_scope_id: self.scope_id, - ..unsafe { mem::zeroed() } - } - } -} - -#[stable(feature = "ip_from_ip", since = "1.16.0")] -impl From<SocketAddrV4> for SocketAddr { - /// Converts a [`SocketAddrV4`] into a [`SocketAddr::V4`]. - fn from(sock4: SocketAddrV4) -> SocketAddr { - SocketAddr::V4(sock4) - } -} - -#[stable(feature = "ip_from_ip", since = "1.16.0")] -impl From<SocketAddrV6> for SocketAddr { - /// Converts a [`SocketAddrV6`] into a [`SocketAddr::V6`]. - fn from(sock6: SocketAddrV6) -> SocketAddr { - SocketAddr::V6(sock6) - } -} - -#[stable(feature = "addr_from_into_ip", since = "1.17.0")] -impl<I: Into<IpAddr>> From<(I, u16)> for SocketAddr { - /// Converts a tuple struct (Into<[`IpAddr`]>, `u16`) into a [`SocketAddr`]. - /// - /// This conversion creates a [`SocketAddr::V4`] for an [`IpAddr::V4`] - /// and creates a [`SocketAddr::V6`] for an [`IpAddr::V6`]. - /// - /// `u16` is treated as port of the newly created [`SocketAddr`]. - fn from(pieces: (I, u16)) -> SocketAddr { - SocketAddr::new(pieces.0.into(), pieces.1) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Display for SocketAddr { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - match *self { - SocketAddr::V4(ref a) => a.fmt(f), - SocketAddr::V6(ref a) => a.fmt(f), - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Debug for SocketAddr { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(self, fmt) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Display for SocketAddrV4 { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - // Fast path: if there's no alignment stuff, write to the output buffer - // directly - if f.precision().is_none() && f.width().is_none() { - write!(f, "{}:{}", self.ip(), self.port()) - } else { - const IPV4_SOCKET_BUF_LEN: usize = (3 * 4) // the segments - + 3 // the separators - + 1 + 5; // the port - let mut buf = [0; IPV4_SOCKET_BUF_LEN]; - let mut buf_slice = &mut buf[..]; - - // Unwrap is fine because writing to a sufficiently-sized - // buffer is infallible - write!(buf_slice, "{}:{}", self.ip(), self.port()).unwrap(); - let len = IPV4_SOCKET_BUF_LEN - buf_slice.len(); - - // This unsafe is OK because we know what is being written to the buffer - let buf = unsafe { crate::str::from_utf8_unchecked(&buf[..len]) }; - f.pad(buf) - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Debug for SocketAddrV4 { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(self, fmt) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Display for SocketAddrV6 { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - // Fast path: if there's no alignment stuff, write to the output - // buffer directly - if f.precision().is_none() && f.width().is_none() { - match self.scope_id() { - 0 => write!(f, "[{}]:{}", self.ip(), self.port()), - scope_id => write!(f, "[{}%{}]:{}", self.ip(), scope_id, self.port()), - } - } else { - const IPV6_SOCKET_BUF_LEN: usize = (4 * 8) // The address - + 7 // The colon separators - + 2 // The brackets - + 1 + 10 // The scope id - + 1 + 5; // The port - - let mut buf = [0; IPV6_SOCKET_BUF_LEN]; - let mut buf_slice = &mut buf[..]; - - match self.scope_id() { - 0 => write!(buf_slice, "[{}]:{}", self.ip(), self.port()), - scope_id => write!(buf_slice, "[{}%{}]:{}", self.ip(), scope_id, self.port()), - } - // Unwrap is fine because writing to a sufficiently-sized - // buffer is infallible - .unwrap(); - let len = IPV6_SOCKET_BUF_LEN - buf_slice.len(); - - // This unsafe is OK because we know what is being written to the buffer - let buf = unsafe { crate::str::from_utf8_unchecked(&buf[..len]) }; - f.pad(buf) - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Debug for SocketAddrV6 { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(self, fmt) - } -} - -#[stable(feature = "socketaddr_ordering", since = "1.45.0")] -impl PartialOrd for SocketAddrV4 { - fn partial_cmp(&self, other: &SocketAddrV4) -> Option<Ordering> { - Some(self.cmp(other)) - } -} - -#[stable(feature = "socketaddr_ordering", since = "1.45.0")] -impl PartialOrd for SocketAddrV6 { - fn partial_cmp(&self, other: &SocketAddrV6) -> Option<Ordering> { - Some(self.cmp(other)) - } -} - -#[stable(feature = "socketaddr_ordering", since = "1.45.0")] -impl Ord for SocketAddrV4 { - fn cmp(&self, other: &SocketAddrV4) -> Ordering { - self.ip().cmp(other.ip()).then(self.port().cmp(&other.port())) - } -} - -#[stable(feature = "socketaddr_ordering", since = "1.45.0")] -impl Ord for SocketAddrV6 { - fn cmp(&self, other: &SocketAddrV6) -> Ordering { - self.ip().cmp(other.ip()).then(self.port().cmp(&other.port())) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl hash::Hash for SocketAddrV4 { - fn hash<H: hash::Hasher>(&self, s: &mut H) { - (self.port, self.ip).hash(s) - } -} -#[stable(feature = "rust1", since = "1.0.0")] -impl hash::Hash for SocketAddrV6 { - fn hash<H: hash::Hasher>(&self, s: &mut H) { - (self.port, &self.ip, self.flowinfo, self.scope_id).hash(s) - } -} - -/// A trait for objects which can be converted or resolved to one or more -/// [`SocketAddr`] values. -/// -/// This trait is used for generic address resolution when constructing network -/// objects. By default it is implemented for the following types: -/// -/// * [`SocketAddr`]: [`to_socket_addrs`] is the identity function. -/// -/// * [`SocketAddrV4`], [`SocketAddrV6`], <code>([IpAddr], [u16])</code>, -/// <code>([Ipv4Addr], [u16])</code>, <code>([Ipv6Addr], [u16])</code>: -/// [`to_socket_addrs`] constructs a [`SocketAddr`] trivially. -/// -/// * <code>(&[str], [u16])</code>: <code>&[str]</code> should be either a string representation -/// of an [`IpAddr`] address as expected by [`FromStr`] implementation or a host -/// name. [`u16`] is the port number. -/// -/// * <code>&[str]</code>: the string should be either a string representation of a -/// [`SocketAddr`] as expected by its [`FromStr`] implementation or a string like -/// `<host_name>:<port>` pair where `<port>` is a [`u16`] value. -/// -/// This trait allows constructing network objects like [`TcpStream`] or -/// [`UdpSocket`] easily with values of various types for the bind/connection -/// address. It is needed because sometimes one type is more appropriate than -/// the other: for simple uses a string like `"localhost:12345"` is much nicer -/// than manual construction of the corresponding [`SocketAddr`], but sometimes -/// [`SocketAddr`] value is *the* main source of the address, and converting it to -/// some other type (e.g., a string) just for it to be converted back to -/// [`SocketAddr`] in constructor methods is pointless. -/// -/// Addresses returned by the operating system that are not IP addresses are -/// silently ignored. -/// -/// [`FromStr`]: crate::str::FromStr "std::str::FromStr" -/// [`TcpStream`]: crate::net::TcpStream "net::TcpStream" -/// [`to_socket_addrs`]: ToSocketAddrs::to_socket_addrs -/// [`UdpSocket`]: crate::net::UdpSocket "net::UdpSocket" -/// -/// # Examples -/// -/// Creating a [`SocketAddr`] iterator that yields one item: -/// -/// ``` -/// use std::net::{ToSocketAddrs, SocketAddr}; -/// -/// let addr = SocketAddr::from(([127, 0, 0, 1], 443)); -/// let mut addrs_iter = addr.to_socket_addrs().unwrap(); -/// -/// assert_eq!(Some(addr), addrs_iter.next()); -/// assert!(addrs_iter.next().is_none()); -/// ``` -/// -/// Creating a [`SocketAddr`] iterator from a hostname: -/// -/// ```no_run -/// use std::net::{SocketAddr, ToSocketAddrs}; -/// -/// // assuming 'localhost' resolves to 127.0.0.1 -/// let mut addrs_iter = "localhost:443".to_socket_addrs().unwrap(); -/// assert_eq!(addrs_iter.next(), Some(SocketAddr::from(([127, 0, 0, 1], 443)))); -/// assert!(addrs_iter.next().is_none()); -/// -/// // assuming 'foo' does not resolve -/// assert!("foo:443".to_socket_addrs().is_err()); -/// ``` -/// -/// Creating a [`SocketAddr`] iterator that yields multiple items: -/// -/// ``` -/// use std::net::{SocketAddr, ToSocketAddrs}; -/// -/// let addr1 = SocketAddr::from(([0, 0, 0, 0], 80)); -/// let addr2 = SocketAddr::from(([127, 0, 0, 1], 443)); -/// let addrs = vec![addr1, addr2]; -/// -/// let mut addrs_iter = (&addrs[..]).to_socket_addrs().unwrap(); -/// -/// assert_eq!(Some(addr1), addrs_iter.next()); -/// assert_eq!(Some(addr2), addrs_iter.next()); -/// assert!(addrs_iter.next().is_none()); -/// ``` -/// -/// Attempting to create a [`SocketAddr`] iterator from an improperly formatted -/// socket address `&str` (missing the port): -/// -/// ``` -/// use std::io; -/// use std::net::ToSocketAddrs; -/// -/// let err = "127.0.0.1".to_socket_addrs().unwrap_err(); -/// assert_eq!(err.kind(), io::ErrorKind::InvalidInput); -/// ``` -/// -/// [`TcpStream::connect`] is an example of an function that utilizes -/// `ToSocketAddrs` as a trait bound on its parameter in order to accept -/// different types: -/// -/// ```no_run -/// use std::net::{TcpStream, Ipv4Addr}; -/// -/// let stream = TcpStream::connect(("127.0.0.1", 443)); -/// // or -/// let stream = TcpStream::connect("127.0.0.1:443"); -/// // or -/// let stream = TcpStream::connect((Ipv4Addr::new(127, 0, 0, 1), 443)); -/// ``` -/// -/// [`TcpStream::connect`]: crate::net::TcpStream::connect -#[stable(feature = "rust1", since = "1.0.0")] -pub trait ToSocketAddrs { - /// Returned iterator over socket addresses which this type may correspond - /// to. - #[stable(feature = "rust1", since = "1.0.0")] - type Iter: Iterator<Item = SocketAddr>; - - /// Converts this object to an iterator of resolved [`SocketAddr`]s. - /// - /// The returned iterator might not actually yield any values depending on the - /// outcome of any resolution performed. - /// - /// Note that this function may block the current thread while resolution is - /// performed. - #[stable(feature = "rust1", since = "1.0.0")] - fn to_socket_addrs(&self) -> io::Result<Self::Iter>; -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for SocketAddr { - type Iter = option::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { - Ok(Some(*self).into_iter()) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for SocketAddrV4 { - type Iter = option::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { - SocketAddr::V4(*self).to_socket_addrs() - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for SocketAddrV6 { - type Iter = option::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { - SocketAddr::V6(*self).to_socket_addrs() - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for (IpAddr, u16) { - type Iter = option::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { - let (ip, port) = *self; - match ip { - IpAddr::V4(ref a) => (*a, port).to_socket_addrs(), - IpAddr::V6(ref a) => (*a, port).to_socket_addrs(), - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for (Ipv4Addr, u16) { - type Iter = option::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { - let (ip, port) = *self; - SocketAddrV4::new(ip, port).to_socket_addrs() - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for (Ipv6Addr, u16) { - type Iter = option::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { - let (ip, port) = *self; - SocketAddrV6::new(ip, port, 0, 0).to_socket_addrs() - } -} - -fn resolve_socket_addr(lh: LookupHost) -> io::Result<vec::IntoIter<SocketAddr>> { - let p = lh.port(); - let v: Vec<_> = lh - .map(|mut a| { - a.set_port(p); - a - }) - .collect(); - Ok(v.into_iter()) -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for (&str, u16) { - type Iter = vec::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { - let (host, port) = *self; - - // try to parse the host as a regular IP address first - if let Ok(addr) = host.parse::<Ipv4Addr>() { - let addr = SocketAddrV4::new(addr, port); - return Ok(vec![SocketAddr::V4(addr)].into_iter()); - } - if let Ok(addr) = host.parse::<Ipv6Addr>() { - let addr = SocketAddrV6::new(addr, port, 0, 0); - return Ok(vec![SocketAddr::V6(addr)].into_iter()); - } - - resolve_socket_addr((host, port).try_into()?) - } -} - -#[stable(feature = "string_u16_to_socket_addrs", since = "1.46.0")] -impl ToSocketAddrs for (String, u16) { - type Iter = vec::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { - (&*self.0, self.1).to_socket_addrs() - } -} - -// accepts strings like 'localhost:12345' -#[stable(feature = "rust1", since = "1.0.0")] -impl ToSocketAddrs for str { - type Iter = vec::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { - // try to parse as a regular SocketAddr first - if let Ok(addr) = self.parse() { - return Ok(vec![addr].into_iter()); - } - - resolve_socket_addr(self.try_into()?) - } -} - -#[stable(feature = "slice_to_socket_addrs", since = "1.8.0")] -impl<'a> ToSocketAddrs for &'a [SocketAddr] { - type Iter = iter::Cloned<slice::Iter<'a, SocketAddr>>; - - fn to_socket_addrs(&self) -> io::Result<Self::Iter> { - Ok(self.iter().cloned()) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl<T: ToSocketAddrs + ?Sized> ToSocketAddrs for &T { - type Iter = T::Iter; - fn to_socket_addrs(&self) -> io::Result<T::Iter> { - (**self).to_socket_addrs() - } -} - -#[stable(feature = "string_to_socket_addrs", since = "1.16.0")] -impl ToSocketAddrs for String { - type Iter = vec::IntoIter<SocketAddr>; - fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { - (&**self).to_socket_addrs() - } -} diff --git a/library/std/src/net/addr/tests.rs b/library/std/src/net/addr/tests.rs deleted file mode 100644 index 585a17451..000000000 --- a/library/std/src/net/addr/tests.rs +++ /dev/null @@ -1,237 +0,0 @@ -use crate::net::test::{sa4, sa6, tsa}; -use crate::net::*; - -#[test] -fn to_socket_addr_ipaddr_u16() { - let a = Ipv4Addr::new(77, 88, 21, 11); - let p = 12345; - let e = SocketAddr::V4(SocketAddrV4::new(a, p)); - assert_eq!(Ok(vec![e]), tsa((a, p))); -} - -#[test] -fn to_socket_addr_str_u16() { - let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 24352); - assert_eq!(Ok(vec![a]), tsa(("77.88.21.11", 24352))); - - let a = sa6(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53); - assert_eq!(Ok(vec![a]), tsa(("2a02:6b8:0:1::1", 53))); - - let a = sa4(Ipv4Addr::new(127, 0, 0, 1), 23924); - #[cfg(not(target_env = "sgx"))] - assert!(tsa(("localhost", 23924)).unwrap().contains(&a)); - #[cfg(target_env = "sgx")] - let _ = a; -} - -#[test] -fn to_socket_addr_str() { - let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 24352); - assert_eq!(Ok(vec![a]), tsa("77.88.21.11:24352")); - - let a = sa6(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53); - assert_eq!(Ok(vec![a]), tsa("[2a02:6b8:0:1::1]:53")); - - let a = sa4(Ipv4Addr::new(127, 0, 0, 1), 23924); - #[cfg(not(target_env = "sgx"))] - assert!(tsa("localhost:23924").unwrap().contains(&a)); - #[cfg(target_env = "sgx")] - let _ = a; -} - -#[test] -fn to_socket_addr_string() { - let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 24352); - assert_eq!(Ok(vec![a]), tsa(&*format!("{}:{}", "77.88.21.11", "24352"))); - assert_eq!(Ok(vec![a]), tsa(&format!("{}:{}", "77.88.21.11", "24352"))); - assert_eq!(Ok(vec![a]), tsa(format!("{}:{}", "77.88.21.11", "24352"))); - - let s = format!("{}:{}", "77.88.21.11", "24352"); - assert_eq!(Ok(vec![a]), tsa(s)); - // s has been moved into the tsa call -} - -#[test] -fn bind_udp_socket_bad() { - // rust-lang/rust#53957: This is a regression test for a parsing problem - // discovered as part of issue rust-lang/rust#23076, where we were - // incorrectly parsing invalid input and then that would result in a - // successful `UdpSocket` binding when we would expect failure. - // - // At one time, this test was written as a call to `tsa` with - // INPUT_23076. However, that structure yields an unreliable test, - // because it ends up passing junk input to the DNS server, and some DNS - // servers will respond with `Ok` to such input, with the ip address of - // the DNS server itself. - // - // This form of the test is more robust: even when the DNS server - // returns its own address, it is still an error to bind a UDP socket to - // a non-local address, and so we still get an error here in that case. - - const INPUT_23076: &str = "1200::AB00:1234::2552:7777:1313:34300"; - - assert!(crate::net::UdpSocket::bind(INPUT_23076).is_err()) -} - -#[test] -fn set_ip() { - fn ip4(low: u8) -> Ipv4Addr { - Ipv4Addr::new(77, 88, 21, low) - } - fn ip6(low: u16) -> Ipv6Addr { - Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, low) - } - - let mut v4 = SocketAddrV4::new(ip4(11), 80); - assert_eq!(v4.ip(), &ip4(11)); - v4.set_ip(ip4(12)); - assert_eq!(v4.ip(), &ip4(12)); - - let mut addr = SocketAddr::V4(v4); - assert_eq!(addr.ip(), IpAddr::V4(ip4(12))); - addr.set_ip(IpAddr::V4(ip4(13))); - assert_eq!(addr.ip(), IpAddr::V4(ip4(13))); - addr.set_ip(IpAddr::V6(ip6(14))); - assert_eq!(addr.ip(), IpAddr::V6(ip6(14))); - - let mut v6 = SocketAddrV6::new(ip6(1), 80, 0, 0); - assert_eq!(v6.ip(), &ip6(1)); - v6.set_ip(ip6(2)); - assert_eq!(v6.ip(), &ip6(2)); - - let mut addr = SocketAddr::V6(v6); - assert_eq!(addr.ip(), IpAddr::V6(ip6(2))); - addr.set_ip(IpAddr::V6(ip6(3))); - assert_eq!(addr.ip(), IpAddr::V6(ip6(3))); - addr.set_ip(IpAddr::V4(ip4(4))); - assert_eq!(addr.ip(), IpAddr::V4(ip4(4))); -} - -#[test] -fn set_port() { - let mut v4 = SocketAddrV4::new(Ipv4Addr::new(77, 88, 21, 11), 80); - assert_eq!(v4.port(), 80); - v4.set_port(443); - assert_eq!(v4.port(), 443); - - let mut addr = SocketAddr::V4(v4); - assert_eq!(addr.port(), 443); - addr.set_port(8080); - assert_eq!(addr.port(), 8080); - - let mut v6 = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 80, 0, 0); - assert_eq!(v6.port(), 80); - v6.set_port(443); - assert_eq!(v6.port(), 443); - - let mut addr = SocketAddr::V6(v6); - assert_eq!(addr.port(), 443); - addr.set_port(8080); - assert_eq!(addr.port(), 8080); -} - -#[test] -fn set_flowinfo() { - let mut v6 = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 80, 10, 0); - assert_eq!(v6.flowinfo(), 10); - v6.set_flowinfo(20); - assert_eq!(v6.flowinfo(), 20); -} - -#[test] -fn set_scope_id() { - let mut v6 = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 80, 0, 10); - assert_eq!(v6.scope_id(), 10); - v6.set_scope_id(20); - assert_eq!(v6.scope_id(), 20); -} - -#[test] -fn is_v4() { - let v4 = SocketAddr::V4(SocketAddrV4::new(Ipv4Addr::new(77, 88, 21, 11), 80)); - assert!(v4.is_ipv4()); - assert!(!v4.is_ipv6()); -} - -#[test] -fn is_v6() { - let v6 = SocketAddr::V6(SocketAddrV6::new( - Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), - 80, - 10, - 0, - )); - assert!(!v6.is_ipv4()); - assert!(v6.is_ipv6()); -} - -#[test] -fn socket_v4_to_str() { - let socket = SocketAddrV4::new(Ipv4Addr::new(192, 168, 0, 1), 8080); - - assert_eq!(format!("{socket}"), "192.168.0.1:8080"); - assert_eq!(format!("{socket:<20}"), "192.168.0.1:8080 "); - assert_eq!(format!("{socket:>20}"), " 192.168.0.1:8080"); - assert_eq!(format!("{socket:^20}"), " 192.168.0.1:8080 "); - assert_eq!(format!("{socket:.10}"), "192.168.0."); -} - -#[test] -fn socket_v6_to_str() { - let mut socket = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53, 0, 0); - - assert_eq!(format!("{socket}"), "[2a02:6b8:0:1::1]:53"); - assert_eq!(format!("{socket:<24}"), "[2a02:6b8:0:1::1]:53 "); - assert_eq!(format!("{socket:>24}"), " [2a02:6b8:0:1::1]:53"); - assert_eq!(format!("{socket:^24}"), " [2a02:6b8:0:1::1]:53 "); - assert_eq!(format!("{socket:.15}"), "[2a02:6b8:0:1::"); - - socket.set_scope_id(5); - - assert_eq!(format!("{socket}"), "[2a02:6b8:0:1::1%5]:53"); - assert_eq!(format!("{socket:<24}"), "[2a02:6b8:0:1::1%5]:53 "); - assert_eq!(format!("{socket:>24}"), " [2a02:6b8:0:1::1%5]:53"); - assert_eq!(format!("{socket:^24}"), " [2a02:6b8:0:1::1%5]:53 "); - assert_eq!(format!("{socket:.18}"), "[2a02:6b8:0:1::1%5"); -} - -#[test] -fn compare() { - let v4_1 = "224.120.45.1:23456".parse::<SocketAddrV4>().unwrap(); - let v4_2 = "224.210.103.5:12345".parse::<SocketAddrV4>().unwrap(); - let v4_3 = "224.210.103.5:23456".parse::<SocketAddrV4>().unwrap(); - let v6_1 = "[2001:db8:f00::1002]:23456".parse::<SocketAddrV6>().unwrap(); - let v6_2 = "[2001:db8:f00::2001]:12345".parse::<SocketAddrV6>().unwrap(); - let v6_3 = "[2001:db8:f00::2001]:23456".parse::<SocketAddrV6>().unwrap(); - - // equality - assert_eq!(v4_1, v4_1); - assert_eq!(v6_1, v6_1); - assert_eq!(SocketAddr::V4(v4_1), SocketAddr::V4(v4_1)); - assert_eq!(SocketAddr::V6(v6_1), SocketAddr::V6(v6_1)); - assert!(v4_1 != v4_2); - assert!(v6_1 != v6_2); - - // compare different addresses - assert!(v4_1 < v4_2); - assert!(v6_1 < v6_2); - assert!(v4_2 > v4_1); - assert!(v6_2 > v6_1); - - // compare the same address with different ports - assert!(v4_2 < v4_3); - assert!(v6_2 < v6_3); - assert!(v4_3 > v4_2); - assert!(v6_3 > v6_2); - - // compare different addresses with the same port - assert!(v4_1 < v4_3); - assert!(v6_1 < v6_3); - assert!(v4_3 > v4_1); - assert!(v6_3 > v6_1); - - // compare with an inferred right-hand side - assert_eq!(v4_1, "224.120.45.1:23456".parse().unwrap()); - assert_eq!(v6_1, "[2001:db8:f00::1002]:23456".parse().unwrap()); - assert_eq!(SocketAddr::V4(v4_1), "224.120.45.1:23456".parse().unwrap()); -} diff --git a/library/std/src/net/display_buffer.rs b/library/std/src/net/display_buffer.rs new file mode 100644 index 000000000..7aadf06e9 --- /dev/null +++ b/library/std/src/net/display_buffer.rs @@ -0,0 +1,40 @@ +use crate::fmt; +use crate::mem::MaybeUninit; +use crate::str; + +/// Used for slow path in `Display` implementations when alignment is required. +pub struct DisplayBuffer<const SIZE: usize> { + buf: [MaybeUninit<u8>; SIZE], + len: usize, +} + +impl<const SIZE: usize> DisplayBuffer<SIZE> { + #[inline] + pub const fn new() -> Self { + Self { buf: MaybeUninit::uninit_array(), len: 0 } + } + + #[inline] + pub fn as_str(&self) -> &str { + // SAFETY: `buf` is only written to by the `fmt::Write::write_str` implementation + // which writes a valid UTF-8 string to `buf` and correctly sets `len`. + unsafe { + let s = MaybeUninit::slice_assume_init_ref(&self.buf[..self.len]); + str::from_utf8_unchecked(s) + } + } +} + +impl<const SIZE: usize> fmt::Write for DisplayBuffer<SIZE> { + fn write_str(&mut self, s: &str) -> fmt::Result { + let bytes = s.as_bytes(); + + if let Some(buf) = self.buf.get_mut(self.len..(self.len + bytes.len())) { + MaybeUninit::write_slice(buf, bytes); + self.len += bytes.len(); + Ok(()) + } else { + Err(fmt::Error) + } + } +} diff --git a/library/std/src/net/ip.rs b/library/std/src/net/ip.rs deleted file mode 100644 index 41ca9ba84..000000000 --- a/library/std/src/net/ip.rs +++ /dev/null @@ -1,2040 +0,0 @@ -// Tests for this module -#[cfg(all(test, not(target_os = "emscripten")))] -mod tests; - -use crate::cmp::Ordering; -use crate::fmt::{self, Write as FmtWrite}; -use crate::io::Write as IoWrite; -use crate::mem::transmute; -use crate::sys::net::netc as c; -use crate::sys_common::{FromInner, IntoInner}; - -/// An IP address, either IPv4 or IPv6. -/// -/// This enum can contain either an [`Ipv4Addr`] or an [`Ipv6Addr`], see their -/// respective documentation for more details. -/// -/// # Examples -/// -/// ``` -/// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; -/// -/// let localhost_v4 = IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)); -/// let localhost_v6 = IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); -/// -/// assert_eq!("127.0.0.1".parse(), Ok(localhost_v4)); -/// assert_eq!("::1".parse(), Ok(localhost_v6)); -/// -/// assert_eq!(localhost_v4.is_ipv6(), false); -/// assert_eq!(localhost_v4.is_ipv4(), true); -/// ``` -#[stable(feature = "ip_addr", since = "1.7.0")] -#[derive(Copy, Clone, Eq, PartialEq, Hash, PartialOrd, Ord)] -pub enum IpAddr { - /// An IPv4 address. - #[stable(feature = "ip_addr", since = "1.7.0")] - V4(#[stable(feature = "ip_addr", since = "1.7.0")] Ipv4Addr), - /// An IPv6 address. - #[stable(feature = "ip_addr", since = "1.7.0")] - V6(#[stable(feature = "ip_addr", since = "1.7.0")] Ipv6Addr), -} - -/// An IPv4 address. -/// -/// IPv4 addresses are defined as 32-bit integers in [IETF RFC 791]. -/// They are usually represented as four octets. -/// -/// See [`IpAddr`] for a type encompassing both IPv4 and IPv6 addresses. -/// -/// [IETF RFC 791]: https://tools.ietf.org/html/rfc791 -/// -/// # Textual representation -/// -/// `Ipv4Addr` provides a [`FromStr`] implementation. The four octets are in decimal -/// notation, divided by `.` (this is called "dot-decimal notation"). -/// Notably, octal numbers (which are indicated with a leading `0`) and hexadecimal numbers (which -/// are indicated with a leading `0x`) are not allowed per [IETF RFC 6943]. -/// -/// [IETF RFC 6943]: https://tools.ietf.org/html/rfc6943#section-3.1.1 -/// [`FromStr`]: crate::str::FromStr -/// -/// # Examples -/// -/// ``` -/// use std::net::Ipv4Addr; -/// -/// let localhost = Ipv4Addr::new(127, 0, 0, 1); -/// assert_eq!("127.0.0.1".parse(), Ok(localhost)); -/// assert_eq!(localhost.is_loopback(), true); -/// assert!("012.004.002.000".parse::<Ipv4Addr>().is_err()); // all octets are in octal -/// assert!("0000000.0.0.0".parse::<Ipv4Addr>().is_err()); // first octet is a zero in octal -/// assert!("0xcb.0x0.0x71.0x00".parse::<Ipv4Addr>().is_err()); // all octets are in hex -/// ``` -#[derive(Copy, Clone, PartialEq, Eq, Hash)] -#[stable(feature = "rust1", since = "1.0.0")] -pub struct Ipv4Addr { - octets: [u8; 4], -} - -/// An IPv6 address. -/// -/// IPv6 addresses are defined as 128-bit integers in [IETF RFC 4291]. -/// They are usually represented as eight 16-bit segments. -/// -/// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 -/// -/// # Embedding IPv4 Addresses -/// -/// See [`IpAddr`] for a type encompassing both IPv4 and IPv6 addresses. -/// -/// To assist in the transition from IPv4 to IPv6 two types of IPv6 addresses that embed an IPv4 address were defined: -/// IPv4-compatible and IPv4-mapped addresses. Of these IPv4-compatible addresses have been officially deprecated. -/// -/// Both types of addresses are not assigned any special meaning by this implementation, -/// other than what the relevant standards prescribe. This means that an address like `::ffff:127.0.0.1`, -/// while representing an IPv4 loopback address, is not itself an IPv6 loopback address; only `::1` is. -/// To handle these so called "IPv4-in-IPv6" addresses, they have to first be converted to their canonical IPv4 address. -/// -/// ### IPv4-Compatible IPv6 Addresses -/// -/// IPv4-compatible IPv6 addresses are defined in [IETF RFC 4291 Section 2.5.5.1], and have been officially deprecated. -/// The RFC describes the format of an "IPv4-Compatible IPv6 address" as follows: -/// -/// ```text -/// | 80 bits | 16 | 32 bits | -/// +--------------------------------------+--------------------------+ -/// |0000..............................0000|0000| IPv4 address | -/// +--------------------------------------+----+---------------------+ -/// ``` -/// So `::a.b.c.d` would be an IPv4-compatible IPv6 address representing the IPv4 address `a.b.c.d`. -/// -/// To convert from an IPv4 address to an IPv4-compatible IPv6 address, use [`Ipv4Addr::to_ipv6_compatible`]. -/// Use [`Ipv6Addr::to_ipv4`] to convert an IPv4-compatible IPv6 address to the canonical IPv4 address. -/// -/// [IETF RFC 4291 Section 2.5.5.1]: https://datatracker.ietf.org/doc/html/rfc4291#section-2.5.5.1 -/// -/// ### IPv4-Mapped IPv6 Addresses -/// -/// IPv4-mapped IPv6 addresses are defined in [IETF RFC 4291 Section 2.5.5.2]. -/// The RFC describes the format of an "IPv4-Mapped IPv6 address" as follows: -/// -/// ```text -/// | 80 bits | 16 | 32 bits | -/// +--------------------------------------+--------------------------+ -/// |0000..............................0000|FFFF| IPv4 address | -/// +--------------------------------------+----+---------------------+ -/// ``` -/// So `::ffff:a.b.c.d` would be an IPv4-mapped IPv6 address representing the IPv4 address `a.b.c.d`. -/// -/// To convert from an IPv4 address to an IPv4-mapped IPv6 address, use [`Ipv4Addr::to_ipv6_mapped`]. -/// Use [`Ipv6Addr::to_ipv4`] to convert an IPv4-mapped IPv6 address to the canonical IPv4 address. -/// Note that this will also convert the IPv6 loopback address `::1` to `0.0.0.1`. Use -/// [`Ipv6Addr::to_ipv4_mapped`] to avoid this. -/// -/// [IETF RFC 4291 Section 2.5.5.2]: https://datatracker.ietf.org/doc/html/rfc4291#section-2.5.5.2 -/// -/// # Textual representation -/// -/// `Ipv6Addr` provides a [`FromStr`] implementation. There are many ways to represent -/// an IPv6 address in text, but in general, each segments is written in hexadecimal -/// notation, and segments are separated by `:`. For more information, see -/// [IETF RFC 5952]. -/// -/// [`FromStr`]: crate::str::FromStr -/// [IETF RFC 5952]: https://tools.ietf.org/html/rfc5952 -/// -/// # Examples -/// -/// ``` -/// use std::net::Ipv6Addr; -/// -/// let localhost = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); -/// assert_eq!("::1".parse(), Ok(localhost)); -/// assert_eq!(localhost.is_loopback(), true); -/// ``` -#[derive(Copy, Clone, PartialEq, Eq, Hash)] -#[stable(feature = "rust1", since = "1.0.0")] -pub struct Ipv6Addr { - octets: [u8; 16], -} - -/// Scope of an [IPv6 multicast address] as defined in [IETF RFC 7346 section 2]. -/// -/// # Stability Guarantees -/// -/// Not all possible values for a multicast scope have been assigned. -/// Future RFCs may introduce new scopes, which will be added as variants to this enum; -/// because of this the enum is marked as `#[non_exhaustive]`. -/// -/// # Examples -/// ``` -/// #![feature(ip)] -/// -/// use std::net::Ipv6Addr; -/// use std::net::Ipv6MulticastScope::*; -/// -/// // An IPv6 multicast address with global scope (`ff0e::`). -/// let address = Ipv6Addr::new(0xff0e, 0, 0, 0, 0, 0, 0, 0); -/// -/// // Will print "Global scope". -/// match address.multicast_scope() { -/// Some(InterfaceLocal) => println!("Interface-Local scope"), -/// Some(LinkLocal) => println!("Link-Local scope"), -/// Some(RealmLocal) => println!("Realm-Local scope"), -/// Some(AdminLocal) => println!("Admin-Local scope"), -/// Some(SiteLocal) => println!("Site-Local scope"), -/// Some(OrganizationLocal) => println!("Organization-Local scope"), -/// Some(Global) => println!("Global scope"), -/// Some(_) => println!("Unknown scope"), -/// None => println!("Not a multicast address!") -/// } -/// -/// ``` -/// -/// [IPv6 multicast address]: Ipv6Addr -/// [IETF RFC 7346 section 2]: https://tools.ietf.org/html/rfc7346#section-2 -#[derive(Copy, PartialEq, Eq, Clone, Hash, Debug)] -#[unstable(feature = "ip", issue = "27709")] -#[non_exhaustive] -pub enum Ipv6MulticastScope { - /// Interface-Local scope. - InterfaceLocal, - /// Link-Local scope. - LinkLocal, - /// Realm-Local scope. - RealmLocal, - /// Admin-Local scope. - AdminLocal, - /// Site-Local scope. - SiteLocal, - /// Organization-Local scope. - OrganizationLocal, - /// Global scope. - Global, -} - -impl IpAddr { - /// Returns [`true`] for the special 'unspecified' address. - /// - /// See the documentation for [`Ipv4Addr::is_unspecified()`] and - /// [`Ipv6Addr::is_unspecified()`] for more details. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(0, 0, 0, 0)).is_unspecified(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)).is_unspecified(), true); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "ip_shared", since = "1.12.0")] - #[must_use] - #[inline] - pub const fn is_unspecified(&self) -> bool { - match self { - IpAddr::V4(ip) => ip.is_unspecified(), - IpAddr::V6(ip) => ip.is_unspecified(), - } - } - - /// Returns [`true`] if this is a loopback address. - /// - /// See the documentation for [`Ipv4Addr::is_loopback()`] and - /// [`Ipv6Addr::is_loopback()`] for more details. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)).is_loopback(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0x1)).is_loopback(), true); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "ip_shared", since = "1.12.0")] - #[must_use] - #[inline] - pub const fn is_loopback(&self) -> bool { - match self { - IpAddr::V4(ip) => ip.is_loopback(), - IpAddr::V6(ip) => ip.is_loopback(), - } - } - - /// Returns [`true`] if the address appears to be globally routable. - /// - /// See the documentation for [`Ipv4Addr::is_global()`] and - /// [`Ipv6Addr::is_global()`] for more details. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(80, 9, 12, 3)).is_global(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0x1c9, 0, 0, 0xafc8, 0, 0x1)).is_global(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ip", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_global(&self) -> bool { - match self { - IpAddr::V4(ip) => ip.is_global(), - IpAddr::V6(ip) => ip.is_global(), - } - } - - /// Returns [`true`] if this is a multicast address. - /// - /// See the documentation for [`Ipv4Addr::is_multicast()`] and - /// [`Ipv6Addr::is_multicast()`] for more details. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(224, 254, 0, 0)).is_multicast(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0)).is_multicast(), true); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "ip_shared", since = "1.12.0")] - #[must_use] - #[inline] - pub const fn is_multicast(&self) -> bool { - match self { - IpAddr::V4(ip) => ip.is_multicast(), - IpAddr::V6(ip) => ip.is_multicast(), - } - } - - /// Returns [`true`] if this address is in a range designated for documentation. - /// - /// See the documentation for [`Ipv4Addr::is_documentation()`] and - /// [`Ipv6Addr::is_documentation()`] for more details. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(203, 0, 113, 6)).is_documentation(), true); - /// assert_eq!( - /// IpAddr::V6(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0)).is_documentation(), - /// true - /// ); - /// ``` - #[rustc_const_unstable(feature = "const_ip", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_documentation(&self) -> bool { - match self { - IpAddr::V4(ip) => ip.is_documentation(), - IpAddr::V6(ip) => ip.is_documentation(), - } - } - - /// Returns [`true`] if this address is in a range designated for benchmarking. - /// - /// See the documentation for [`Ipv4Addr::is_benchmarking()`] and - /// [`Ipv6Addr::is_benchmarking()`] for more details. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(198, 19, 255, 255)).is_benchmarking(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0x2001, 0x2, 0, 0, 0, 0, 0, 0)).is_benchmarking(), true); - /// ``` - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_benchmarking(&self) -> bool { - match self { - IpAddr::V4(ip) => ip.is_benchmarking(), - IpAddr::V6(ip) => ip.is_benchmarking(), - } - } - - /// Returns [`true`] if this address is an [`IPv4` address], and [`false`] - /// otherwise. - /// - /// [`IPv4` address]: IpAddr::V4 - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(203, 0, 113, 6)).is_ipv4(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0)).is_ipv4(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "ipaddr_checker", since = "1.16.0")] - #[must_use] - #[inline] - pub const fn is_ipv4(&self) -> bool { - matches!(self, IpAddr::V4(_)) - } - - /// Returns [`true`] if this address is an [`IPv6` address], and [`false`] - /// otherwise. - /// - /// [`IPv6` address]: IpAddr::V6 - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(203, 0, 113, 6)).is_ipv6(), false); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0)).is_ipv6(), true); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "ipaddr_checker", since = "1.16.0")] - #[must_use] - #[inline] - pub const fn is_ipv6(&self) -> bool { - matches!(self, IpAddr::V6(_)) - } - - /// Converts this address to an `IpAddr::V4` if it is an IPv4-mapped IPv6 addresses, otherwise it - /// return `self` as-is. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)).to_canonical().is_loopback(), true); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1)).is_loopback(), false); - /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1)).to_canonical().is_loopback(), true); - /// ``` - #[inline] - #[must_use = "this returns the result of the operation, \ - without modifying the original"] - #[rustc_const_unstable(feature = "const_ip", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - pub const fn to_canonical(&self) -> IpAddr { - match self { - &v4 @ IpAddr::V4(_) => v4, - IpAddr::V6(v6) => v6.to_canonical(), - } - } -} - -impl Ipv4Addr { - /// Creates a new IPv4 address from four eight-bit octets. - /// - /// The result will represent the IP address `a`.`b`.`c`.`d`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::new(127, 0, 0, 1); - /// ``` - #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use] - #[inline] - pub const fn new(a: u8, b: u8, c: u8, d: u8) -> Ipv4Addr { - Ipv4Addr { octets: [a, b, c, d] } - } - - /// An IPv4 address with the address pointing to localhost: `127.0.0.1` - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::LOCALHOST; - /// assert_eq!(addr, Ipv4Addr::new(127, 0, 0, 1)); - /// ``` - #[stable(feature = "ip_constructors", since = "1.30.0")] - pub const LOCALHOST: Self = Ipv4Addr::new(127, 0, 0, 1); - - /// An IPv4 address representing an unspecified address: `0.0.0.0` - /// - /// This corresponds to the constant `INADDR_ANY` in other languages. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::UNSPECIFIED; - /// assert_eq!(addr, Ipv4Addr::new(0, 0, 0, 0)); - /// ``` - #[doc(alias = "INADDR_ANY")] - #[stable(feature = "ip_constructors", since = "1.30.0")] - pub const UNSPECIFIED: Self = Ipv4Addr::new(0, 0, 0, 0); - - /// An IPv4 address representing the broadcast address: `255.255.255.255` - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::BROADCAST; - /// assert_eq!(addr, Ipv4Addr::new(255, 255, 255, 255)); - /// ``` - #[stable(feature = "ip_constructors", since = "1.30.0")] - pub const BROADCAST: Self = Ipv4Addr::new(255, 255, 255, 255); - - /// Returns the four eight-bit integers that make up this address. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::new(127, 0, 0, 1); - /// assert_eq!(addr.octets(), [127, 0, 0, 1]); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use] - #[inline] - pub const fn octets(&self) -> [u8; 4] { - self.octets - } - - /// Returns [`true`] for the special 'unspecified' address (`0.0.0.0`). - /// - /// This property is defined in _UNIX Network Programming, Second Edition_, - /// W. Richard Stevens, p. 891; see also [ip7]. - /// - /// [ip7]: https://man7.org/linux/man-pages/man7/ip.7.html - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(0, 0, 0, 0).is_unspecified(), true); - /// assert_eq!(Ipv4Addr::new(45, 22, 13, 197).is_unspecified(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] - #[stable(feature = "ip_shared", since = "1.12.0")] - #[must_use] - #[inline] - pub const fn is_unspecified(&self) -> bool { - u32::from_be_bytes(self.octets) == 0 - } - - /// Returns [`true`] if this is a loopback address (`127.0.0.0/8`). - /// - /// This property is defined by [IETF RFC 1122]. - /// - /// [IETF RFC 1122]: https://tools.ietf.org/html/rfc1122 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(127, 0, 0, 1).is_loopback(), true); - /// assert_eq!(Ipv4Addr::new(45, 22, 13, 197).is_loopback(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_loopback(&self) -> bool { - self.octets()[0] == 127 - } - - /// Returns [`true`] if this is a private address. - /// - /// The private address ranges are defined in [IETF RFC 1918] and include: - /// - /// - `10.0.0.0/8` - /// - `172.16.0.0/12` - /// - `192.168.0.0/16` - /// - /// [IETF RFC 1918]: https://tools.ietf.org/html/rfc1918 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(10, 0, 0, 1).is_private(), true); - /// assert_eq!(Ipv4Addr::new(10, 10, 10, 10).is_private(), true); - /// assert_eq!(Ipv4Addr::new(172, 16, 10, 10).is_private(), true); - /// assert_eq!(Ipv4Addr::new(172, 29, 45, 14).is_private(), true); - /// assert_eq!(Ipv4Addr::new(172, 32, 0, 2).is_private(), false); - /// assert_eq!(Ipv4Addr::new(192, 168, 0, 2).is_private(), true); - /// assert_eq!(Ipv4Addr::new(192, 169, 0, 2).is_private(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_private(&self) -> bool { - match self.octets() { - [10, ..] => true, - [172, b, ..] if b >= 16 && b <= 31 => true, - [192, 168, ..] => true, - _ => false, - } - } - - /// Returns [`true`] if the address is link-local (`169.254.0.0/16`). - /// - /// This property is defined by [IETF RFC 3927]. - /// - /// [IETF RFC 3927]: https://tools.ietf.org/html/rfc3927 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(169, 254, 0, 0).is_link_local(), true); - /// assert_eq!(Ipv4Addr::new(169, 254, 10, 65).is_link_local(), true); - /// assert_eq!(Ipv4Addr::new(16, 89, 10, 65).is_link_local(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_link_local(&self) -> bool { - matches!(self.octets(), [169, 254, ..]) - } - - /// Returns [`true`] if the address appears to be globally routable. - /// See [iana-ipv4-special-registry][ipv4-sr]. - /// - /// The following return [`false`]: - /// - /// - private addresses (see [`Ipv4Addr::is_private()`]) - /// - the loopback address (see [`Ipv4Addr::is_loopback()`]) - /// - the link-local address (see [`Ipv4Addr::is_link_local()`]) - /// - the broadcast address (see [`Ipv4Addr::is_broadcast()`]) - /// - addresses used for documentation (see [`Ipv4Addr::is_documentation()`]) - /// - the unspecified address (see [`Ipv4Addr::is_unspecified()`]), and the whole - /// `0.0.0.0/8` block - /// - addresses reserved for future protocols, except - /// `192.0.0.9/32` and `192.0.0.10/32` which are globally routable - /// - addresses reserved for future use (see [`Ipv4Addr::is_reserved()`] - /// - addresses reserved for networking devices benchmarking (see - /// [`Ipv4Addr::is_benchmarking()`]) - /// - /// [ipv4-sr]: https://www.iana.org/assignments/iana-ipv4-special-registry/iana-ipv4-special-registry.xhtml - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv4Addr; - /// - /// // private addresses are not global - /// assert_eq!(Ipv4Addr::new(10, 254, 0, 0).is_global(), false); - /// assert_eq!(Ipv4Addr::new(192, 168, 10, 65).is_global(), false); - /// assert_eq!(Ipv4Addr::new(172, 16, 10, 65).is_global(), false); - /// - /// // the 0.0.0.0/8 block is not global - /// assert_eq!(Ipv4Addr::new(0, 1, 2, 3).is_global(), false); - /// // in particular, the unspecified address is not global - /// assert_eq!(Ipv4Addr::new(0, 0, 0, 0).is_global(), false); - /// - /// // the loopback address is not global - /// assert_eq!(Ipv4Addr::new(127, 0, 0, 1).is_global(), false); - /// - /// // link local addresses are not global - /// assert_eq!(Ipv4Addr::new(169, 254, 45, 1).is_global(), false); - /// - /// // the broadcast address is not global - /// assert_eq!(Ipv4Addr::new(255, 255, 255, 255).is_global(), false); - /// - /// // the address space designated for documentation is not global - /// assert_eq!(Ipv4Addr::new(192, 0, 2, 255).is_global(), false); - /// assert_eq!(Ipv4Addr::new(198, 51, 100, 65).is_global(), false); - /// assert_eq!(Ipv4Addr::new(203, 0, 113, 6).is_global(), false); - /// - /// // shared addresses are not global - /// assert_eq!(Ipv4Addr::new(100, 100, 0, 0).is_global(), false); - /// - /// // addresses reserved for protocol assignment are not global - /// assert_eq!(Ipv4Addr::new(192, 0, 0, 0).is_global(), false); - /// assert_eq!(Ipv4Addr::new(192, 0, 0, 255).is_global(), false); - /// - /// // addresses reserved for future use are not global - /// assert_eq!(Ipv4Addr::new(250, 10, 20, 30).is_global(), false); - /// - /// // addresses reserved for network devices benchmarking are not global - /// assert_eq!(Ipv4Addr::new(198, 18, 0, 0).is_global(), false); - /// - /// // All the other addresses are global - /// assert_eq!(Ipv4Addr::new(1, 1, 1, 1).is_global(), true); - /// assert_eq!(Ipv4Addr::new(80, 9, 12, 3).is_global(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_global(&self) -> bool { - // check if this address is 192.0.0.9 or 192.0.0.10. These addresses are the only two - // globally routable addresses in the 192.0.0.0/24 range. - if u32::from_be_bytes(self.octets()) == 0xc0000009 - || u32::from_be_bytes(self.octets()) == 0xc000000a - { - return true; - } - !self.is_private() - && !self.is_loopback() - && !self.is_link_local() - && !self.is_broadcast() - && !self.is_documentation() - && !self.is_shared() - // addresses reserved for future protocols (`192.0.0.0/24`) - && !(self.octets()[0] == 192 && self.octets()[1] == 0 && self.octets()[2] == 0) - && !self.is_reserved() - && !self.is_benchmarking() - // Make sure the address is not in 0.0.0.0/8 - && self.octets()[0] != 0 - } - - /// Returns [`true`] if this address is part of the Shared Address Space defined in - /// [IETF RFC 6598] (`100.64.0.0/10`). - /// - /// [IETF RFC 6598]: https://tools.ietf.org/html/rfc6598 - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(100, 64, 0, 0).is_shared(), true); - /// assert_eq!(Ipv4Addr::new(100, 127, 255, 255).is_shared(), true); - /// assert_eq!(Ipv4Addr::new(100, 128, 0, 0).is_shared(), false); - /// ``` - #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_shared(&self) -> bool { - self.octets()[0] == 100 && (self.octets()[1] & 0b1100_0000 == 0b0100_0000) - } - - /// Returns [`true`] if this address part of the `198.18.0.0/15` range, which is reserved for - /// network devices benchmarking. This range is defined in [IETF RFC 2544] as `192.18.0.0` - /// through `198.19.255.255` but [errata 423] corrects it to `198.18.0.0/15`. - /// - /// [IETF RFC 2544]: https://tools.ietf.org/html/rfc2544 - /// [errata 423]: https://www.rfc-editor.org/errata/eid423 - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(198, 17, 255, 255).is_benchmarking(), false); - /// assert_eq!(Ipv4Addr::new(198, 18, 0, 0).is_benchmarking(), true); - /// assert_eq!(Ipv4Addr::new(198, 19, 255, 255).is_benchmarking(), true); - /// assert_eq!(Ipv4Addr::new(198, 20, 0, 0).is_benchmarking(), false); - /// ``` - #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_benchmarking(&self) -> bool { - self.octets()[0] == 198 && (self.octets()[1] & 0xfe) == 18 - } - - /// Returns [`true`] if this address is reserved by IANA for future use. [IETF RFC 1112] - /// defines the block of reserved addresses as `240.0.0.0/4`. This range normally includes the - /// broadcast address `255.255.255.255`, but this implementation explicitly excludes it, since - /// it is obviously not reserved for future use. - /// - /// [IETF RFC 1112]: https://tools.ietf.org/html/rfc1112 - /// - /// # Warning - /// - /// As IANA assigns new addresses, this method will be - /// updated. This may result in non-reserved addresses being - /// treated as reserved in code that relies on an outdated version - /// of this method. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(240, 0, 0, 0).is_reserved(), true); - /// assert_eq!(Ipv4Addr::new(255, 255, 255, 254).is_reserved(), true); - /// - /// assert_eq!(Ipv4Addr::new(239, 255, 255, 255).is_reserved(), false); - /// // The broadcast address is not considered as reserved for future use by this implementation - /// assert_eq!(Ipv4Addr::new(255, 255, 255, 255).is_reserved(), false); - /// ``` - #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_reserved(&self) -> bool { - self.octets()[0] & 240 == 240 && !self.is_broadcast() - } - - /// Returns [`true`] if this is a multicast address (`224.0.0.0/4`). - /// - /// Multicast addresses have a most significant octet between `224` and `239`, - /// and is defined by [IETF RFC 5771]. - /// - /// [IETF RFC 5771]: https://tools.ietf.org/html/rfc5771 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(224, 254, 0, 0).is_multicast(), true); - /// assert_eq!(Ipv4Addr::new(236, 168, 10, 65).is_multicast(), true); - /// assert_eq!(Ipv4Addr::new(172, 16, 10, 65).is_multicast(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_multicast(&self) -> bool { - self.octets()[0] >= 224 && self.octets()[0] <= 239 - } - - /// Returns [`true`] if this is a broadcast address (`255.255.255.255`). - /// - /// A broadcast address has all octets set to `255` as defined in [IETF RFC 919]. - /// - /// [IETF RFC 919]: https://tools.ietf.org/html/rfc919 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(255, 255, 255, 255).is_broadcast(), true); - /// assert_eq!(Ipv4Addr::new(236, 168, 10, 65).is_broadcast(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_broadcast(&self) -> bool { - u32::from_be_bytes(self.octets()) == u32::from_be_bytes(Self::BROADCAST.octets()) - } - - /// Returns [`true`] if this address is in a range designated for documentation. - /// - /// This is defined in [IETF RFC 5737]: - /// - /// - `192.0.2.0/24` (TEST-NET-1) - /// - `198.51.100.0/24` (TEST-NET-2) - /// - `203.0.113.0/24` (TEST-NET-3) - /// - /// [IETF RFC 5737]: https://tools.ietf.org/html/rfc5737 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// assert_eq!(Ipv4Addr::new(192, 0, 2, 255).is_documentation(), true); - /// assert_eq!(Ipv4Addr::new(198, 51, 100, 65).is_documentation(), true); - /// assert_eq!(Ipv4Addr::new(203, 0, 113, 6).is_documentation(), true); - /// assert_eq!(Ipv4Addr::new(193, 34, 17, 19).is_documentation(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_documentation(&self) -> bool { - matches!(self.octets(), [192, 0, 2, _] | [198, 51, 100, _] | [203, 0, 113, _]) - } - - /// Converts this address to an [IPv4-compatible] [`IPv6` address]. - /// - /// `a.b.c.d` becomes `::a.b.c.d` - /// - /// Note that IPv4-compatible addresses have been officially deprecated. - /// If you don't explicitly need an IPv4-compatible address for legacy reasons, consider using `to_ipv6_mapped` instead. - /// - /// [IPv4-compatible]: Ipv6Addr#ipv4-compatible-ipv6-addresses - /// [`IPv6` address]: Ipv6Addr - /// - /// # Examples - /// - /// ``` - /// use std::net::{Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!( - /// Ipv4Addr::new(192, 0, 2, 255).to_ipv6_compatible(), - /// Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0xc000, 0x2ff) - /// ); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use = "this returns the result of the operation, \ - without modifying the original"] - #[inline] - pub const fn to_ipv6_compatible(&self) -> Ipv6Addr { - let [a, b, c, d] = self.octets(); - Ipv6Addr { octets: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, a, b, c, d] } - } - - /// Converts this address to an [IPv4-mapped] [`IPv6` address]. - /// - /// `a.b.c.d` becomes `::ffff:a.b.c.d` - /// - /// [IPv4-mapped]: Ipv6Addr#ipv4-mapped-ipv6-addresses - /// [`IPv6` address]: Ipv6Addr - /// - /// # Examples - /// - /// ``` - /// use std::net::{Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(Ipv4Addr::new(192, 0, 2, 255).to_ipv6_mapped(), - /// Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc000, 0x2ff)); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use = "this returns the result of the operation, \ - without modifying the original"] - #[inline] - pub const fn to_ipv6_mapped(&self) -> Ipv6Addr { - let [a, b, c, d] = self.octets(); - Ipv6Addr { octets: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xFF, 0xFF, a, b, c, d] } - } -} - -#[stable(feature = "ip_addr", since = "1.7.0")] -impl fmt::Display for IpAddr { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - match self { - IpAddr::V4(ip) => ip.fmt(fmt), - IpAddr::V6(ip) => ip.fmt(fmt), - } - } -} - -#[stable(feature = "ip_addr", since = "1.7.0")] -impl fmt::Debug for IpAddr { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(self, fmt) - } -} - -#[stable(feature = "ip_from_ip", since = "1.16.0")] -impl From<Ipv4Addr> for IpAddr { - /// Copies this address to a new `IpAddr::V4`. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr}; - /// - /// let addr = Ipv4Addr::new(127, 0, 0, 1); - /// - /// assert_eq!( - /// IpAddr::V4(addr), - /// IpAddr::from(addr) - /// ) - /// ``` - #[inline] - fn from(ipv4: Ipv4Addr) -> IpAddr { - IpAddr::V4(ipv4) - } -} - -#[stable(feature = "ip_from_ip", since = "1.16.0")] -impl From<Ipv6Addr> for IpAddr { - /// Copies this address to a new `IpAddr::V6`. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv6Addr}; - /// - /// let addr = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff); - /// - /// assert_eq!( - /// IpAddr::V6(addr), - /// IpAddr::from(addr) - /// ); - /// ``` - #[inline] - fn from(ipv6: Ipv6Addr) -> IpAddr { - IpAddr::V6(ipv6) - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Display for Ipv4Addr { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - let octets = self.octets(); - // Fast Path: if there's no alignment stuff, write directly to the buffer - if fmt.precision().is_none() && fmt.width().is_none() { - write!(fmt, "{}.{}.{}.{}", octets[0], octets[1], octets[2], octets[3]) - } else { - const IPV4_BUF_LEN: usize = 15; // Long enough for the longest possible IPv4 address - let mut buf = [0u8; IPV4_BUF_LEN]; - let mut buf_slice = &mut buf[..]; - - // Note: The call to write should never fail, hence the unwrap - write!(buf_slice, "{}.{}.{}.{}", octets[0], octets[1], octets[2], octets[3]).unwrap(); - let len = IPV4_BUF_LEN - buf_slice.len(); - - // This unsafe is OK because we know what is being written to the buffer - let buf = unsafe { crate::str::from_utf8_unchecked(&buf[..len]) }; - fmt.pad(buf) - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Debug for Ipv4Addr { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(self, fmt) - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialEq<Ipv4Addr> for IpAddr { - #[inline] - fn eq(&self, other: &Ipv4Addr) -> bool { - match self { - IpAddr::V4(v4) => v4 == other, - IpAddr::V6(_) => false, - } - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialEq<IpAddr> for Ipv4Addr { - #[inline] - fn eq(&self, other: &IpAddr) -> bool { - match other { - IpAddr::V4(v4) => self == v4, - IpAddr::V6(_) => false, - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl PartialOrd for Ipv4Addr { - #[inline] - fn partial_cmp(&self, other: &Ipv4Addr) -> Option<Ordering> { - Some(self.cmp(other)) - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialOrd<Ipv4Addr> for IpAddr { - #[inline] - fn partial_cmp(&self, other: &Ipv4Addr) -> Option<Ordering> { - match self { - IpAddr::V4(v4) => v4.partial_cmp(other), - IpAddr::V6(_) => Some(Ordering::Greater), - } - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialOrd<IpAddr> for Ipv4Addr { - #[inline] - fn partial_cmp(&self, other: &IpAddr) -> Option<Ordering> { - match other { - IpAddr::V4(v4) => self.partial_cmp(v4), - IpAddr::V6(_) => Some(Ordering::Less), - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Ord for Ipv4Addr { - #[inline] - fn cmp(&self, other: &Ipv4Addr) -> Ordering { - self.octets.cmp(&other.octets) - } -} - -impl IntoInner<c::in_addr> for Ipv4Addr { - #[inline] - fn into_inner(self) -> c::in_addr { - // `s_addr` is stored as BE on all machines and the array is in BE order. - // So the native endian conversion method is used so that it's never swapped. - c::in_addr { s_addr: u32::from_ne_bytes(self.octets) } - } -} -impl FromInner<c::in_addr> for Ipv4Addr { - fn from_inner(addr: c::in_addr) -> Ipv4Addr { - Ipv4Addr { octets: addr.s_addr.to_ne_bytes() } - } -} - -#[stable(feature = "ip_u32", since = "1.1.0")] -impl From<Ipv4Addr> for u32 { - /// Converts an `Ipv4Addr` into a host byte order `u32`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::new(0x12, 0x34, 0x56, 0x78); - /// assert_eq!(0x12345678, u32::from(addr)); - /// ``` - #[inline] - fn from(ip: Ipv4Addr) -> u32 { - u32::from_be_bytes(ip.octets) - } -} - -#[stable(feature = "ip_u32", since = "1.1.0")] -impl From<u32> for Ipv4Addr { - /// Converts a host byte order `u32` into an `Ipv4Addr`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::from(0x12345678); - /// assert_eq!(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78), addr); - /// ``` - #[inline] - fn from(ip: u32) -> Ipv4Addr { - Ipv4Addr { octets: ip.to_be_bytes() } - } -} - -#[stable(feature = "from_slice_v4", since = "1.9.0")] -impl From<[u8; 4]> for Ipv4Addr { - /// Creates an `Ipv4Addr` from a four element byte array. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv4Addr; - /// - /// let addr = Ipv4Addr::from([13u8, 12u8, 11u8, 10u8]); - /// assert_eq!(Ipv4Addr::new(13, 12, 11, 10), addr); - /// ``` - #[inline] - fn from(octets: [u8; 4]) -> Ipv4Addr { - Ipv4Addr { octets } - } -} - -#[stable(feature = "ip_from_slice", since = "1.17.0")] -impl From<[u8; 4]> for IpAddr { - /// Creates an `IpAddr::V4` from a four element byte array. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv4Addr}; - /// - /// let addr = IpAddr::from([13u8, 12u8, 11u8, 10u8]); - /// assert_eq!(IpAddr::V4(Ipv4Addr::new(13, 12, 11, 10)), addr); - /// ``` - #[inline] - fn from(octets: [u8; 4]) -> IpAddr { - IpAddr::V4(Ipv4Addr::from(octets)) - } -} - -impl Ipv6Addr { - /// Creates a new IPv6 address from eight 16-bit segments. - /// - /// The result will represent the IP address `a:b:c:d:e:f:g:h`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff); - /// ``` - #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use] - #[inline] - pub const fn new(a: u16, b: u16, c: u16, d: u16, e: u16, f: u16, g: u16, h: u16) -> Ipv6Addr { - let addr16 = [ - a.to_be(), - b.to_be(), - c.to_be(), - d.to_be(), - e.to_be(), - f.to_be(), - g.to_be(), - h.to_be(), - ]; - Ipv6Addr { - // All elements in `addr16` are big endian. - // SAFETY: `[u16; 8]` is always safe to transmute to `[u8; 16]`. - octets: unsafe { transmute::<_, [u8; 16]>(addr16) }, - } - } - - /// An IPv6 address representing localhost: `::1`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::LOCALHOST; - /// assert_eq!(addr, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); - /// ``` - #[stable(feature = "ip_constructors", since = "1.30.0")] - pub const LOCALHOST: Self = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); - - /// An IPv6 address representing the unspecified address: `::` - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::UNSPECIFIED; - /// assert_eq!(addr, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)); - /// ``` - #[stable(feature = "ip_constructors", since = "1.30.0")] - pub const UNSPECIFIED: Self = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0); - - /// Returns the eight 16-bit segments that make up this address. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).segments(), - /// [0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff]); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use] - #[inline] - pub const fn segments(&self) -> [u16; 8] { - // All elements in `self.octets` must be big endian. - // SAFETY: `[u8; 16]` is always safe to transmute to `[u16; 8]`. - let [a, b, c, d, e, f, g, h] = unsafe { transmute::<_, [u16; 8]>(self.octets) }; - // We want native endian u16 - [ - u16::from_be(a), - u16::from_be(b), - u16::from_be(c), - u16::from_be(d), - u16::from_be(e), - u16::from_be(f), - u16::from_be(g), - u16::from_be(h), - ] - } - - /// Returns [`true`] for the special 'unspecified' address (`::`). - /// - /// This property is defined in [IETF RFC 4291]. - /// - /// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_unspecified(), false); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0).is_unspecified(), true); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_unspecified(&self) -> bool { - u128::from_be_bytes(self.octets()) == u128::from_be_bytes(Ipv6Addr::UNSPECIFIED.octets()) - } - - /// Returns [`true`] if this is the [loopback address] (`::1`), - /// as defined in [IETF RFC 4291 section 2.5.3]. - /// - /// Contrary to IPv4, in IPv6 there is only one loopback address. - /// - /// [loopback address]: Ipv6Addr::LOCALHOST - /// [IETF RFC 4291 section 2.5.3]: https://tools.ietf.org/html/rfc4291#section-2.5.3 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_loopback(), false); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0x1).is_loopback(), true); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_loopback(&self) -> bool { - u128::from_be_bytes(self.octets()) == u128::from_be_bytes(Ipv6Addr::LOCALHOST.octets()) - } - - /// Returns [`true`] if the address appears to be globally routable. - /// - /// The following return [`false`]: - /// - /// - the loopback address - /// - link-local and unique local unicast addresses - /// - interface-, link-, realm-, admin- and site-local multicast addresses - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_global(), true); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0x1).is_global(), false); - /// assert_eq!(Ipv6Addr::new(0, 0, 0x1c9, 0, 0, 0xafc8, 0, 0x1).is_global(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_global(&self) -> bool { - match self.multicast_scope() { - Some(Ipv6MulticastScope::Global) => true, - None => self.is_unicast_global(), - _ => false, - } - } - - /// Returns [`true`] if this is a unique local address (`fc00::/7`). - /// - /// This property is defined in [IETF RFC 4193]. - /// - /// [IETF RFC 4193]: https://tools.ietf.org/html/rfc4193 - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_unique_local(), false); - /// assert_eq!(Ipv6Addr::new(0xfc02, 0, 0, 0, 0, 0, 0, 0).is_unique_local(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_unique_local(&self) -> bool { - (self.segments()[0] & 0xfe00) == 0xfc00 - } - - /// Returns [`true`] if this is a unicast address, as defined by [IETF RFC 4291]. - /// Any address that is not a [multicast address] (`ff00::/8`) is unicast. - /// - /// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 - /// [multicast address]: Ipv6Addr::is_multicast - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// // The unspecified and loopback addresses are unicast. - /// assert_eq!(Ipv6Addr::UNSPECIFIED.is_unicast(), true); - /// assert_eq!(Ipv6Addr::LOCALHOST.is_unicast(), true); - /// - /// // Any address that is not a multicast address (`ff00::/8`) is unicast. - /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_unicast(), true); - /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).is_unicast(), false); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_unicast(&self) -> bool { - !self.is_multicast() - } - - /// Returns `true` if the address is a unicast address with link-local scope, - /// as defined in [RFC 4291]. - /// - /// A unicast address has link-local scope if it has the prefix `fe80::/10`, as per [RFC 4291 section 2.4]. - /// Note that this encompasses more addresses than those defined in [RFC 4291 section 2.5.6], - /// which describes "Link-Local IPv6 Unicast Addresses" as having the following stricter format: - /// - /// ```text - /// | 10 bits | 54 bits | 64 bits | - /// +----------+-------------------------+----------------------------+ - /// |1111111010| 0 | interface ID | - /// +----------+-------------------------+----------------------------+ - /// ``` - /// So while currently the only addresses with link-local scope an application will encounter are all in `fe80::/64`, - /// this might change in the future with the publication of new standards. More addresses in `fe80::/10` could be allocated, - /// and those addresses will have link-local scope. - /// - /// Also note that while [RFC 4291 section 2.5.3] mentions about the [loopback address] (`::1`) that "it is treated as having Link-Local scope", - /// this does not mean that the loopback address actually has link-local scope and this method will return `false` on it. - /// - /// [RFC 4291]: https://tools.ietf.org/html/rfc4291 - /// [RFC 4291 section 2.4]: https://tools.ietf.org/html/rfc4291#section-2.4 - /// [RFC 4291 section 2.5.3]: https://tools.ietf.org/html/rfc4291#section-2.5.3 - /// [RFC 4291 section 2.5.6]: https://tools.ietf.org/html/rfc4291#section-2.5.6 - /// [loopback address]: Ipv6Addr::LOCALHOST - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// // The loopback address (`::1`) does not actually have link-local scope. - /// assert_eq!(Ipv6Addr::LOCALHOST.is_unicast_link_local(), false); - /// - /// // Only addresses in `fe80::/10` have link-local scope. - /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_unicast_link_local(), false); - /// assert_eq!(Ipv6Addr::new(0xfe80, 0, 0, 0, 0, 0, 0, 0).is_unicast_link_local(), true); - /// - /// // Addresses outside the stricter `fe80::/64` also have link-local scope. - /// assert_eq!(Ipv6Addr::new(0xfe80, 0, 0, 1, 0, 0, 0, 0).is_unicast_link_local(), true); - /// assert_eq!(Ipv6Addr::new(0xfe81, 0, 0, 0, 0, 0, 0, 0).is_unicast_link_local(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_unicast_link_local(&self) -> bool { - (self.segments()[0] & 0xffc0) == 0xfe80 - } - - /// Returns [`true`] if this is an address reserved for documentation - /// (`2001:db8::/32`). - /// - /// This property is defined in [IETF RFC 3849]. - /// - /// [IETF RFC 3849]: https://tools.ietf.org/html/rfc3849 - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_documentation(), false); - /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_documentation(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_documentation(&self) -> bool { - (self.segments()[0] == 0x2001) && (self.segments()[1] == 0xdb8) - } - - /// Returns [`true`] if this is an address reserved for benchmarking (`2001:2::/48`). - /// - /// This property is defined in [IETF RFC 5180], where it is mistakenly specified as covering the range `2001:0200::/48`. - /// This is corrected in [IETF RFC Errata 1752] to `2001:0002::/48`. - /// - /// [IETF RFC 5180]: https://tools.ietf.org/html/rfc5180 - /// [IETF RFC Errata 1752]: https://www.rfc-editor.org/errata_search.php?eid=1752 - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc613, 0x0).is_benchmarking(), false); - /// assert_eq!(Ipv6Addr::new(0x2001, 0x2, 0, 0, 0, 0, 0, 0).is_benchmarking(), true); - /// ``` - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_benchmarking(&self) -> bool { - (self.segments()[0] == 0x2001) && (self.segments()[1] == 0x2) && (self.segments()[2] == 0) - } - - /// Returns [`true`] if the address is a globally routable unicast address. - /// - /// The following return false: - /// - /// - the loopback address - /// - the link-local addresses - /// - unique local addresses - /// - the unspecified address - /// - the address range reserved for documentation - /// - /// This method returns [`true`] for site-local addresses as per [RFC 4291 section 2.5.7] - /// - /// ```no_rust - /// The special behavior of [the site-local unicast] prefix defined in [RFC3513] must no longer - /// be supported in new implementations (i.e., new implementations must treat this prefix as - /// Global Unicast). - /// ``` - /// - /// [RFC 4291 section 2.5.7]: https://tools.ietf.org/html/rfc4291#section-2.5.7 - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_unicast_global(), false); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_unicast_global(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn is_unicast_global(&self) -> bool { - self.is_unicast() - && !self.is_loopback() - && !self.is_unicast_link_local() - && !self.is_unique_local() - && !self.is_unspecified() - && !self.is_documentation() - } - - /// Returns the address's multicast scope if the address is multicast. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// - /// use std::net::{Ipv6Addr, Ipv6MulticastScope}; - /// - /// assert_eq!( - /// Ipv6Addr::new(0xff0e, 0, 0, 0, 0, 0, 0, 0).multicast_scope(), - /// Some(Ipv6MulticastScope::Global) - /// ); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).multicast_scope(), None); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use] - #[inline] - pub const fn multicast_scope(&self) -> Option<Ipv6MulticastScope> { - if self.is_multicast() { - match self.segments()[0] & 0x000f { - 1 => Some(Ipv6MulticastScope::InterfaceLocal), - 2 => Some(Ipv6MulticastScope::LinkLocal), - 3 => Some(Ipv6MulticastScope::RealmLocal), - 4 => Some(Ipv6MulticastScope::AdminLocal), - 5 => Some(Ipv6MulticastScope::SiteLocal), - 8 => Some(Ipv6MulticastScope::OrganizationLocal), - 14 => Some(Ipv6MulticastScope::Global), - _ => None, - } - } else { - None - } - } - - /// Returns [`true`] if this is a multicast address (`ff00::/8`). - /// - /// This property is defined by [IETF RFC 4291]. - /// - /// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).is_multicast(), true); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_multicast(), false); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(since = "1.7.0", feature = "ip_17")] - #[must_use] - #[inline] - pub const fn is_multicast(&self) -> bool { - (self.segments()[0] & 0xff00) == 0xff00 - } - - /// Converts this address to an [`IPv4` address] if it's an [IPv4-mapped] address, - /// as defined in [IETF RFC 4291 section 2.5.5.2], otherwise returns [`None`]. - /// - /// `::ffff:a.b.c.d` becomes `a.b.c.d`. - /// All addresses *not* starting with `::ffff` will return `None`. - /// - /// [`IPv4` address]: Ipv4Addr - /// [IPv4-mapped]: Ipv6Addr - /// [IETF RFC 4291 section 2.5.5.2]: https://tools.ietf.org/html/rfc4291#section-2.5.5.2 - /// - /// # Examples - /// - /// ``` - /// use std::net::{Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).to_ipv4_mapped(), None); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).to_ipv4_mapped(), - /// Some(Ipv4Addr::new(192, 10, 2, 255))); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1).to_ipv4_mapped(), None); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[stable(feature = "ipv6_to_ipv4_mapped", since = "1.63.0")] - #[must_use = "this returns the result of the operation, \ - without modifying the original"] - #[inline] - pub const fn to_ipv4_mapped(&self) -> Option<Ipv4Addr> { - match self.octets() { - [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, a, b, c, d] => { - Some(Ipv4Addr::new(a, b, c, d)) - } - _ => None, - } - } - - /// Converts this address to an [`IPv4` address] if it is either - /// an [IPv4-compatible] address as defined in [IETF RFC 4291 section 2.5.5.1], - /// or an [IPv4-mapped] address as defined in [IETF RFC 4291 section 2.5.5.2], - /// otherwise returns [`None`]. - /// - /// Note that this will return an [`IPv4` address] for the IPv6 loopback address `::1`. Use - /// [`Ipv6Addr::to_ipv4_mapped`] to avoid this. - /// - /// `::a.b.c.d` and `::ffff:a.b.c.d` become `a.b.c.d`. `::1` becomes `0.0.0.1`. - /// All addresses *not* starting with either all zeroes or `::ffff` will return `None`. - /// - /// [`IPv4` address]: Ipv4Addr - /// [IPv4-compatible]: Ipv6Addr#ipv4-compatible-ipv6-addresses - /// [IPv4-mapped]: Ipv6Addr#ipv4-mapped-ipv6-addresses - /// [IETF RFC 4291 section 2.5.5.1]: https://tools.ietf.org/html/rfc4291#section-2.5.5.1 - /// [IETF RFC 4291 section 2.5.5.2]: https://tools.ietf.org/html/rfc4291#section-2.5.5.2 - /// - /// # Examples - /// - /// ``` - /// use std::net::{Ipv4Addr, Ipv6Addr}; - /// - /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).to_ipv4(), None); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).to_ipv4(), - /// Some(Ipv4Addr::new(192, 10, 2, 255))); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1).to_ipv4(), - /// Some(Ipv4Addr::new(0, 0, 0, 1))); - /// ``` - #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] - #[stable(feature = "rust1", since = "1.0.0")] - #[must_use = "this returns the result of the operation, \ - without modifying the original"] - #[inline] - pub const fn to_ipv4(&self) -> Option<Ipv4Addr> { - if let [0, 0, 0, 0, 0, 0 | 0xffff, ab, cd] = self.segments() { - let [a, b] = ab.to_be_bytes(); - let [c, d] = cd.to_be_bytes(); - Some(Ipv4Addr::new(a, b, c, d)) - } else { - None - } - } - - /// Converts this address to an `IpAddr::V4` if it is an IPv4-mapped addresses, otherwise it - /// returns self wrapped in an `IpAddr::V6`. - /// - /// # Examples - /// - /// ``` - /// #![feature(ip)] - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1).is_loopback(), false); - /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1).to_canonical().is_loopback(), true); - /// ``` - #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] - #[unstable(feature = "ip", issue = "27709")] - #[must_use = "this returns the result of the operation, \ - without modifying the original"] - #[inline] - pub const fn to_canonical(&self) -> IpAddr { - if let Some(mapped) = self.to_ipv4_mapped() { - return IpAddr::V4(mapped); - } - IpAddr::V6(*self) - } - - /// Returns the sixteen eight-bit integers the IPv6 address consists of. - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).octets(), - /// [255, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]); - /// ``` - #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] - #[stable(feature = "ipv6_to_octets", since = "1.12.0")] - #[must_use] - #[inline] - pub const fn octets(&self) -> [u8; 16] { - self.octets - } -} - -/// Write an Ipv6Addr, conforming to the canonical style described by -/// [RFC 5952](https://tools.ietf.org/html/rfc5952). -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Display for Ipv6Addr { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - // If there are no alignment requirements, write out the IP address to - // f. Otherwise, write it to a local buffer, then use f.pad. - if f.precision().is_none() && f.width().is_none() { - let segments = self.segments(); - - // Special case for :: and ::1; otherwise they get written with the - // IPv4 formatter - if self.is_unspecified() { - f.write_str("::") - } else if self.is_loopback() { - f.write_str("::1") - } else if let Some(ipv4) = self.to_ipv4() { - match segments[5] { - // IPv4 Compatible address - 0 => write!(f, "::{}", ipv4), - // IPv4 Mapped address - 0xffff => write!(f, "::ffff:{}", ipv4), - _ => unreachable!(), - } - } else { - #[derive(Copy, Clone, Default)] - struct Span { - start: usize, - len: usize, - } - - // Find the inner 0 span - let zeroes = { - let mut longest = Span::default(); - let mut current = Span::default(); - - for (i, &segment) in segments.iter().enumerate() { - if segment == 0 { - if current.len == 0 { - current.start = i; - } - - current.len += 1; - - if current.len > longest.len { - longest = current; - } - } else { - current = Span::default(); - } - } - - longest - }; - - /// Write a colon-separated part of the address - #[inline] - fn fmt_subslice(f: &mut fmt::Formatter<'_>, chunk: &[u16]) -> fmt::Result { - if let Some((first, tail)) = chunk.split_first() { - write!(f, "{:x}", first)?; - for segment in tail { - f.write_char(':')?; - write!(f, "{:x}", segment)?; - } - } - Ok(()) - } - - if zeroes.len > 1 { - fmt_subslice(f, &segments[..zeroes.start])?; - f.write_str("::")?; - fmt_subslice(f, &segments[zeroes.start + zeroes.len..]) - } else { - fmt_subslice(f, &segments) - } - } - } else { - // Slow path: write the address to a local buffer, then use f.pad. - // Defined recursively by using the fast path to write to the - // buffer. - - // This is the largest possible size of an IPv6 address - const IPV6_BUF_LEN: usize = (4 * 8) + 7; - let mut buf = [0u8; IPV6_BUF_LEN]; - let mut buf_slice = &mut buf[..]; - - // Note: This call to write should never fail, so unwrap is okay. - write!(buf_slice, "{}", self).unwrap(); - let len = IPV6_BUF_LEN - buf_slice.len(); - - // This is safe because we know exactly what can be in this buffer - let buf = unsafe { crate::str::from_utf8_unchecked(&buf[..len]) }; - f.pad(buf) - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl fmt::Debug for Ipv6Addr { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(self, fmt) - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialEq<IpAddr> for Ipv6Addr { - #[inline] - fn eq(&self, other: &IpAddr) -> bool { - match other { - IpAddr::V4(_) => false, - IpAddr::V6(v6) => self == v6, - } - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialEq<Ipv6Addr> for IpAddr { - #[inline] - fn eq(&self, other: &Ipv6Addr) -> bool { - match self { - IpAddr::V4(_) => false, - IpAddr::V6(v6) => v6 == other, - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl PartialOrd for Ipv6Addr { - #[inline] - fn partial_cmp(&self, other: &Ipv6Addr) -> Option<Ordering> { - Some(self.cmp(other)) - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialOrd<Ipv6Addr> for IpAddr { - #[inline] - fn partial_cmp(&self, other: &Ipv6Addr) -> Option<Ordering> { - match self { - IpAddr::V4(_) => Some(Ordering::Less), - IpAddr::V6(v6) => v6.partial_cmp(other), - } - } -} - -#[stable(feature = "ip_cmp", since = "1.16.0")] -impl PartialOrd<IpAddr> for Ipv6Addr { - #[inline] - fn partial_cmp(&self, other: &IpAddr) -> Option<Ordering> { - match other { - IpAddr::V4(_) => Some(Ordering::Greater), - IpAddr::V6(v6) => self.partial_cmp(v6), - } - } -} - -#[stable(feature = "rust1", since = "1.0.0")] -impl Ord for Ipv6Addr { - #[inline] - fn cmp(&self, other: &Ipv6Addr) -> Ordering { - self.segments().cmp(&other.segments()) - } -} - -impl IntoInner<c::in6_addr> for Ipv6Addr { - fn into_inner(self) -> c::in6_addr { - c::in6_addr { s6_addr: self.octets } - } -} -impl FromInner<c::in6_addr> for Ipv6Addr { - #[inline] - fn from_inner(addr: c::in6_addr) -> Ipv6Addr { - Ipv6Addr { octets: addr.s6_addr } - } -} - -#[stable(feature = "i128", since = "1.26.0")] -impl From<Ipv6Addr> for u128 { - /// Convert an `Ipv6Addr` into a host byte order `u128`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::new( - /// 0x1020, 0x3040, 0x5060, 0x7080, - /// 0x90A0, 0xB0C0, 0xD0E0, 0xF00D, - /// ); - /// assert_eq!(0x102030405060708090A0B0C0D0E0F00D_u128, u128::from(addr)); - /// ``` - #[inline] - fn from(ip: Ipv6Addr) -> u128 { - u128::from_be_bytes(ip.octets) - } -} -#[stable(feature = "i128", since = "1.26.0")] -impl From<u128> for Ipv6Addr { - /// Convert a host byte order `u128` into an `Ipv6Addr`. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::from(0x102030405060708090A0B0C0D0E0F00D_u128); - /// assert_eq!( - /// Ipv6Addr::new( - /// 0x1020, 0x3040, 0x5060, 0x7080, - /// 0x90A0, 0xB0C0, 0xD0E0, 0xF00D, - /// ), - /// addr); - /// ``` - #[inline] - fn from(ip: u128) -> Ipv6Addr { - Ipv6Addr::from(ip.to_be_bytes()) - } -} - -#[stable(feature = "ipv6_from_octets", since = "1.9.0")] -impl From<[u8; 16]> for Ipv6Addr { - /// Creates an `Ipv6Addr` from a sixteen element byte array. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::from([ - /// 25u8, 24u8, 23u8, 22u8, 21u8, 20u8, 19u8, 18u8, - /// 17u8, 16u8, 15u8, 14u8, 13u8, 12u8, 11u8, 10u8, - /// ]); - /// assert_eq!( - /// Ipv6Addr::new( - /// 0x1918, 0x1716, - /// 0x1514, 0x1312, - /// 0x1110, 0x0f0e, - /// 0x0d0c, 0x0b0a - /// ), - /// addr - /// ); - /// ``` - #[inline] - fn from(octets: [u8; 16]) -> Ipv6Addr { - Ipv6Addr { octets } - } -} - -#[stable(feature = "ipv6_from_segments", since = "1.16.0")] -impl From<[u16; 8]> for Ipv6Addr { - /// Creates an `Ipv6Addr` from an eight element 16-bit array. - /// - /// # Examples - /// - /// ``` - /// use std::net::Ipv6Addr; - /// - /// let addr = Ipv6Addr::from([ - /// 525u16, 524u16, 523u16, 522u16, - /// 521u16, 520u16, 519u16, 518u16, - /// ]); - /// assert_eq!( - /// Ipv6Addr::new( - /// 0x20d, 0x20c, - /// 0x20b, 0x20a, - /// 0x209, 0x208, - /// 0x207, 0x206 - /// ), - /// addr - /// ); - /// ``` - #[inline] - fn from(segments: [u16; 8]) -> Ipv6Addr { - let [a, b, c, d, e, f, g, h] = segments; - Ipv6Addr::new(a, b, c, d, e, f, g, h) - } -} - -#[stable(feature = "ip_from_slice", since = "1.17.0")] -impl From<[u8; 16]> for IpAddr { - /// Creates an `IpAddr::V6` from a sixteen element byte array. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv6Addr}; - /// - /// let addr = IpAddr::from([ - /// 25u8, 24u8, 23u8, 22u8, 21u8, 20u8, 19u8, 18u8, - /// 17u8, 16u8, 15u8, 14u8, 13u8, 12u8, 11u8, 10u8, - /// ]); - /// assert_eq!( - /// IpAddr::V6(Ipv6Addr::new( - /// 0x1918, 0x1716, - /// 0x1514, 0x1312, - /// 0x1110, 0x0f0e, - /// 0x0d0c, 0x0b0a - /// )), - /// addr - /// ); - /// ``` - #[inline] - fn from(octets: [u8; 16]) -> IpAddr { - IpAddr::V6(Ipv6Addr::from(octets)) - } -} - -#[stable(feature = "ip_from_slice", since = "1.17.0")] -impl From<[u16; 8]> for IpAddr { - /// Creates an `IpAddr::V6` from an eight element 16-bit array. - /// - /// # Examples - /// - /// ``` - /// use std::net::{IpAddr, Ipv6Addr}; - /// - /// let addr = IpAddr::from([ - /// 525u16, 524u16, 523u16, 522u16, - /// 521u16, 520u16, 519u16, 518u16, - /// ]); - /// assert_eq!( - /// IpAddr::V6(Ipv6Addr::new( - /// 0x20d, 0x20c, - /// 0x20b, 0x20a, - /// 0x209, 0x208, - /// 0x207, 0x206 - /// )), - /// addr - /// ); - /// ``` - #[inline] - fn from(segments: [u16; 8]) -> IpAddr { - IpAddr::V6(Ipv6Addr::from(segments)) - } -} diff --git a/library/std/src/net/ip/tests.rs b/library/std/src/net/ip/tests.rs deleted file mode 100644 index c29509331..000000000 --- a/library/std/src/net/ip/tests.rs +++ /dev/null @@ -1,969 +0,0 @@ -use crate::net::test::{sa4, sa6, tsa}; -use crate::net::*; -use crate::str::FromStr; - -#[test] -fn test_from_str_ipv4() { - assert_eq!(Ok(Ipv4Addr::new(127, 0, 0, 1)), "127.0.0.1".parse()); - assert_eq!(Ok(Ipv4Addr::new(255, 255, 255, 255)), "255.255.255.255".parse()); - assert_eq!(Ok(Ipv4Addr::new(0, 0, 0, 0)), "0.0.0.0".parse()); - - // out of range - let none: Option<Ipv4Addr> = "256.0.0.1".parse().ok(); - assert_eq!(None, none); - // too short - let none: Option<Ipv4Addr> = "255.0.0".parse().ok(); - assert_eq!(None, none); - // too long - let none: Option<Ipv4Addr> = "255.0.0.1.2".parse().ok(); - assert_eq!(None, none); - // no number between dots - let none: Option<Ipv4Addr> = "255.0..1".parse().ok(); - assert_eq!(None, none); - // octal - let none: Option<Ipv4Addr> = "255.0.0.01".parse().ok(); - assert_eq!(None, none); - // octal zero - let none: Option<Ipv4Addr> = "255.0.0.00".parse().ok(); - assert_eq!(None, none); - let none: Option<Ipv4Addr> = "255.0.00.0".parse().ok(); - assert_eq!(None, none); -} - -#[test] -fn test_from_str_ipv6() { - assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)), "0:0:0:0:0:0:0:0".parse()); - assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)), "0:0:0:0:0:0:0:1".parse()); - - assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)), "::1".parse()); - assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)), "::".parse()); - - assert_eq!(Ok(Ipv6Addr::new(0x2a02, 0x6b8, 0, 0, 0, 0, 0x11, 0x11)), "2a02:6b8::11:11".parse()); - - // too long group - let none: Option<Ipv6Addr> = "::00000".parse().ok(); - assert_eq!(None, none); - // too short - let none: Option<Ipv6Addr> = "1:2:3:4:5:6:7".parse().ok(); - assert_eq!(None, none); - // too long - let none: Option<Ipv6Addr> = "1:2:3:4:5:6:7:8:9".parse().ok(); - assert_eq!(None, none); - // triple colon - let none: Option<Ipv6Addr> = "1:2:::6:7:8".parse().ok(); - assert_eq!(None, none); - // two double colons - let none: Option<Ipv6Addr> = "1:2::6::8".parse().ok(); - assert_eq!(None, none); - // `::` indicating zero groups of zeros - let none: Option<Ipv6Addr> = "1:2:3:4::5:6:7:8".parse().ok(); - assert_eq!(None, none); -} - -#[test] -fn test_from_str_ipv4_in_ipv6() { - assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 49152, 545)), "::192.0.2.33".parse()); - assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0xFFFF, 49152, 545)), "::FFFF:192.0.2.33".parse()); - assert_eq!( - Ok(Ipv6Addr::new(0x64, 0xff9b, 0, 0, 0, 0, 49152, 545)), - "64:ff9b::192.0.2.33".parse() - ); - assert_eq!( - Ok(Ipv6Addr::new(0x2001, 0xdb8, 0x122, 0xc000, 0x2, 0x2100, 49152, 545)), - "2001:db8:122:c000:2:2100:192.0.2.33".parse() - ); - - // colon after v4 - let none: Option<Ipv4Addr> = "::127.0.0.1:".parse().ok(); - assert_eq!(None, none); - // not enough groups - let none: Option<Ipv6Addr> = "1:2:3:4:5:127.0.0.1".parse().ok(); - assert_eq!(None, none); - // too many groups - let none: Option<Ipv6Addr> = "1:2:3:4:5:6:7:127.0.0.1".parse().ok(); - assert_eq!(None, none); -} - -#[test] -fn test_from_str_socket_addr() { - assert_eq!(Ok(sa4(Ipv4Addr::new(77, 88, 21, 11), 80)), "77.88.21.11:80".parse()); - assert_eq!(Ok(SocketAddrV4::new(Ipv4Addr::new(77, 88, 21, 11), 80)), "77.88.21.11:80".parse()); - assert_eq!( - Ok(sa6(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53)), - "[2a02:6b8:0:1::1]:53".parse() - ); - assert_eq!( - Ok(SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53, 0, 0)), - "[2a02:6b8:0:1::1]:53".parse() - ); - assert_eq!(Ok(sa6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x7F00, 1), 22)), "[::127.0.0.1]:22".parse()); - assert_eq!( - Ok(SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x7F00, 1), 22, 0, 0)), - "[::127.0.0.1]:22".parse() - ); - - // without port - let none: Option<SocketAddr> = "127.0.0.1".parse().ok(); - assert_eq!(None, none); - // without port - let none: Option<SocketAddr> = "127.0.0.1:".parse().ok(); - assert_eq!(None, none); - // wrong brackets around v4 - let none: Option<SocketAddr> = "[127.0.0.1]:22".parse().ok(); - assert_eq!(None, none); - // port out of range - let none: Option<SocketAddr> = "127.0.0.1:123456".parse().ok(); - assert_eq!(None, none); -} - -#[test] -fn ipv4_addr_to_string() { - assert_eq!(Ipv4Addr::new(127, 0, 0, 1).to_string(), "127.0.0.1"); - // Short address - assert_eq!(Ipv4Addr::new(1, 1, 1, 1).to_string(), "1.1.1.1"); - // Long address - assert_eq!(Ipv4Addr::new(127, 127, 127, 127).to_string(), "127.127.127.127"); - - // Test padding - assert_eq!(&format!("{:16}", Ipv4Addr::new(1, 1, 1, 1)), "1.1.1.1 "); - assert_eq!(&format!("{:>16}", Ipv4Addr::new(1, 1, 1, 1)), " 1.1.1.1"); -} - -#[test] -fn ipv6_addr_to_string() { - // ipv4-mapped address - let a1 = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc000, 0x280); - assert_eq!(a1.to_string(), "::ffff:192.0.2.128"); - - // ipv4-compatible address - let a1 = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0xc000, 0x280); - assert_eq!(a1.to_string(), "::192.0.2.128"); - - // v6 address with no zero segments - assert_eq!(Ipv6Addr::new(8, 9, 10, 11, 12, 13, 14, 15).to_string(), "8:9:a:b:c:d:e:f"); - - // longest possible IPv6 length - assert_eq!( - Ipv6Addr::new(0x1111, 0x2222, 0x3333, 0x4444, 0x5555, 0x6666, 0x7777, 0x8888).to_string(), - "1111:2222:3333:4444:5555:6666:7777:8888" - ); - // padding - assert_eq!(&format!("{:20}", Ipv6Addr::new(1, 2, 3, 4, 5, 6, 7, 8)), "1:2:3:4:5:6:7:8 "); - assert_eq!(&format!("{:>20}", Ipv6Addr::new(1, 2, 3, 4, 5, 6, 7, 8)), " 1:2:3:4:5:6:7:8"); - - // reduce a single run of zeros - assert_eq!( - "ae::ffff:102:304", - Ipv6Addr::new(0xae, 0, 0, 0, 0, 0xffff, 0x0102, 0x0304).to_string() - ); - - // don't reduce just a single zero segment - assert_eq!("1:2:3:4:5:6:0:8", Ipv6Addr::new(1, 2, 3, 4, 5, 6, 0, 8).to_string()); - - // 'any' address - assert_eq!("::", Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0).to_string()); - - // loopback address - assert_eq!("::1", Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1).to_string()); - - // ends in zeros - assert_eq!("1::", Ipv6Addr::new(1, 0, 0, 0, 0, 0, 0, 0).to_string()); - - // two runs of zeros, second one is longer - assert_eq!("1:0:0:4::8", Ipv6Addr::new(1, 0, 0, 4, 0, 0, 0, 8).to_string()); - - // two runs of zeros, equal length - assert_eq!("1::4:5:0:0:8", Ipv6Addr::new(1, 0, 0, 4, 5, 0, 0, 8).to_string()); - - // don't prefix `0x` to each segment in `dbg!`. - assert_eq!("1::4:5:0:0:8", &format!("{:#?}", Ipv6Addr::new(1, 0, 0, 4, 5, 0, 0, 8))); -} - -#[test] -fn ipv4_to_ipv6() { - assert_eq!( - Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678), - Ipv4Addr::new(0x12, 0x34, 0x56, 0x78).to_ipv6_mapped() - ); - assert_eq!( - Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x1234, 0x5678), - Ipv4Addr::new(0x12, 0x34, 0x56, 0x78).to_ipv6_compatible() - ); -} - -#[test] -fn ipv6_to_ipv4_mapped() { - assert_eq!( - Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678).to_ipv4_mapped(), - Some(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78)) - ); - assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x1234, 0x5678).to_ipv4_mapped(), None); -} - -#[test] -fn ipv6_to_ipv4() { - assert_eq!( - Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678).to_ipv4(), - Some(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78)) - ); - assert_eq!( - Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x1234, 0x5678).to_ipv4(), - Some(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78)) - ); - assert_eq!(Ipv6Addr::new(0, 0, 1, 0, 0, 0, 0x1234, 0x5678).to_ipv4(), None); -} - -#[test] -fn ip_properties() { - macro_rules! ip { - ($s:expr) => { - IpAddr::from_str($s).unwrap() - }; - } - - macro_rules! check { - ($s:expr) => { - check!($s, 0); - }; - - ($s:expr, $mask:expr) => {{ - let unspec: u8 = 1 << 0; - let loopback: u8 = 1 << 1; - let global: u8 = 1 << 2; - let multicast: u8 = 1 << 3; - let doc: u8 = 1 << 4; - let benchmarking: u8 = 1 << 5; - - if ($mask & unspec) == unspec { - assert!(ip!($s).is_unspecified()); - } else { - assert!(!ip!($s).is_unspecified()); - } - - if ($mask & loopback) == loopback { - assert!(ip!($s).is_loopback()); - } else { - assert!(!ip!($s).is_loopback()); - } - - if ($mask & global) == global { - assert!(ip!($s).is_global()); - } else { - assert!(!ip!($s).is_global()); - } - - if ($mask & multicast) == multicast { - assert!(ip!($s).is_multicast()); - } else { - assert!(!ip!($s).is_multicast()); - } - - if ($mask & doc) == doc { - assert!(ip!($s).is_documentation()); - } else { - assert!(!ip!($s).is_documentation()); - } - - if ($mask & benchmarking) == benchmarking { - assert!(ip!($s).is_benchmarking()); - } else { - assert!(!ip!($s).is_benchmarking()); - } - }}; - } - - let unspec: u8 = 1 << 0; - let loopback: u8 = 1 << 1; - let global: u8 = 1 << 2; - let multicast: u8 = 1 << 3; - let doc: u8 = 1 << 4; - let benchmarking: u8 = 1 << 5; - - check!("0.0.0.0", unspec); - check!("0.0.0.1"); - check!("0.1.0.0"); - check!("10.9.8.7"); - check!("127.1.2.3", loopback); - check!("172.31.254.253"); - check!("169.254.253.242"); - check!("192.0.2.183", doc); - check!("192.1.2.183", global); - check!("192.168.254.253"); - check!("198.51.100.0", doc); - check!("203.0.113.0", doc); - check!("203.2.113.0", global); - check!("224.0.0.0", global | multicast); - check!("239.255.255.255", global | multicast); - check!("255.255.255.255"); - // make sure benchmarking addresses are not global - check!("198.18.0.0", benchmarking); - check!("198.18.54.2", benchmarking); - check!("198.19.255.255", benchmarking); - // make sure addresses reserved for protocol assignment are not global - check!("192.0.0.0"); - check!("192.0.0.255"); - check!("192.0.0.100"); - // make sure reserved addresses are not global - check!("240.0.0.0"); - check!("251.54.1.76"); - check!("254.255.255.255"); - // make sure shared addresses are not global - check!("100.64.0.0"); - check!("100.127.255.255"); - check!("100.100.100.0"); - - check!("::", unspec); - check!("::1", loopback); - check!("::0.0.0.2", global); - check!("1::", global); - check!("fc00::"); - check!("fdff:ffff::"); - check!("fe80:ffff::"); - check!("febf:ffff::"); - check!("fec0::", global); - check!("ff01::", multicast); - check!("ff02::", multicast); - check!("ff03::", multicast); - check!("ff04::", multicast); - check!("ff05::", multicast); - check!("ff08::", multicast); - check!("ff0e::", global | multicast); - check!("2001:db8:85a3::8a2e:370:7334", doc); - check!("2001:2::ac32:23ff:21", global | benchmarking); - check!("102:304:506:708:90a:b0c:d0e:f10", global); -} - -#[test] -fn ipv4_properties() { - macro_rules! ip { - ($s:expr) => { - Ipv4Addr::from_str($s).unwrap() - }; - } - - macro_rules! check { - ($s:expr) => { - check!($s, 0); - }; - - ($s:expr, $mask:expr) => {{ - let unspec: u16 = 1 << 0; - let loopback: u16 = 1 << 1; - let private: u16 = 1 << 2; - let link_local: u16 = 1 << 3; - let global: u16 = 1 << 4; - let multicast: u16 = 1 << 5; - let broadcast: u16 = 1 << 6; - let documentation: u16 = 1 << 7; - let benchmarking: u16 = 1 << 8; - let reserved: u16 = 1 << 10; - let shared: u16 = 1 << 11; - - if ($mask & unspec) == unspec { - assert!(ip!($s).is_unspecified()); - } else { - assert!(!ip!($s).is_unspecified()); - } - - if ($mask & loopback) == loopback { - assert!(ip!($s).is_loopback()); - } else { - assert!(!ip!($s).is_loopback()); - } - - if ($mask & private) == private { - assert!(ip!($s).is_private()); - } else { - assert!(!ip!($s).is_private()); - } - - if ($mask & link_local) == link_local { - assert!(ip!($s).is_link_local()); - } else { - assert!(!ip!($s).is_link_local()); - } - - if ($mask & global) == global { - assert!(ip!($s).is_global()); - } else { - assert!(!ip!($s).is_global()); - } - - if ($mask & multicast) == multicast { - assert!(ip!($s).is_multicast()); - } else { - assert!(!ip!($s).is_multicast()); - } - - if ($mask & broadcast) == broadcast { - assert!(ip!($s).is_broadcast()); - } else { - assert!(!ip!($s).is_broadcast()); - } - - if ($mask & documentation) == documentation { - assert!(ip!($s).is_documentation()); - } else { - assert!(!ip!($s).is_documentation()); - } - - if ($mask & benchmarking) == benchmarking { - assert!(ip!($s).is_benchmarking()); - } else { - assert!(!ip!($s).is_benchmarking()); - } - - if ($mask & reserved) == reserved { - assert!(ip!($s).is_reserved()); - } else { - assert!(!ip!($s).is_reserved()); - } - - if ($mask & shared) == shared { - assert!(ip!($s).is_shared()); - } else { - assert!(!ip!($s).is_shared()); - } - }}; - } - - let unspec: u16 = 1 << 0; - let loopback: u16 = 1 << 1; - let private: u16 = 1 << 2; - let link_local: u16 = 1 << 3; - let global: u16 = 1 << 4; - let multicast: u16 = 1 << 5; - let broadcast: u16 = 1 << 6; - let documentation: u16 = 1 << 7; - let benchmarking: u16 = 1 << 8; - let reserved: u16 = 1 << 10; - let shared: u16 = 1 << 11; - - check!("0.0.0.0", unspec); - check!("0.0.0.1"); - check!("0.1.0.0"); - check!("10.9.8.7", private); - check!("127.1.2.3", loopback); - check!("172.31.254.253", private); - check!("169.254.253.242", link_local); - check!("192.0.2.183", documentation); - check!("192.1.2.183", global); - check!("192.168.254.253", private); - check!("198.51.100.0", documentation); - check!("203.0.113.0", documentation); - check!("203.2.113.0", global); - check!("224.0.0.0", global | multicast); - check!("239.255.255.255", global | multicast); - check!("255.255.255.255", broadcast); - check!("198.18.0.0", benchmarking); - check!("198.18.54.2", benchmarking); - check!("198.19.255.255", benchmarking); - check!("192.0.0.0"); - check!("192.0.0.255"); - check!("192.0.0.100"); - check!("240.0.0.0", reserved); - check!("251.54.1.76", reserved); - check!("254.255.255.255", reserved); - check!("100.64.0.0", shared); - check!("100.127.255.255", shared); - check!("100.100.100.0", shared); -} - -#[test] -fn ipv6_properties() { - macro_rules! ip { - ($s:expr) => { - Ipv6Addr::from_str($s).unwrap() - }; - } - - macro_rules! check { - ($s:expr, &[$($octet:expr),*], $mask:expr) => { - assert_eq!($s, ip!($s).to_string()); - let octets = &[$($octet),*]; - assert_eq!(&ip!($s).octets(), octets); - assert_eq!(Ipv6Addr::from(*octets), ip!($s)); - - let unspecified: u32 = 1 << 0; - let loopback: u32 = 1 << 1; - let unique_local: u32 = 1 << 2; - let global: u32 = 1 << 3; - let unicast_link_local: u32 = 1 << 4; - let unicast_global: u32 = 1 << 7; - let documentation: u32 = 1 << 8; - let benchmarking: u32 = 1 << 16; - let multicast_interface_local: u32 = 1 << 9; - let multicast_link_local: u32 = 1 << 10; - let multicast_realm_local: u32 = 1 << 11; - let multicast_admin_local: u32 = 1 << 12; - let multicast_site_local: u32 = 1 << 13; - let multicast_organization_local: u32 = 1 << 14; - let multicast_global: u32 = 1 << 15; - let multicast: u32 = multicast_interface_local - | multicast_admin_local - | multicast_global - | multicast_link_local - | multicast_realm_local - | multicast_site_local - | multicast_organization_local; - - if ($mask & unspecified) == unspecified { - assert!(ip!($s).is_unspecified()); - } else { - assert!(!ip!($s).is_unspecified()); - } - if ($mask & loopback) == loopback { - assert!(ip!($s).is_loopback()); - } else { - assert!(!ip!($s).is_loopback()); - } - if ($mask & unique_local) == unique_local { - assert!(ip!($s).is_unique_local()); - } else { - assert!(!ip!($s).is_unique_local()); - } - if ($mask & global) == global { - assert!(ip!($s).is_global()); - } else { - assert!(!ip!($s).is_global()); - } - if ($mask & unicast_link_local) == unicast_link_local { - assert!(ip!($s).is_unicast_link_local()); - } else { - assert!(!ip!($s).is_unicast_link_local()); - } - if ($mask & unicast_global) == unicast_global { - assert!(ip!($s).is_unicast_global()); - } else { - assert!(!ip!($s).is_unicast_global()); - } - if ($mask & documentation) == documentation { - assert!(ip!($s).is_documentation()); - } else { - assert!(!ip!($s).is_documentation()); - } - if ($mask & benchmarking) == benchmarking { - assert!(ip!($s).is_benchmarking()); - } else { - assert!(!ip!($s).is_benchmarking()); - } - if ($mask & multicast) != 0 { - assert!(ip!($s).multicast_scope().is_some()); - assert!(ip!($s).is_multicast()); - } else { - assert!(ip!($s).multicast_scope().is_none()); - assert!(!ip!($s).is_multicast()); - } - if ($mask & multicast_interface_local) == multicast_interface_local { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::InterfaceLocal); - } - if ($mask & multicast_link_local) == multicast_link_local { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::LinkLocal); - } - if ($mask & multicast_realm_local) == multicast_realm_local { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::RealmLocal); - } - if ($mask & multicast_admin_local) == multicast_admin_local { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::AdminLocal); - } - if ($mask & multicast_site_local) == multicast_site_local { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::SiteLocal); - } - if ($mask & multicast_organization_local) == multicast_organization_local { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::OrganizationLocal); - } - if ($mask & multicast_global) == multicast_global { - assert_eq!(ip!($s).multicast_scope().unwrap(), - Ipv6MulticastScope::Global); - } - } - } - - let unspecified: u32 = 1 << 0; - let loopback: u32 = 1 << 1; - let unique_local: u32 = 1 << 2; - let global: u32 = 1 << 3; - let unicast_link_local: u32 = 1 << 4; - let unicast_global: u32 = 1 << 7; - let documentation: u32 = 1 << 8; - let benchmarking: u32 = 1 << 16; - let multicast_interface_local: u32 = 1 << 9; - let multicast_link_local: u32 = 1 << 10; - let multicast_realm_local: u32 = 1 << 11; - let multicast_admin_local: u32 = 1 << 12; - let multicast_site_local: u32 = 1 << 13; - let multicast_organization_local: u32 = 1 << 14; - let multicast_global: u32 = 1 << 15; - - check!("::", &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unspecified); - - check!("::1", &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1], loopback); - - check!("::0.0.0.2", &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2], global | unicast_global); - - check!("1::", &[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], global | unicast_global); - - check!("fc00::", &[0xfc, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unique_local); - - check!( - "fdff:ffff::", - &[0xfd, 0xff, 0xff, 0xff, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - unique_local - ); - - check!( - "fe80:ffff::", - &[0xfe, 0x80, 0xff, 0xff, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - unicast_link_local - ); - - check!("fe80::", &[0xfe, 0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_link_local); - - check!( - "febf:ffff::", - &[0xfe, 0xbf, 0xff, 0xff, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - unicast_link_local - ); - - check!("febf::", &[0xfe, 0xbf, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_link_local); - - check!( - "febf:ffff:ffff:ffff:ffff:ffff:ffff:ffff", - &[ - 0xfe, 0xbf, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff - ], - unicast_link_local - ); - - check!( - "fe80::ffff:ffff:ffff:ffff", - &[ - 0xfe, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, - 0xff, 0xff - ], - unicast_link_local - ); - - check!( - "fe80:0:0:1::", - &[0xfe, 0x80, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0], - unicast_link_local - ); - - check!( - "fec0::", - &[0xfe, 0xc0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - unicast_global | global - ); - - check!( - "ff01::", - &[0xff, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - multicast_interface_local - ); - - check!("ff02::", &[0xff, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], multicast_link_local); - - check!("ff03::", &[0xff, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], multicast_realm_local); - - check!("ff04::", &[0xff, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], multicast_admin_local); - - check!("ff05::", &[0xff, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], multicast_site_local); - - check!( - "ff08::", - &[0xff, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - multicast_organization_local - ); - - check!( - "ff0e::", - &[0xff, 0xe, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], - multicast_global | global - ); - - check!( - "2001:db8:85a3::8a2e:370:7334", - &[0x20, 1, 0xd, 0xb8, 0x85, 0xa3, 0, 0, 0, 0, 0x8a, 0x2e, 3, 0x70, 0x73, 0x34], - documentation - ); - - check!( - "2001:2::ac32:23ff:21", - &[0x20, 1, 0, 2, 0, 0, 0, 0, 0, 0, 0xac, 0x32, 0x23, 0xff, 0, 0x21], - global | unicast_global | benchmarking - ); - - check!( - "102:304:506:708:90a:b0c:d0e:f10", - &[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], - global | unicast_global - ); -} - -#[test] -fn to_socket_addr_socketaddr() { - let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 12345); - assert_eq!(Ok(vec![a]), tsa(a)); -} - -#[test] -fn test_ipv4_to_int() { - let a = Ipv4Addr::new(0x11, 0x22, 0x33, 0x44); - assert_eq!(u32::from(a), 0x11223344); -} - -#[test] -fn test_int_to_ipv4() { - let a = Ipv4Addr::new(0x11, 0x22, 0x33, 0x44); - assert_eq!(Ipv4Addr::from(0x11223344), a); -} - -#[test] -fn test_ipv6_to_int() { - let a = Ipv6Addr::new(0x1122, 0x3344, 0x5566, 0x7788, 0x99aa, 0xbbcc, 0xddee, 0xff11); - assert_eq!(u128::from(a), 0x112233445566778899aabbccddeeff11u128); -} - -#[test] -fn test_int_to_ipv6() { - let a = Ipv6Addr::new(0x1122, 0x3344, 0x5566, 0x7788, 0x99aa, 0xbbcc, 0xddee, 0xff11); - assert_eq!(Ipv6Addr::from(0x112233445566778899aabbccddeeff11u128), a); -} - -#[test] -fn ipv4_from_constructors() { - assert_eq!(Ipv4Addr::LOCALHOST, Ipv4Addr::new(127, 0, 0, 1)); - assert!(Ipv4Addr::LOCALHOST.is_loopback()); - assert_eq!(Ipv4Addr::UNSPECIFIED, Ipv4Addr::new(0, 0, 0, 0)); - assert!(Ipv4Addr::UNSPECIFIED.is_unspecified()); - assert_eq!(Ipv4Addr::BROADCAST, Ipv4Addr::new(255, 255, 255, 255)); - assert!(Ipv4Addr::BROADCAST.is_broadcast()); -} - -#[test] -fn ipv6_from_constructors() { - assert_eq!(Ipv6Addr::LOCALHOST, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); - assert!(Ipv6Addr::LOCALHOST.is_loopback()); - assert_eq!(Ipv6Addr::UNSPECIFIED, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)); - assert!(Ipv6Addr::UNSPECIFIED.is_unspecified()); -} - -#[test] -fn ipv4_from_octets() { - assert_eq!(Ipv4Addr::from([127, 0, 0, 1]), Ipv4Addr::new(127, 0, 0, 1)) -} - -#[test] -fn ipv6_from_segments() { - let from_u16s = - Ipv6Addr::from([0x0011, 0x2233, 0x4455, 0x6677, 0x8899, 0xaabb, 0xccdd, 0xeeff]); - let new = Ipv6Addr::new(0x0011, 0x2233, 0x4455, 0x6677, 0x8899, 0xaabb, 0xccdd, 0xeeff); - assert_eq!(new, from_u16s); -} - -#[test] -fn ipv6_from_octets() { - let from_u16s = - Ipv6Addr::from([0x0011, 0x2233, 0x4455, 0x6677, 0x8899, 0xaabb, 0xccdd, 0xeeff]); - let from_u8s = Ipv6Addr::from([ - 0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77, 0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, - 0xff, - ]); - assert_eq!(from_u16s, from_u8s); -} - -#[test] -fn cmp() { - let v41 = Ipv4Addr::new(100, 64, 3, 3); - let v42 = Ipv4Addr::new(192, 0, 2, 2); - let v61 = "2001:db8:f00::1002".parse::<Ipv6Addr>().unwrap(); - let v62 = "2001:db8:f00::2001".parse::<Ipv6Addr>().unwrap(); - assert!(v41 < v42); - assert!(v61 < v62); - - assert_eq!(v41, IpAddr::V4(v41)); - assert_eq!(v61, IpAddr::V6(v61)); - assert!(v41 != IpAddr::V4(v42)); - assert!(v61 != IpAddr::V6(v62)); - - assert!(v41 < IpAddr::V4(v42)); - assert!(v61 < IpAddr::V6(v62)); - assert!(IpAddr::V4(v41) < v42); - assert!(IpAddr::V6(v61) < v62); - - assert!(v41 < IpAddr::V6(v61)); - assert!(IpAddr::V4(v41) < v61); -} - -#[test] -fn is_v4() { - let ip = IpAddr::V4(Ipv4Addr::new(100, 64, 3, 3)); - assert!(ip.is_ipv4()); - assert!(!ip.is_ipv6()); -} - -#[test] -fn is_v6() { - let ip = IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678)); - assert!(!ip.is_ipv4()); - assert!(ip.is_ipv6()); -} - -#[test] -fn ipv4_const() { - // test that the methods of `Ipv4Addr` are usable in a const context - - const IP_ADDRESS: Ipv4Addr = Ipv4Addr::new(127, 0, 0, 1); - assert_eq!(IP_ADDRESS, Ipv4Addr::LOCALHOST); - - const OCTETS: [u8; 4] = IP_ADDRESS.octets(); - assert_eq!(OCTETS, [127, 0, 0, 1]); - - const IS_UNSPECIFIED: bool = IP_ADDRESS.is_unspecified(); - assert!(!IS_UNSPECIFIED); - - const IS_LOOPBACK: bool = IP_ADDRESS.is_loopback(); - assert!(IS_LOOPBACK); - - const IS_PRIVATE: bool = IP_ADDRESS.is_private(); - assert!(!IS_PRIVATE); - - const IS_LINK_LOCAL: bool = IP_ADDRESS.is_link_local(); - assert!(!IS_LINK_LOCAL); - - const IS_GLOBAL: bool = IP_ADDRESS.is_global(); - assert!(!IS_GLOBAL); - - const IS_SHARED: bool = IP_ADDRESS.is_shared(); - assert!(!IS_SHARED); - - const IS_BENCHMARKING: bool = IP_ADDRESS.is_benchmarking(); - assert!(!IS_BENCHMARKING); - - const IS_RESERVED: bool = IP_ADDRESS.is_reserved(); - assert!(!IS_RESERVED); - - const IS_MULTICAST: bool = IP_ADDRESS.is_multicast(); - assert!(!IS_MULTICAST); - - const IS_BROADCAST: bool = IP_ADDRESS.is_broadcast(); - assert!(!IS_BROADCAST); - - const IS_DOCUMENTATION: bool = IP_ADDRESS.is_documentation(); - assert!(!IS_DOCUMENTATION); - - const IP_V6_COMPATIBLE: Ipv6Addr = IP_ADDRESS.to_ipv6_compatible(); - assert_eq!( - IP_V6_COMPATIBLE, - Ipv6Addr::from([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 127, 0, 0, 1]) - ); - - const IP_V6_MAPPED: Ipv6Addr = IP_ADDRESS.to_ipv6_mapped(); - assert_eq!( - IP_V6_MAPPED, - Ipv6Addr::from([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 255, 255, 127, 0, 0, 1]) - ); -} - -#[test] -fn ipv6_const() { - // test that the methods of `Ipv6Addr` are usable in a const context - - const IP_ADDRESS: Ipv6Addr = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); - assert_eq!(IP_ADDRESS, Ipv6Addr::LOCALHOST); - - const SEGMENTS: [u16; 8] = IP_ADDRESS.segments(); - assert_eq!(SEGMENTS, [0, 0, 0, 0, 0, 0, 0, 1]); - - const OCTETS: [u8; 16] = IP_ADDRESS.octets(); - assert_eq!(OCTETS, [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]); - - const IS_UNSPECIFIED: bool = IP_ADDRESS.is_unspecified(); - assert!(!IS_UNSPECIFIED); - - const IS_LOOPBACK: bool = IP_ADDRESS.is_loopback(); - assert!(IS_LOOPBACK); - - const IS_GLOBAL: bool = IP_ADDRESS.is_global(); - assert!(!IS_GLOBAL); - - const IS_UNIQUE_LOCAL: bool = IP_ADDRESS.is_unique_local(); - assert!(!IS_UNIQUE_LOCAL); - - const IS_UNICAST_LINK_LOCAL: bool = IP_ADDRESS.is_unicast_link_local(); - assert!(!IS_UNICAST_LINK_LOCAL); - - const IS_DOCUMENTATION: bool = IP_ADDRESS.is_documentation(); - assert!(!IS_DOCUMENTATION); - - const IS_BENCHMARKING: bool = IP_ADDRESS.is_benchmarking(); - assert!(!IS_BENCHMARKING); - - const IS_UNICAST_GLOBAL: bool = IP_ADDRESS.is_unicast_global(); - assert!(!IS_UNICAST_GLOBAL); - - const MULTICAST_SCOPE: Option<Ipv6MulticastScope> = IP_ADDRESS.multicast_scope(); - assert_eq!(MULTICAST_SCOPE, None); - - const IS_MULTICAST: bool = IP_ADDRESS.is_multicast(); - assert!(!IS_MULTICAST); - - const IP_V4: Option<Ipv4Addr> = IP_ADDRESS.to_ipv4(); - assert_eq!(IP_V4.unwrap(), Ipv4Addr::new(0, 0, 0, 1)); -} - -#[test] -fn ip_const() { - // test that the methods of `IpAddr` are usable in a const context - - const IP_ADDRESS: IpAddr = IpAddr::V4(Ipv4Addr::LOCALHOST); - - const IS_UNSPECIFIED: bool = IP_ADDRESS.is_unspecified(); - assert!(!IS_UNSPECIFIED); - - const IS_LOOPBACK: bool = IP_ADDRESS.is_loopback(); - assert!(IS_LOOPBACK); - - const IS_GLOBAL: bool = IP_ADDRESS.is_global(); - assert!(!IS_GLOBAL); - - const IS_MULTICAST: bool = IP_ADDRESS.is_multicast(); - assert!(!IS_MULTICAST); - - const IS_IP_V4: bool = IP_ADDRESS.is_ipv4(); - assert!(IS_IP_V4); - - const IS_IP_V6: bool = IP_ADDRESS.is_ipv6(); - assert!(!IS_IP_V6); -} - -#[test] -fn structural_match() { - // test that all IP types can be structurally matched upon - - const IPV4: Ipv4Addr = Ipv4Addr::LOCALHOST; - match IPV4 { - Ipv4Addr::LOCALHOST => {} - _ => unreachable!(), - } - - const IPV6: Ipv6Addr = Ipv6Addr::LOCALHOST; - match IPV6 { - Ipv6Addr::LOCALHOST => {} - _ => unreachable!(), - } - - const IP: IpAddr = IpAddr::V4(Ipv4Addr::LOCALHOST); - match IP { - IpAddr::V4(Ipv4Addr::LOCALHOST) => {} - _ => unreachable!(), - } -} diff --git a/library/std/src/net/ip_addr.rs b/library/std/src/net/ip_addr.rs new file mode 100644 index 000000000..4f14fc280 --- /dev/null +++ b/library/std/src/net/ip_addr.rs @@ -0,0 +1,2095 @@ +// Tests for this module +#[cfg(all(test, not(target_os = "emscripten")))] +mod tests; + +use crate::cmp::Ordering; +use crate::fmt::{self, Write}; +use crate::mem::transmute; +use crate::sys::net::netc as c; +use crate::sys_common::{FromInner, IntoInner}; + +use super::display_buffer::DisplayBuffer; + +/// An IP address, either IPv4 or IPv6. +/// +/// This enum can contain either an [`Ipv4Addr`] or an [`Ipv6Addr`], see their +/// respective documentation for more details. +/// +/// # Examples +/// +/// ``` +/// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; +/// +/// let localhost_v4 = IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)); +/// let localhost_v6 = IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); +/// +/// assert_eq!("127.0.0.1".parse(), Ok(localhost_v4)); +/// assert_eq!("::1".parse(), Ok(localhost_v6)); +/// +/// assert_eq!(localhost_v4.is_ipv6(), false); +/// assert_eq!(localhost_v4.is_ipv4(), true); +/// ``` +#[cfg_attr(not(test), rustc_diagnostic_item = "IpAddr")] +#[stable(feature = "ip_addr", since = "1.7.0")] +#[derive(Copy, Clone, Eq, PartialEq, Hash, PartialOrd, Ord)] +pub enum IpAddr { + /// An IPv4 address. + #[stable(feature = "ip_addr", since = "1.7.0")] + V4(#[stable(feature = "ip_addr", since = "1.7.0")] Ipv4Addr), + /// An IPv6 address. + #[stable(feature = "ip_addr", since = "1.7.0")] + V6(#[stable(feature = "ip_addr", since = "1.7.0")] Ipv6Addr), +} + +/// An IPv4 address. +/// +/// IPv4 addresses are defined as 32-bit integers in [IETF RFC 791]. +/// They are usually represented as four octets. +/// +/// See [`IpAddr`] for a type encompassing both IPv4 and IPv6 addresses. +/// +/// [IETF RFC 791]: https://tools.ietf.org/html/rfc791 +/// +/// # Textual representation +/// +/// `Ipv4Addr` provides a [`FromStr`] implementation. The four octets are in decimal +/// notation, divided by `.` (this is called "dot-decimal notation"). +/// Notably, octal numbers (which are indicated with a leading `0`) and hexadecimal numbers (which +/// are indicated with a leading `0x`) are not allowed per [IETF RFC 6943]. +/// +/// [IETF RFC 6943]: https://tools.ietf.org/html/rfc6943#section-3.1.1 +/// [`FromStr`]: crate::str::FromStr +/// +/// # Examples +/// +/// ``` +/// use std::net::Ipv4Addr; +/// +/// let localhost = Ipv4Addr::new(127, 0, 0, 1); +/// assert_eq!("127.0.0.1".parse(), Ok(localhost)); +/// assert_eq!(localhost.is_loopback(), true); +/// assert!("012.004.002.000".parse::<Ipv4Addr>().is_err()); // all octets are in octal +/// assert!("0000000.0.0.0".parse::<Ipv4Addr>().is_err()); // first octet is a zero in octal +/// assert!("0xcb.0x0.0x71.0x00".parse::<Ipv4Addr>().is_err()); // all octets are in hex +/// ``` +#[derive(Copy, Clone, PartialEq, Eq, Hash)] +#[cfg_attr(not(test), rustc_diagnostic_item = "Ipv4Addr")] +#[stable(feature = "rust1", since = "1.0.0")] +pub struct Ipv4Addr { + octets: [u8; 4], +} + +/// An IPv6 address. +/// +/// IPv6 addresses are defined as 128-bit integers in [IETF RFC 4291]. +/// They are usually represented as eight 16-bit segments. +/// +/// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 +/// +/// # Embedding IPv4 Addresses +/// +/// See [`IpAddr`] for a type encompassing both IPv4 and IPv6 addresses. +/// +/// To assist in the transition from IPv4 to IPv6 two types of IPv6 addresses that embed an IPv4 address were defined: +/// IPv4-compatible and IPv4-mapped addresses. Of these IPv4-compatible addresses have been officially deprecated. +/// +/// Both types of addresses are not assigned any special meaning by this implementation, +/// other than what the relevant standards prescribe. This means that an address like `::ffff:127.0.0.1`, +/// while representing an IPv4 loopback address, is not itself an IPv6 loopback address; only `::1` is. +/// To handle these so called "IPv4-in-IPv6" addresses, they have to first be converted to their canonical IPv4 address. +/// +/// ### IPv4-Compatible IPv6 Addresses +/// +/// IPv4-compatible IPv6 addresses are defined in [IETF RFC 4291 Section 2.5.5.1], and have been officially deprecated. +/// The RFC describes the format of an "IPv4-Compatible IPv6 address" as follows: +/// +/// ```text +/// | 80 bits | 16 | 32 bits | +/// +--------------------------------------+--------------------------+ +/// |0000..............................0000|0000| IPv4 address | +/// +--------------------------------------+----+---------------------+ +/// ``` +/// So `::a.b.c.d` would be an IPv4-compatible IPv6 address representing the IPv4 address `a.b.c.d`. +/// +/// To convert from an IPv4 address to an IPv4-compatible IPv6 address, use [`Ipv4Addr::to_ipv6_compatible`]. +/// Use [`Ipv6Addr::to_ipv4`] to convert an IPv4-compatible IPv6 address to the canonical IPv4 address. +/// +/// [IETF RFC 4291 Section 2.5.5.1]: https://datatracker.ietf.org/doc/html/rfc4291#section-2.5.5.1 +/// +/// ### IPv4-Mapped IPv6 Addresses +/// +/// IPv4-mapped IPv6 addresses are defined in [IETF RFC 4291 Section 2.5.5.2]. +/// The RFC describes the format of an "IPv4-Mapped IPv6 address" as follows: +/// +/// ```text +/// | 80 bits | 16 | 32 bits | +/// +--------------------------------------+--------------------------+ +/// |0000..............................0000|FFFF| IPv4 address | +/// +--------------------------------------+----+---------------------+ +/// ``` +/// So `::ffff:a.b.c.d` would be an IPv4-mapped IPv6 address representing the IPv4 address `a.b.c.d`. +/// +/// To convert from an IPv4 address to an IPv4-mapped IPv6 address, use [`Ipv4Addr::to_ipv6_mapped`]. +/// Use [`Ipv6Addr::to_ipv4`] to convert an IPv4-mapped IPv6 address to the canonical IPv4 address. +/// Note that this will also convert the IPv6 loopback address `::1` to `0.0.0.1`. Use +/// [`Ipv6Addr::to_ipv4_mapped`] to avoid this. +/// +/// [IETF RFC 4291 Section 2.5.5.2]: https://datatracker.ietf.org/doc/html/rfc4291#section-2.5.5.2 +/// +/// # Textual representation +/// +/// `Ipv6Addr` provides a [`FromStr`] implementation. There are many ways to represent +/// an IPv6 address in text, but in general, each segments is written in hexadecimal +/// notation, and segments are separated by `:`. For more information, see +/// [IETF RFC 5952]. +/// +/// [`FromStr`]: crate::str::FromStr +/// [IETF RFC 5952]: https://tools.ietf.org/html/rfc5952 +/// +/// # Examples +/// +/// ``` +/// use std::net::Ipv6Addr; +/// +/// let localhost = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); +/// assert_eq!("::1".parse(), Ok(localhost)); +/// assert_eq!(localhost.is_loopback(), true); +/// ``` +#[derive(Copy, Clone, PartialEq, Eq, Hash)] +#[cfg_attr(not(test), rustc_diagnostic_item = "Ipv6Addr")] +#[stable(feature = "rust1", since = "1.0.0")] +pub struct Ipv6Addr { + octets: [u8; 16], +} + +/// Scope of an [IPv6 multicast address] as defined in [IETF RFC 7346 section 2]. +/// +/// # Stability Guarantees +/// +/// Not all possible values for a multicast scope have been assigned. +/// Future RFCs may introduce new scopes, which will be added as variants to this enum; +/// because of this the enum is marked as `#[non_exhaustive]`. +/// +/// # Examples +/// ``` +/// #![feature(ip)] +/// +/// use std::net::Ipv6Addr; +/// use std::net::Ipv6MulticastScope::*; +/// +/// // An IPv6 multicast address with global scope (`ff0e::`). +/// let address = Ipv6Addr::new(0xff0e, 0, 0, 0, 0, 0, 0, 0); +/// +/// // Will print "Global scope". +/// match address.multicast_scope() { +/// Some(InterfaceLocal) => println!("Interface-Local scope"), +/// Some(LinkLocal) => println!("Link-Local scope"), +/// Some(RealmLocal) => println!("Realm-Local scope"), +/// Some(AdminLocal) => println!("Admin-Local scope"), +/// Some(SiteLocal) => println!("Site-Local scope"), +/// Some(OrganizationLocal) => println!("Organization-Local scope"), +/// Some(Global) => println!("Global scope"), +/// Some(_) => println!("Unknown scope"), +/// None => println!("Not a multicast address!") +/// } +/// +/// ``` +/// +/// [IPv6 multicast address]: Ipv6Addr +/// [IETF RFC 7346 section 2]: https://tools.ietf.org/html/rfc7346#section-2 +#[derive(Copy, PartialEq, Eq, Clone, Hash, Debug)] +#[unstable(feature = "ip", issue = "27709")] +#[non_exhaustive] +pub enum Ipv6MulticastScope { + /// Interface-Local scope. + InterfaceLocal, + /// Link-Local scope. + LinkLocal, + /// Realm-Local scope. + RealmLocal, + /// Admin-Local scope. + AdminLocal, + /// Site-Local scope. + SiteLocal, + /// Organization-Local scope. + OrganizationLocal, + /// Global scope. + Global, +} + +impl IpAddr { + /// Returns [`true`] for the special 'unspecified' address. + /// + /// See the documentation for [`Ipv4Addr::is_unspecified()`] and + /// [`Ipv6Addr::is_unspecified()`] for more details. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(0, 0, 0, 0)).is_unspecified(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)).is_unspecified(), true); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "ip_shared", since = "1.12.0")] + #[must_use] + #[inline] + pub const fn is_unspecified(&self) -> bool { + match self { + IpAddr::V4(ip) => ip.is_unspecified(), + IpAddr::V6(ip) => ip.is_unspecified(), + } + } + + /// Returns [`true`] if this is a loopback address. + /// + /// See the documentation for [`Ipv4Addr::is_loopback()`] and + /// [`Ipv6Addr::is_loopback()`] for more details. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)).is_loopback(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0x1)).is_loopback(), true); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "ip_shared", since = "1.12.0")] + #[must_use] + #[inline] + pub const fn is_loopback(&self) -> bool { + match self { + IpAddr::V4(ip) => ip.is_loopback(), + IpAddr::V6(ip) => ip.is_loopback(), + } + } + + /// Returns [`true`] if the address appears to be globally routable. + /// + /// See the documentation for [`Ipv4Addr::is_global()`] and + /// [`Ipv6Addr::is_global()`] for more details. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(80, 9, 12, 3)).is_global(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0x1c9, 0, 0, 0xafc8, 0, 0x1)).is_global(), true); + /// ``` + #[rustc_const_unstable(feature = "const_ip", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_global(&self) -> bool { + match self { + IpAddr::V4(ip) => ip.is_global(), + IpAddr::V6(ip) => ip.is_global(), + } + } + + /// Returns [`true`] if this is a multicast address. + /// + /// See the documentation for [`Ipv4Addr::is_multicast()`] and + /// [`Ipv6Addr::is_multicast()`] for more details. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(224, 254, 0, 0)).is_multicast(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0)).is_multicast(), true); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "ip_shared", since = "1.12.0")] + #[must_use] + #[inline] + pub const fn is_multicast(&self) -> bool { + match self { + IpAddr::V4(ip) => ip.is_multicast(), + IpAddr::V6(ip) => ip.is_multicast(), + } + } + + /// Returns [`true`] if this address is in a range designated for documentation. + /// + /// See the documentation for [`Ipv4Addr::is_documentation()`] and + /// [`Ipv6Addr::is_documentation()`] for more details. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(203, 0, 113, 6)).is_documentation(), true); + /// assert_eq!( + /// IpAddr::V6(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0)).is_documentation(), + /// true + /// ); + /// ``` + #[rustc_const_unstable(feature = "const_ip", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_documentation(&self) -> bool { + match self { + IpAddr::V4(ip) => ip.is_documentation(), + IpAddr::V6(ip) => ip.is_documentation(), + } + } + + /// Returns [`true`] if this address is in a range designated for benchmarking. + /// + /// See the documentation for [`Ipv4Addr::is_benchmarking()`] and + /// [`Ipv6Addr::is_benchmarking()`] for more details. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(198, 19, 255, 255)).is_benchmarking(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0x2001, 0x2, 0, 0, 0, 0, 0, 0)).is_benchmarking(), true); + /// ``` + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_benchmarking(&self) -> bool { + match self { + IpAddr::V4(ip) => ip.is_benchmarking(), + IpAddr::V6(ip) => ip.is_benchmarking(), + } + } + + /// Returns [`true`] if this address is an [`IPv4` address], and [`false`] + /// otherwise. + /// + /// [`IPv4` address]: IpAddr::V4 + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(203, 0, 113, 6)).is_ipv4(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0)).is_ipv4(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "ipaddr_checker", since = "1.16.0")] + #[must_use] + #[inline] + pub const fn is_ipv4(&self) -> bool { + matches!(self, IpAddr::V4(_)) + } + + /// Returns [`true`] if this address is an [`IPv6` address], and [`false`] + /// otherwise. + /// + /// [`IPv6` address]: IpAddr::V6 + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(203, 0, 113, 6)).is_ipv6(), false); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0)).is_ipv6(), true); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "ipaddr_checker", since = "1.16.0")] + #[must_use] + #[inline] + pub const fn is_ipv6(&self) -> bool { + matches!(self, IpAddr::V6(_)) + } + + /// Converts this address to an `IpAddr::V4` if it is an IPv4-mapped IPv6 addresses, otherwise it + /// return `self` as-is. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)).to_canonical().is_loopback(), true); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1)).is_loopback(), false); + /// assert_eq!(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1)).to_canonical().is_loopback(), true); + /// ``` + #[inline] + #[must_use = "this returns the result of the operation, \ + without modifying the original"] + #[rustc_const_unstable(feature = "const_ip", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + pub const fn to_canonical(&self) -> IpAddr { + match self { + &v4 @ IpAddr::V4(_) => v4, + IpAddr::V6(v6) => v6.to_canonical(), + } + } +} + +impl Ipv4Addr { + /// Creates a new IPv4 address from four eight-bit octets. + /// + /// The result will represent the IP address `a`.`b`.`c`.`d`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::new(127, 0, 0, 1); + /// ``` + #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use] + #[inline] + pub const fn new(a: u8, b: u8, c: u8, d: u8) -> Ipv4Addr { + Ipv4Addr { octets: [a, b, c, d] } + } + + /// An IPv4 address with the address pointing to localhost: `127.0.0.1` + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::LOCALHOST; + /// assert_eq!(addr, Ipv4Addr::new(127, 0, 0, 1)); + /// ``` + #[stable(feature = "ip_constructors", since = "1.30.0")] + pub const LOCALHOST: Self = Ipv4Addr::new(127, 0, 0, 1); + + /// An IPv4 address representing an unspecified address: `0.0.0.0` + /// + /// This corresponds to the constant `INADDR_ANY` in other languages. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::UNSPECIFIED; + /// assert_eq!(addr, Ipv4Addr::new(0, 0, 0, 0)); + /// ``` + #[doc(alias = "INADDR_ANY")] + #[stable(feature = "ip_constructors", since = "1.30.0")] + pub const UNSPECIFIED: Self = Ipv4Addr::new(0, 0, 0, 0); + + /// An IPv4 address representing the broadcast address: `255.255.255.255` + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::BROADCAST; + /// assert_eq!(addr, Ipv4Addr::new(255, 255, 255, 255)); + /// ``` + #[stable(feature = "ip_constructors", since = "1.30.0")] + pub const BROADCAST: Self = Ipv4Addr::new(255, 255, 255, 255); + + /// Returns the four eight-bit integers that make up this address. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::new(127, 0, 0, 1); + /// assert_eq!(addr.octets(), [127, 0, 0, 1]); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use] + #[inline] + pub const fn octets(&self) -> [u8; 4] { + self.octets + } + + /// Returns [`true`] for the special 'unspecified' address (`0.0.0.0`). + /// + /// This property is defined in _UNIX Network Programming, Second Edition_, + /// W. Richard Stevens, p. 891; see also [ip7]. + /// + /// [ip7]: https://man7.org/linux/man-pages/man7/ip.7.html + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(0, 0, 0, 0).is_unspecified(), true); + /// assert_eq!(Ipv4Addr::new(45, 22, 13, 197).is_unspecified(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] + #[stable(feature = "ip_shared", since = "1.12.0")] + #[must_use] + #[inline] + pub const fn is_unspecified(&self) -> bool { + u32::from_be_bytes(self.octets) == 0 + } + + /// Returns [`true`] if this is a loopback address (`127.0.0.0/8`). + /// + /// This property is defined by [IETF RFC 1122]. + /// + /// [IETF RFC 1122]: https://tools.ietf.org/html/rfc1122 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(127, 0, 0, 1).is_loopback(), true); + /// assert_eq!(Ipv4Addr::new(45, 22, 13, 197).is_loopback(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_loopback(&self) -> bool { + self.octets()[0] == 127 + } + + /// Returns [`true`] if this is a private address. + /// + /// The private address ranges are defined in [IETF RFC 1918] and include: + /// + /// - `10.0.0.0/8` + /// - `172.16.0.0/12` + /// - `192.168.0.0/16` + /// + /// [IETF RFC 1918]: https://tools.ietf.org/html/rfc1918 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(10, 0, 0, 1).is_private(), true); + /// assert_eq!(Ipv4Addr::new(10, 10, 10, 10).is_private(), true); + /// assert_eq!(Ipv4Addr::new(172, 16, 10, 10).is_private(), true); + /// assert_eq!(Ipv4Addr::new(172, 29, 45, 14).is_private(), true); + /// assert_eq!(Ipv4Addr::new(172, 32, 0, 2).is_private(), false); + /// assert_eq!(Ipv4Addr::new(192, 168, 0, 2).is_private(), true); + /// assert_eq!(Ipv4Addr::new(192, 169, 0, 2).is_private(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_private(&self) -> bool { + match self.octets() { + [10, ..] => true, + [172, b, ..] if b >= 16 && b <= 31 => true, + [192, 168, ..] => true, + _ => false, + } + } + + /// Returns [`true`] if the address is link-local (`169.254.0.0/16`). + /// + /// This property is defined by [IETF RFC 3927]. + /// + /// [IETF RFC 3927]: https://tools.ietf.org/html/rfc3927 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(169, 254, 0, 0).is_link_local(), true); + /// assert_eq!(Ipv4Addr::new(169, 254, 10, 65).is_link_local(), true); + /// assert_eq!(Ipv4Addr::new(16, 89, 10, 65).is_link_local(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_link_local(&self) -> bool { + matches!(self.octets(), [169, 254, ..]) + } + + /// Returns [`true`] if the address appears to be globally reachable + /// as specified by the [IANA IPv4 Special-Purpose Address Registry]. + /// Whether or not an address is practically reachable will depend on your network configuration. + /// + /// Most IPv4 addresses are globally reachable; + /// unless they are specifically defined as *not* globally reachable. + /// + /// Non-exhaustive list of notable addresses that are not globally reachable: + /// + /// - The [unspecified address] ([`is_unspecified`](Ipv4Addr::is_unspecified)) + /// - Addresses reserved for private use ([`is_private`](Ipv4Addr::is_private)) + /// - Addresses in the shared address space ([`is_shared`](Ipv4Addr::is_shared)) + /// - Loopback addresses ([`is_loopback`](Ipv4Addr::is_loopback)) + /// - Link-local addresses ([`is_link_local`](Ipv4Addr::is_link_local)) + /// - Addresses reserved for documentation ([`is_documentation`](Ipv4Addr::is_documentation)) + /// - Addresses reserved for benchmarking ([`is_benchmarking`](Ipv4Addr::is_benchmarking)) + /// - Reserved addresses ([`is_reserved`](Ipv4Addr::is_reserved)) + /// - The [broadcast address] ([`is_broadcast`](Ipv4Addr::is_broadcast)) + /// + /// For the complete overview of which addresses are globally reachable, see the table at the [IANA IPv4 Special-Purpose Address Registry]. + /// + /// [IANA IPv4 Special-Purpose Address Registry]: https://www.iana.org/assignments/iana-ipv4-special-registry/iana-ipv4-special-registry.xhtml + /// [unspecified address]: Ipv4Addr::UNSPECIFIED + /// [broadcast address]: Ipv4Addr::BROADCAST + + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv4Addr; + /// + /// // Most IPv4 addresses are globally reachable: + /// assert_eq!(Ipv4Addr::new(80, 9, 12, 3).is_global(), true); + /// + /// // However some addresses have been assigned a special meaning + /// // that makes them not globally reachable. Some examples are: + /// + /// // The unspecified address (`0.0.0.0`) + /// assert_eq!(Ipv4Addr::UNSPECIFIED.is_global(), false); + /// + /// // Addresses reserved for private use (`10.0.0.0/8`, `172.16.0.0/12`, 192.168.0.0/16) + /// assert_eq!(Ipv4Addr::new(10, 254, 0, 0).is_global(), false); + /// assert_eq!(Ipv4Addr::new(192, 168, 10, 65).is_global(), false); + /// assert_eq!(Ipv4Addr::new(172, 16, 10, 65).is_global(), false); + /// + /// // Addresses in the shared address space (`100.64.0.0/10`) + /// assert_eq!(Ipv4Addr::new(100, 100, 0, 0).is_global(), false); + /// + /// // The loopback addresses (`127.0.0.0/8`) + /// assert_eq!(Ipv4Addr::LOCALHOST.is_global(), false); + /// + /// // Link-local addresses (`169.254.0.0/16`) + /// assert_eq!(Ipv4Addr::new(169, 254, 45, 1).is_global(), false); + /// + /// // Addresses reserved for documentation (`192.0.2.0/24`, `198.51.100.0/24`, `203.0.113.0/24`) + /// assert_eq!(Ipv4Addr::new(192, 0, 2, 255).is_global(), false); + /// assert_eq!(Ipv4Addr::new(198, 51, 100, 65).is_global(), false); + /// assert_eq!(Ipv4Addr::new(203, 0, 113, 6).is_global(), false); + /// + /// // Addresses reserved for benchmarking (`198.18.0.0/15`) + /// assert_eq!(Ipv4Addr::new(198, 18, 0, 0).is_global(), false); + /// + /// // Reserved addresses (`240.0.0.0/4`) + /// assert_eq!(Ipv4Addr::new(250, 10, 20, 30).is_global(), false); + /// + /// // The broadcast address (`255.255.255.255`) + /// assert_eq!(Ipv4Addr::BROADCAST.is_global(), false); + /// + /// // For a complete overview see the IANA IPv4 Special-Purpose Address Registry. + /// ``` + #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_global(&self) -> bool { + !(self.octets()[0] == 0 // "This network" + || self.is_private() + || self.is_shared() + || self.is_loopback() + || self.is_link_local() + // addresses reserved for future protocols (`192.0.0.0/24`) + ||(self.octets()[0] == 192 && self.octets()[1] == 0 && self.octets()[2] == 0) + || self.is_documentation() + || self.is_benchmarking() + || self.is_reserved() + || self.is_broadcast()) + } + + /// Returns [`true`] if this address is part of the Shared Address Space defined in + /// [IETF RFC 6598] (`100.64.0.0/10`). + /// + /// [IETF RFC 6598]: https://tools.ietf.org/html/rfc6598 + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(100, 64, 0, 0).is_shared(), true); + /// assert_eq!(Ipv4Addr::new(100, 127, 255, 255).is_shared(), true); + /// assert_eq!(Ipv4Addr::new(100, 128, 0, 0).is_shared(), false); + /// ``` + #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_shared(&self) -> bool { + self.octets()[0] == 100 && (self.octets()[1] & 0b1100_0000 == 0b0100_0000) + } + + /// Returns [`true`] if this address part of the `198.18.0.0/15` range, which is reserved for + /// network devices benchmarking. This range is defined in [IETF RFC 2544] as `192.18.0.0` + /// through `198.19.255.255` but [errata 423] corrects it to `198.18.0.0/15`. + /// + /// [IETF RFC 2544]: https://tools.ietf.org/html/rfc2544 + /// [errata 423]: https://www.rfc-editor.org/errata/eid423 + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(198, 17, 255, 255).is_benchmarking(), false); + /// assert_eq!(Ipv4Addr::new(198, 18, 0, 0).is_benchmarking(), true); + /// assert_eq!(Ipv4Addr::new(198, 19, 255, 255).is_benchmarking(), true); + /// assert_eq!(Ipv4Addr::new(198, 20, 0, 0).is_benchmarking(), false); + /// ``` + #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_benchmarking(&self) -> bool { + self.octets()[0] == 198 && (self.octets()[1] & 0xfe) == 18 + } + + /// Returns [`true`] if this address is reserved by IANA for future use. [IETF RFC 1112] + /// defines the block of reserved addresses as `240.0.0.0/4`. This range normally includes the + /// broadcast address `255.255.255.255`, but this implementation explicitly excludes it, since + /// it is obviously not reserved for future use. + /// + /// [IETF RFC 1112]: https://tools.ietf.org/html/rfc1112 + /// + /// # Warning + /// + /// As IANA assigns new addresses, this method will be + /// updated. This may result in non-reserved addresses being + /// treated as reserved in code that relies on an outdated version + /// of this method. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(240, 0, 0, 0).is_reserved(), true); + /// assert_eq!(Ipv4Addr::new(255, 255, 255, 254).is_reserved(), true); + /// + /// assert_eq!(Ipv4Addr::new(239, 255, 255, 255).is_reserved(), false); + /// // The broadcast address is not considered as reserved for future use by this implementation + /// assert_eq!(Ipv4Addr::new(255, 255, 255, 255).is_reserved(), false); + /// ``` + #[rustc_const_unstable(feature = "const_ipv4", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_reserved(&self) -> bool { + self.octets()[0] & 240 == 240 && !self.is_broadcast() + } + + /// Returns [`true`] if this is a multicast address (`224.0.0.0/4`). + /// + /// Multicast addresses have a most significant octet between `224` and `239`, + /// and is defined by [IETF RFC 5771]. + /// + /// [IETF RFC 5771]: https://tools.ietf.org/html/rfc5771 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(224, 254, 0, 0).is_multicast(), true); + /// assert_eq!(Ipv4Addr::new(236, 168, 10, 65).is_multicast(), true); + /// assert_eq!(Ipv4Addr::new(172, 16, 10, 65).is_multicast(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_multicast(&self) -> bool { + self.octets()[0] >= 224 && self.octets()[0] <= 239 + } + + /// Returns [`true`] if this is a broadcast address (`255.255.255.255`). + /// + /// A broadcast address has all octets set to `255` as defined in [IETF RFC 919]. + /// + /// [IETF RFC 919]: https://tools.ietf.org/html/rfc919 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(255, 255, 255, 255).is_broadcast(), true); + /// assert_eq!(Ipv4Addr::new(236, 168, 10, 65).is_broadcast(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_broadcast(&self) -> bool { + u32::from_be_bytes(self.octets()) == u32::from_be_bytes(Self::BROADCAST.octets()) + } + + /// Returns [`true`] if this address is in a range designated for documentation. + /// + /// This is defined in [IETF RFC 5737]: + /// + /// - `192.0.2.0/24` (TEST-NET-1) + /// - `198.51.100.0/24` (TEST-NET-2) + /// - `203.0.113.0/24` (TEST-NET-3) + /// + /// [IETF RFC 5737]: https://tools.ietf.org/html/rfc5737 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// assert_eq!(Ipv4Addr::new(192, 0, 2, 255).is_documentation(), true); + /// assert_eq!(Ipv4Addr::new(198, 51, 100, 65).is_documentation(), true); + /// assert_eq!(Ipv4Addr::new(203, 0, 113, 6).is_documentation(), true); + /// assert_eq!(Ipv4Addr::new(193, 34, 17, 19).is_documentation(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_documentation(&self) -> bool { + matches!(self.octets(), [192, 0, 2, _] | [198, 51, 100, _] | [203, 0, 113, _]) + } + + /// Converts this address to an [IPv4-compatible] [`IPv6` address]. + /// + /// `a.b.c.d` becomes `::a.b.c.d` + /// + /// Note that IPv4-compatible addresses have been officially deprecated. + /// If you don't explicitly need an IPv4-compatible address for legacy reasons, consider using `to_ipv6_mapped` instead. + /// + /// [IPv4-compatible]: Ipv6Addr#ipv4-compatible-ipv6-addresses + /// [`IPv6` address]: Ipv6Addr + /// + /// # Examples + /// + /// ``` + /// use std::net::{Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!( + /// Ipv4Addr::new(192, 0, 2, 255).to_ipv6_compatible(), + /// Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0xc000, 0x2ff) + /// ); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use = "this returns the result of the operation, \ + without modifying the original"] + #[inline] + pub const fn to_ipv6_compatible(&self) -> Ipv6Addr { + let [a, b, c, d] = self.octets(); + Ipv6Addr { octets: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, a, b, c, d] } + } + + /// Converts this address to an [IPv4-mapped] [`IPv6` address]. + /// + /// `a.b.c.d` becomes `::ffff:a.b.c.d` + /// + /// [IPv4-mapped]: Ipv6Addr#ipv4-mapped-ipv6-addresses + /// [`IPv6` address]: Ipv6Addr + /// + /// # Examples + /// + /// ``` + /// use std::net::{Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(Ipv4Addr::new(192, 0, 2, 255).to_ipv6_mapped(), + /// Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc000, 0x2ff)); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use = "this returns the result of the operation, \ + without modifying the original"] + #[inline] + pub const fn to_ipv6_mapped(&self) -> Ipv6Addr { + let [a, b, c, d] = self.octets(); + Ipv6Addr { octets: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xFF, 0xFF, a, b, c, d] } + } +} + +#[stable(feature = "ip_addr", since = "1.7.0")] +impl fmt::Display for IpAddr { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + match self { + IpAddr::V4(ip) => ip.fmt(fmt), + IpAddr::V6(ip) => ip.fmt(fmt), + } + } +} + +#[stable(feature = "ip_addr", since = "1.7.0")] +impl fmt::Debug for IpAddr { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self, fmt) + } +} + +#[stable(feature = "ip_from_ip", since = "1.16.0")] +impl From<Ipv4Addr> for IpAddr { + /// Copies this address to a new `IpAddr::V4`. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr}; + /// + /// let addr = Ipv4Addr::new(127, 0, 0, 1); + /// + /// assert_eq!( + /// IpAddr::V4(addr), + /// IpAddr::from(addr) + /// ) + /// ``` + #[inline] + fn from(ipv4: Ipv4Addr) -> IpAddr { + IpAddr::V4(ipv4) + } +} + +#[stable(feature = "ip_from_ip", since = "1.16.0")] +impl From<Ipv6Addr> for IpAddr { + /// Copies this address to a new `IpAddr::V6`. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv6Addr}; + /// + /// let addr = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff); + /// + /// assert_eq!( + /// IpAddr::V6(addr), + /// IpAddr::from(addr) + /// ); + /// ``` + #[inline] + fn from(ipv6: Ipv6Addr) -> IpAddr { + IpAddr::V6(ipv6) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Display for Ipv4Addr { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + let octets = self.octets(); + + // If there are no alignment requirements, write the IP address directly to `f`. + // Otherwise, write it to a local buffer and then use `f.pad`. + if fmt.precision().is_none() && fmt.width().is_none() { + write!(fmt, "{}.{}.{}.{}", octets[0], octets[1], octets[2], octets[3]) + } else { + const LONGEST_IPV4_ADDR: &str = "255.255.255.255"; + + let mut buf = DisplayBuffer::<{ LONGEST_IPV4_ADDR.len() }>::new(); + // Buffer is long enough for the longest possible IPv4 address, so this should never fail. + write!(buf, "{}.{}.{}.{}", octets[0], octets[1], octets[2], octets[3]).unwrap(); + + fmt.pad(buf.as_str()) + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Debug for Ipv4Addr { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self, fmt) + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialEq<Ipv4Addr> for IpAddr { + #[inline] + fn eq(&self, other: &Ipv4Addr) -> bool { + match self { + IpAddr::V4(v4) => v4 == other, + IpAddr::V6(_) => false, + } + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialEq<IpAddr> for Ipv4Addr { + #[inline] + fn eq(&self, other: &IpAddr) -> bool { + match other { + IpAddr::V4(v4) => self == v4, + IpAddr::V6(_) => false, + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl PartialOrd for Ipv4Addr { + #[inline] + fn partial_cmp(&self, other: &Ipv4Addr) -> Option<Ordering> { + Some(self.cmp(other)) + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialOrd<Ipv4Addr> for IpAddr { + #[inline] + fn partial_cmp(&self, other: &Ipv4Addr) -> Option<Ordering> { + match self { + IpAddr::V4(v4) => v4.partial_cmp(other), + IpAddr::V6(_) => Some(Ordering::Greater), + } + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialOrd<IpAddr> for Ipv4Addr { + #[inline] + fn partial_cmp(&self, other: &IpAddr) -> Option<Ordering> { + match other { + IpAddr::V4(v4) => self.partial_cmp(v4), + IpAddr::V6(_) => Some(Ordering::Less), + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl Ord for Ipv4Addr { + #[inline] + fn cmp(&self, other: &Ipv4Addr) -> Ordering { + self.octets.cmp(&other.octets) + } +} + +impl IntoInner<c::in_addr> for Ipv4Addr { + #[inline] + fn into_inner(self) -> c::in_addr { + // `s_addr` is stored as BE on all machines and the array is in BE order. + // So the native endian conversion method is used so that it's never swapped. + c::in_addr { s_addr: u32::from_ne_bytes(self.octets) } + } +} +impl FromInner<c::in_addr> for Ipv4Addr { + fn from_inner(addr: c::in_addr) -> Ipv4Addr { + Ipv4Addr { octets: addr.s_addr.to_ne_bytes() } + } +} + +#[stable(feature = "ip_u32", since = "1.1.0")] +impl From<Ipv4Addr> for u32 { + /// Converts an `Ipv4Addr` into a host byte order `u32`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::new(0x12, 0x34, 0x56, 0x78); + /// assert_eq!(0x12345678, u32::from(addr)); + /// ``` + #[inline] + fn from(ip: Ipv4Addr) -> u32 { + u32::from_be_bytes(ip.octets) + } +} + +#[stable(feature = "ip_u32", since = "1.1.0")] +impl From<u32> for Ipv4Addr { + /// Converts a host byte order `u32` into an `Ipv4Addr`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::from(0x12345678); + /// assert_eq!(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78), addr); + /// ``` + #[inline] + fn from(ip: u32) -> Ipv4Addr { + Ipv4Addr { octets: ip.to_be_bytes() } + } +} + +#[stable(feature = "from_slice_v4", since = "1.9.0")] +impl From<[u8; 4]> for Ipv4Addr { + /// Creates an `Ipv4Addr` from a four element byte array. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv4Addr; + /// + /// let addr = Ipv4Addr::from([13u8, 12u8, 11u8, 10u8]); + /// assert_eq!(Ipv4Addr::new(13, 12, 11, 10), addr); + /// ``` + #[inline] + fn from(octets: [u8; 4]) -> Ipv4Addr { + Ipv4Addr { octets } + } +} + +#[stable(feature = "ip_from_slice", since = "1.17.0")] +impl From<[u8; 4]> for IpAddr { + /// Creates an `IpAddr::V4` from a four element byte array. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr}; + /// + /// let addr = IpAddr::from([13u8, 12u8, 11u8, 10u8]); + /// assert_eq!(IpAddr::V4(Ipv4Addr::new(13, 12, 11, 10)), addr); + /// ``` + #[inline] + fn from(octets: [u8; 4]) -> IpAddr { + IpAddr::V4(Ipv4Addr::from(octets)) + } +} + +impl Ipv6Addr { + /// Creates a new IPv6 address from eight 16-bit segments. + /// + /// The result will represent the IP address `a:b:c:d:e:f:g:h`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff); + /// ``` + #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use] + #[inline] + pub const fn new(a: u16, b: u16, c: u16, d: u16, e: u16, f: u16, g: u16, h: u16) -> Ipv6Addr { + let addr16 = [ + a.to_be(), + b.to_be(), + c.to_be(), + d.to_be(), + e.to_be(), + f.to_be(), + g.to_be(), + h.to_be(), + ]; + Ipv6Addr { + // All elements in `addr16` are big endian. + // SAFETY: `[u16; 8]` is always safe to transmute to `[u8; 16]`. + octets: unsafe { transmute::<_, [u8; 16]>(addr16) }, + } + } + + /// An IPv6 address representing localhost: `::1`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::LOCALHOST; + /// assert_eq!(addr, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); + /// ``` + #[stable(feature = "ip_constructors", since = "1.30.0")] + pub const LOCALHOST: Self = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); + + /// An IPv6 address representing the unspecified address: `::` + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::UNSPECIFIED; + /// assert_eq!(addr, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)); + /// ``` + #[stable(feature = "ip_constructors", since = "1.30.0")] + pub const UNSPECIFIED: Self = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0); + + /// Returns the eight 16-bit segments that make up this address. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).segments(), + /// [0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff]); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use] + #[inline] + pub const fn segments(&self) -> [u16; 8] { + // All elements in `self.octets` must be big endian. + // SAFETY: `[u8; 16]` is always safe to transmute to `[u16; 8]`. + let [a, b, c, d, e, f, g, h] = unsafe { transmute::<_, [u16; 8]>(self.octets) }; + // We want native endian u16 + [ + u16::from_be(a), + u16::from_be(b), + u16::from_be(c), + u16::from_be(d), + u16::from_be(e), + u16::from_be(f), + u16::from_be(g), + u16::from_be(h), + ] + } + + /// Returns [`true`] for the special 'unspecified' address (`::`). + /// + /// This property is defined in [IETF RFC 4291]. + /// + /// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_unspecified(), false); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0).is_unspecified(), true); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_unspecified(&self) -> bool { + u128::from_be_bytes(self.octets()) == u128::from_be_bytes(Ipv6Addr::UNSPECIFIED.octets()) + } + + /// Returns [`true`] if this is the [loopback address] (`::1`), + /// as defined in [IETF RFC 4291 section 2.5.3]. + /// + /// Contrary to IPv4, in IPv6 there is only one loopback address. + /// + /// [loopback address]: Ipv6Addr::LOCALHOST + /// [IETF RFC 4291 section 2.5.3]: https://tools.ietf.org/html/rfc4291#section-2.5.3 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_loopback(), false); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0x1).is_loopback(), true); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_loopback(&self) -> bool { + u128::from_be_bytes(self.octets()) == u128::from_be_bytes(Ipv6Addr::LOCALHOST.octets()) + } + + /// Returns [`true`] if the address appears to be globally reachable + /// as specified by the [IANA IPv6 Special-Purpose Address Registry]. + /// Whether or not an address is practically reachable will depend on your network configuration. + /// + /// Most IPv6 addresses are globally reachable; + /// unless they are specifically defined as *not* globally reachable. + /// + /// Non-exhaustive list of notable addresses that are not globally reachable: + /// - The [unspecified address] ([`is_unspecified`](Ipv6Addr::is_unspecified)) + /// - The [loopback address] ([`is_loopback`](Ipv6Addr::is_loopback)) + /// - IPv4-mapped addresses + /// - Addresses reserved for benchmarking + /// - Addresses reserved for documentation ([`is_documentation`](Ipv6Addr::is_documentation)) + /// - Unique local addresses ([`is_unique_local`](Ipv6Addr::is_unique_local)) + /// - Unicast addresses with link-local scope ([`is_unicast_link_local`](Ipv6Addr::is_unicast_link_local)) + /// + /// For the complete overview of which addresses are globally reachable, see the table at the [IANA IPv6 Special-Purpose Address Registry]. + /// + /// Note that an address having global scope is not the same as being globally reachable, + /// and there is no direct relation between the two concepts: There exist addresses with global scope + /// that are not globally reachable (for example unique local addresses), + /// and addresses that are globally reachable without having global scope + /// (multicast addresses with non-global scope). + /// + /// [IANA IPv6 Special-Purpose Address Registry]: https://www.iana.org/assignments/iana-ipv6-special-registry/iana-ipv6-special-registry.xhtml + /// [unspecified address]: Ipv6Addr::UNSPECIFIED + /// [loopback address]: Ipv6Addr::LOCALHOST + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// // Most IPv6 addresses are globally reachable: + /// assert_eq!(Ipv6Addr::new(0x26, 0, 0x1c9, 0, 0, 0xafc8, 0x10, 0x1).is_global(), true); + /// + /// // However some addresses have been assigned a special meaning + /// // that makes them not globally reachable. Some examples are: + /// + /// // The unspecified address (`::`) + /// assert_eq!(Ipv6Addr::UNSPECIFIED.is_global(), false); + /// + /// // The loopback address (`::1`) + /// assert_eq!(Ipv6Addr::LOCALHOST.is_global(), false); + /// + /// // IPv4-mapped addresses (`::ffff:0:0/96`) + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_global(), false); + /// + /// // Addresses reserved for benchmarking (`2001:2::/48`) + /// assert_eq!(Ipv6Addr::new(0x2001, 2, 0, 0, 0, 0, 0, 1,).is_global(), false); + /// + /// // Addresses reserved for documentation (`2001:db8::/32`) + /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 1).is_global(), false); + /// + /// // Unique local addresses (`fc00::/7`) + /// assert_eq!(Ipv6Addr::new(0xfc02, 0, 0, 0, 0, 0, 0, 1).is_global(), false); + /// + /// // Unicast addresses with link-local scope (`fe80::/10`) + /// assert_eq!(Ipv6Addr::new(0xfe81, 0, 0, 0, 0, 0, 0, 1).is_global(), false); + /// + /// // For a complete overview see the IANA IPv6 Special-Purpose Address Registry. + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_global(&self) -> bool { + !(self.is_unspecified() + || self.is_loopback() + // IPv4-mapped Address (`::ffff:0:0/96`) + || matches!(self.segments(), [0, 0, 0, 0, 0, 0xffff, _, _]) + // IPv4-IPv6 Translat. (`64:ff9b:1::/48`) + || matches!(self.segments(), [0x64, 0xff9b, 1, _, _, _, _, _]) + // Discard-Only Address Block (`100::/64`) + || matches!(self.segments(), [0x100, 0, 0, 0, _, _, _, _]) + // IETF Protocol Assignments (`2001::/23`) + || (matches!(self.segments(), [0x2001, b, _, _, _, _, _, _] if b < 0x200) + && !( + // Port Control Protocol Anycast (`2001:1::1`) + u128::from_be_bytes(self.octets()) == 0x2001_0001_0000_0000_0000_0000_0000_0001 + // Traversal Using Relays around NAT Anycast (`2001:1::2`) + || u128::from_be_bytes(self.octets()) == 0x2001_0001_0000_0000_0000_0000_0000_0002 + // AMT (`2001:3::/32`) + || matches!(self.segments(), [0x2001, 3, _, _, _, _, _, _]) + // AS112-v6 (`2001:4:112::/48`) + || matches!(self.segments(), [0x2001, 4, 0x112, _, _, _, _, _]) + // ORCHIDv2 (`2001:20::/28`) + || matches!(self.segments(), [0x2001, b, _, _, _, _, _, _] if b >= 0x20 && b <= 0x2F) + )) + || self.is_documentation() + || self.is_unique_local() + || self.is_unicast_link_local()) + } + + /// Returns [`true`] if this is a unique local address (`fc00::/7`). + /// + /// This property is defined in [IETF RFC 4193]. + /// + /// [IETF RFC 4193]: https://tools.ietf.org/html/rfc4193 + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_unique_local(), false); + /// assert_eq!(Ipv6Addr::new(0xfc02, 0, 0, 0, 0, 0, 0, 0).is_unique_local(), true); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_unique_local(&self) -> bool { + (self.segments()[0] & 0xfe00) == 0xfc00 + } + + /// Returns [`true`] if this is a unicast address, as defined by [IETF RFC 4291]. + /// Any address that is not a [multicast address] (`ff00::/8`) is unicast. + /// + /// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 + /// [multicast address]: Ipv6Addr::is_multicast + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// // The unspecified and loopback addresses are unicast. + /// assert_eq!(Ipv6Addr::UNSPECIFIED.is_unicast(), true); + /// assert_eq!(Ipv6Addr::LOCALHOST.is_unicast(), true); + /// + /// // Any address that is not a multicast address (`ff00::/8`) is unicast. + /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_unicast(), true); + /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).is_unicast(), false); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_unicast(&self) -> bool { + !self.is_multicast() + } + + /// Returns `true` if the address is a unicast address with link-local scope, + /// as defined in [RFC 4291]. + /// + /// A unicast address has link-local scope if it has the prefix `fe80::/10`, as per [RFC 4291 section 2.4]. + /// Note that this encompasses more addresses than those defined in [RFC 4291 section 2.5.6], + /// which describes "Link-Local IPv6 Unicast Addresses" as having the following stricter format: + /// + /// ```text + /// | 10 bits | 54 bits | 64 bits | + /// +----------+-------------------------+----------------------------+ + /// |1111111010| 0 | interface ID | + /// +----------+-------------------------+----------------------------+ + /// ``` + /// So while currently the only addresses with link-local scope an application will encounter are all in `fe80::/64`, + /// this might change in the future with the publication of new standards. More addresses in `fe80::/10` could be allocated, + /// and those addresses will have link-local scope. + /// + /// Also note that while [RFC 4291 section 2.5.3] mentions about the [loopback address] (`::1`) that "it is treated as having Link-Local scope", + /// this does not mean that the loopback address actually has link-local scope and this method will return `false` on it. + /// + /// [RFC 4291]: https://tools.ietf.org/html/rfc4291 + /// [RFC 4291 section 2.4]: https://tools.ietf.org/html/rfc4291#section-2.4 + /// [RFC 4291 section 2.5.3]: https://tools.ietf.org/html/rfc4291#section-2.5.3 + /// [RFC 4291 section 2.5.6]: https://tools.ietf.org/html/rfc4291#section-2.5.6 + /// [loopback address]: Ipv6Addr::LOCALHOST + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// // The loopback address (`::1`) does not actually have link-local scope. + /// assert_eq!(Ipv6Addr::LOCALHOST.is_unicast_link_local(), false); + /// + /// // Only addresses in `fe80::/10` have link-local scope. + /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_unicast_link_local(), false); + /// assert_eq!(Ipv6Addr::new(0xfe80, 0, 0, 0, 0, 0, 0, 0).is_unicast_link_local(), true); + /// + /// // Addresses outside the stricter `fe80::/64` also have link-local scope. + /// assert_eq!(Ipv6Addr::new(0xfe80, 0, 0, 1, 0, 0, 0, 0).is_unicast_link_local(), true); + /// assert_eq!(Ipv6Addr::new(0xfe81, 0, 0, 0, 0, 0, 0, 0).is_unicast_link_local(), true); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_unicast_link_local(&self) -> bool { + (self.segments()[0] & 0xffc0) == 0xfe80 + } + + /// Returns [`true`] if this is an address reserved for documentation + /// (`2001:db8::/32`). + /// + /// This property is defined in [IETF RFC 3849]. + /// + /// [IETF RFC 3849]: https://tools.ietf.org/html/rfc3849 + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_documentation(), false); + /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_documentation(), true); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_documentation(&self) -> bool { + (self.segments()[0] == 0x2001) && (self.segments()[1] == 0xdb8) + } + + /// Returns [`true`] if this is an address reserved for benchmarking (`2001:2::/48`). + /// + /// This property is defined in [IETF RFC 5180], where it is mistakenly specified as covering the range `2001:0200::/48`. + /// This is corrected in [IETF RFC Errata 1752] to `2001:0002::/48`. + /// + /// [IETF RFC 5180]: https://tools.ietf.org/html/rfc5180 + /// [IETF RFC Errata 1752]: https://www.rfc-editor.org/errata_search.php?eid=1752 + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc613, 0x0).is_benchmarking(), false); + /// assert_eq!(Ipv6Addr::new(0x2001, 0x2, 0, 0, 0, 0, 0, 0).is_benchmarking(), true); + /// ``` + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_benchmarking(&self) -> bool { + (self.segments()[0] == 0x2001) && (self.segments()[1] == 0x2) && (self.segments()[2] == 0) + } + + /// Returns [`true`] if the address is a globally routable unicast address. + /// + /// The following return false: + /// + /// - the loopback address + /// - the link-local addresses + /// - unique local addresses + /// - the unspecified address + /// - the address range reserved for documentation + /// + /// This method returns [`true`] for site-local addresses as per [RFC 4291 section 2.5.7] + /// + /// ```no_rust + /// The special behavior of [the site-local unicast] prefix defined in [RFC3513] must no longer + /// be supported in new implementations (i.e., new implementations must treat this prefix as + /// Global Unicast). + /// ``` + /// + /// [RFC 4291 section 2.5.7]: https://tools.ietf.org/html/rfc4291#section-2.5.7 + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 0).is_unicast_global(), false); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_unicast_global(), true); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn is_unicast_global(&self) -> bool { + self.is_unicast() + && !self.is_loopback() + && !self.is_unicast_link_local() + && !self.is_unique_local() + && !self.is_unspecified() + && !self.is_documentation() + && !self.is_benchmarking() + } + + /// Returns the address's multicast scope if the address is multicast. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// + /// use std::net::{Ipv6Addr, Ipv6MulticastScope}; + /// + /// assert_eq!( + /// Ipv6Addr::new(0xff0e, 0, 0, 0, 0, 0, 0, 0).multicast_scope(), + /// Some(Ipv6MulticastScope::Global) + /// ); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).multicast_scope(), None); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use] + #[inline] + pub const fn multicast_scope(&self) -> Option<Ipv6MulticastScope> { + if self.is_multicast() { + match self.segments()[0] & 0x000f { + 1 => Some(Ipv6MulticastScope::InterfaceLocal), + 2 => Some(Ipv6MulticastScope::LinkLocal), + 3 => Some(Ipv6MulticastScope::RealmLocal), + 4 => Some(Ipv6MulticastScope::AdminLocal), + 5 => Some(Ipv6MulticastScope::SiteLocal), + 8 => Some(Ipv6MulticastScope::OrganizationLocal), + 14 => Some(Ipv6MulticastScope::Global), + _ => None, + } + } else { + None + } + } + + /// Returns [`true`] if this is a multicast address (`ff00::/8`). + /// + /// This property is defined by [IETF RFC 4291]. + /// + /// [IETF RFC 4291]: https://tools.ietf.org/html/rfc4291 + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).is_multicast(), true); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).is_multicast(), false); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(since = "1.7.0", feature = "ip_17")] + #[must_use] + #[inline] + pub const fn is_multicast(&self) -> bool { + (self.segments()[0] & 0xff00) == 0xff00 + } + + /// Converts this address to an [`IPv4` address] if it's an [IPv4-mapped] address, + /// as defined in [IETF RFC 4291 section 2.5.5.2], otherwise returns [`None`]. + /// + /// `::ffff:a.b.c.d` becomes `a.b.c.d`. + /// All addresses *not* starting with `::ffff` will return `None`. + /// + /// [`IPv4` address]: Ipv4Addr + /// [IPv4-mapped]: Ipv6Addr + /// [IETF RFC 4291 section 2.5.5.2]: https://tools.ietf.org/html/rfc4291#section-2.5.5.2 + /// + /// # Examples + /// + /// ``` + /// use std::net::{Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).to_ipv4_mapped(), None); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).to_ipv4_mapped(), + /// Some(Ipv4Addr::new(192, 10, 2, 255))); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1).to_ipv4_mapped(), None); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[stable(feature = "ipv6_to_ipv4_mapped", since = "1.63.0")] + #[must_use = "this returns the result of the operation, \ + without modifying the original"] + #[inline] + pub const fn to_ipv4_mapped(&self) -> Option<Ipv4Addr> { + match self.octets() { + [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, a, b, c, d] => { + Some(Ipv4Addr::new(a, b, c, d)) + } + _ => None, + } + } + + /// Converts this address to an [`IPv4` address] if it is either + /// an [IPv4-compatible] address as defined in [IETF RFC 4291 section 2.5.5.1], + /// or an [IPv4-mapped] address as defined in [IETF RFC 4291 section 2.5.5.2], + /// otherwise returns [`None`]. + /// + /// Note that this will return an [`IPv4` address] for the IPv6 loopback address `::1`. Use + /// [`Ipv6Addr::to_ipv4_mapped`] to avoid this. + /// + /// `::a.b.c.d` and `::ffff:a.b.c.d` become `a.b.c.d`. `::1` becomes `0.0.0.1`. + /// All addresses *not* starting with either all zeroes or `::ffff` will return `None`. + /// + /// [`IPv4` address]: Ipv4Addr + /// [IPv4-compatible]: Ipv6Addr#ipv4-compatible-ipv6-addresses + /// [IPv4-mapped]: Ipv6Addr#ipv4-mapped-ipv6-addresses + /// [IETF RFC 4291 section 2.5.5.1]: https://tools.ietf.org/html/rfc4291#section-2.5.5.1 + /// [IETF RFC 4291 section 2.5.5.2]: https://tools.ietf.org/html/rfc4291#section-2.5.5.2 + /// + /// # Examples + /// + /// ``` + /// use std::net::{Ipv4Addr, Ipv6Addr}; + /// + /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).to_ipv4(), None); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc00a, 0x2ff).to_ipv4(), + /// Some(Ipv4Addr::new(192, 10, 2, 255))); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1).to_ipv4(), + /// Some(Ipv4Addr::new(0, 0, 0, 1))); + /// ``` + #[rustc_const_stable(feature = "const_ip_50", since = "1.50.0")] + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use = "this returns the result of the operation, \ + without modifying the original"] + #[inline] + pub const fn to_ipv4(&self) -> Option<Ipv4Addr> { + if let [0, 0, 0, 0, 0, 0 | 0xffff, ab, cd] = self.segments() { + let [a, b] = ab.to_be_bytes(); + let [c, d] = cd.to_be_bytes(); + Some(Ipv4Addr::new(a, b, c, d)) + } else { + None + } + } + + /// Converts this address to an `IpAddr::V4` if it is an IPv4-mapped addresses, otherwise it + /// returns self wrapped in an `IpAddr::V6`. + /// + /// # Examples + /// + /// ``` + /// #![feature(ip)] + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1).is_loopback(), false); + /// assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x7f00, 0x1).to_canonical().is_loopback(), true); + /// ``` + #[rustc_const_unstable(feature = "const_ipv6", issue = "76205")] + #[unstable(feature = "ip", issue = "27709")] + #[must_use = "this returns the result of the operation, \ + without modifying the original"] + #[inline] + pub const fn to_canonical(&self) -> IpAddr { + if let Some(mapped) = self.to_ipv4_mapped() { + return IpAddr::V4(mapped); + } + IpAddr::V6(*self) + } + + /// Returns the sixteen eight-bit integers the IPv6 address consists of. + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// assert_eq!(Ipv6Addr::new(0xff00, 0, 0, 0, 0, 0, 0, 0).octets(), + /// [255, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]); + /// ``` + #[rustc_const_stable(feature = "const_ip_32", since = "1.32.0")] + #[stable(feature = "ipv6_to_octets", since = "1.12.0")] + #[must_use] + #[inline] + pub const fn octets(&self) -> [u8; 16] { + self.octets + } +} + +/// Write an Ipv6Addr, conforming to the canonical style described by +/// [RFC 5952](https://tools.ietf.org/html/rfc5952). +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Display for Ipv6Addr { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + // If there are no alignment requirements, write the IP address directly to `f`. + // Otherwise, write it to a local buffer and then use `f.pad`. + if f.precision().is_none() && f.width().is_none() { + let segments = self.segments(); + + // Special case for :: and ::1; otherwise they get written with the + // IPv4 formatter + if self.is_unspecified() { + f.write_str("::") + } else if self.is_loopback() { + f.write_str("::1") + } else if let Some(ipv4) = self.to_ipv4() { + match segments[5] { + // IPv4 Compatible address + 0 => write!(f, "::{}", ipv4), + // IPv4 Mapped address + 0xffff => write!(f, "::ffff:{}", ipv4), + _ => unreachable!(), + } + } else { + #[derive(Copy, Clone, Default)] + struct Span { + start: usize, + len: usize, + } + + // Find the inner 0 span + let zeroes = { + let mut longest = Span::default(); + let mut current = Span::default(); + + for (i, &segment) in segments.iter().enumerate() { + if segment == 0 { + if current.len == 0 { + current.start = i; + } + + current.len += 1; + + if current.len > longest.len { + longest = current; + } + } else { + current = Span::default(); + } + } + + longest + }; + + /// Write a colon-separated part of the address + #[inline] + fn fmt_subslice(f: &mut fmt::Formatter<'_>, chunk: &[u16]) -> fmt::Result { + if let Some((first, tail)) = chunk.split_first() { + write!(f, "{:x}", first)?; + for segment in tail { + f.write_char(':')?; + write!(f, "{:x}", segment)?; + } + } + Ok(()) + } + + if zeroes.len > 1 { + fmt_subslice(f, &segments[..zeroes.start])?; + f.write_str("::")?; + fmt_subslice(f, &segments[zeroes.start + zeroes.len..]) + } else { + fmt_subslice(f, &segments) + } + } + } else { + const LONGEST_IPV6_ADDR: &str = "ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff"; + + let mut buf = DisplayBuffer::<{ LONGEST_IPV6_ADDR.len() }>::new(); + // Buffer is long enough for the longest possible IPv6 address, so this should never fail. + write!(buf, "{}", self).unwrap(); + + f.pad(buf.as_str()) + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Debug for Ipv6Addr { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self, fmt) + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialEq<IpAddr> for Ipv6Addr { + #[inline] + fn eq(&self, other: &IpAddr) -> bool { + match other { + IpAddr::V4(_) => false, + IpAddr::V6(v6) => self == v6, + } + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialEq<Ipv6Addr> for IpAddr { + #[inline] + fn eq(&self, other: &Ipv6Addr) -> bool { + match self { + IpAddr::V4(_) => false, + IpAddr::V6(v6) => v6 == other, + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl PartialOrd for Ipv6Addr { + #[inline] + fn partial_cmp(&self, other: &Ipv6Addr) -> Option<Ordering> { + Some(self.cmp(other)) + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialOrd<Ipv6Addr> for IpAddr { + #[inline] + fn partial_cmp(&self, other: &Ipv6Addr) -> Option<Ordering> { + match self { + IpAddr::V4(_) => Some(Ordering::Less), + IpAddr::V6(v6) => v6.partial_cmp(other), + } + } +} + +#[stable(feature = "ip_cmp", since = "1.16.0")] +impl PartialOrd<IpAddr> for Ipv6Addr { + #[inline] + fn partial_cmp(&self, other: &IpAddr) -> Option<Ordering> { + match other { + IpAddr::V4(_) => Some(Ordering::Greater), + IpAddr::V6(v6) => self.partial_cmp(v6), + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl Ord for Ipv6Addr { + #[inline] + fn cmp(&self, other: &Ipv6Addr) -> Ordering { + self.segments().cmp(&other.segments()) + } +} + +impl IntoInner<c::in6_addr> for Ipv6Addr { + fn into_inner(self) -> c::in6_addr { + c::in6_addr { s6_addr: self.octets } + } +} +impl FromInner<c::in6_addr> for Ipv6Addr { + #[inline] + fn from_inner(addr: c::in6_addr) -> Ipv6Addr { + Ipv6Addr { octets: addr.s6_addr } + } +} + +#[stable(feature = "i128", since = "1.26.0")] +impl From<Ipv6Addr> for u128 { + /// Convert an `Ipv6Addr` into a host byte order `u128`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::new( + /// 0x1020, 0x3040, 0x5060, 0x7080, + /// 0x90A0, 0xB0C0, 0xD0E0, 0xF00D, + /// ); + /// assert_eq!(0x102030405060708090A0B0C0D0E0F00D_u128, u128::from(addr)); + /// ``` + #[inline] + fn from(ip: Ipv6Addr) -> u128 { + u128::from_be_bytes(ip.octets) + } +} +#[stable(feature = "i128", since = "1.26.0")] +impl From<u128> for Ipv6Addr { + /// Convert a host byte order `u128` into an `Ipv6Addr`. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::from(0x102030405060708090A0B0C0D0E0F00D_u128); + /// assert_eq!( + /// Ipv6Addr::new( + /// 0x1020, 0x3040, 0x5060, 0x7080, + /// 0x90A0, 0xB0C0, 0xD0E0, 0xF00D, + /// ), + /// addr); + /// ``` + #[inline] + fn from(ip: u128) -> Ipv6Addr { + Ipv6Addr::from(ip.to_be_bytes()) + } +} + +#[stable(feature = "ipv6_from_octets", since = "1.9.0")] +impl From<[u8; 16]> for Ipv6Addr { + /// Creates an `Ipv6Addr` from a sixteen element byte array. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::from([ + /// 25u8, 24u8, 23u8, 22u8, 21u8, 20u8, 19u8, 18u8, + /// 17u8, 16u8, 15u8, 14u8, 13u8, 12u8, 11u8, 10u8, + /// ]); + /// assert_eq!( + /// Ipv6Addr::new( + /// 0x1918, 0x1716, + /// 0x1514, 0x1312, + /// 0x1110, 0x0f0e, + /// 0x0d0c, 0x0b0a + /// ), + /// addr + /// ); + /// ``` + #[inline] + fn from(octets: [u8; 16]) -> Ipv6Addr { + Ipv6Addr { octets } + } +} + +#[stable(feature = "ipv6_from_segments", since = "1.16.0")] +impl From<[u16; 8]> for Ipv6Addr { + /// Creates an `Ipv6Addr` from an eight element 16-bit array. + /// + /// # Examples + /// + /// ``` + /// use std::net::Ipv6Addr; + /// + /// let addr = Ipv6Addr::from([ + /// 525u16, 524u16, 523u16, 522u16, + /// 521u16, 520u16, 519u16, 518u16, + /// ]); + /// assert_eq!( + /// Ipv6Addr::new( + /// 0x20d, 0x20c, + /// 0x20b, 0x20a, + /// 0x209, 0x208, + /// 0x207, 0x206 + /// ), + /// addr + /// ); + /// ``` + #[inline] + fn from(segments: [u16; 8]) -> Ipv6Addr { + let [a, b, c, d, e, f, g, h] = segments; + Ipv6Addr::new(a, b, c, d, e, f, g, h) + } +} + +#[stable(feature = "ip_from_slice", since = "1.17.0")] +impl From<[u8; 16]> for IpAddr { + /// Creates an `IpAddr::V6` from a sixteen element byte array. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv6Addr}; + /// + /// let addr = IpAddr::from([ + /// 25u8, 24u8, 23u8, 22u8, 21u8, 20u8, 19u8, 18u8, + /// 17u8, 16u8, 15u8, 14u8, 13u8, 12u8, 11u8, 10u8, + /// ]); + /// assert_eq!( + /// IpAddr::V6(Ipv6Addr::new( + /// 0x1918, 0x1716, + /// 0x1514, 0x1312, + /// 0x1110, 0x0f0e, + /// 0x0d0c, 0x0b0a + /// )), + /// addr + /// ); + /// ``` + #[inline] + fn from(octets: [u8; 16]) -> IpAddr { + IpAddr::V6(Ipv6Addr::from(octets)) + } +} + +#[stable(feature = "ip_from_slice", since = "1.17.0")] +impl From<[u16; 8]> for IpAddr { + /// Creates an `IpAddr::V6` from an eight element 16-bit array. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv6Addr}; + /// + /// let addr = IpAddr::from([ + /// 525u16, 524u16, 523u16, 522u16, + /// 521u16, 520u16, 519u16, 518u16, + /// ]); + /// assert_eq!( + /// IpAddr::V6(Ipv6Addr::new( + /// 0x20d, 0x20c, + /// 0x20b, 0x20a, + /// 0x209, 0x208, + /// 0x207, 0x206 + /// )), + /// addr + /// ); + /// ``` + #[inline] + fn from(segments: [u16; 8]) -> IpAddr { + IpAddr::V6(Ipv6Addr::from(segments)) + } +} diff --git a/library/std/src/net/ip_addr/tests.rs b/library/std/src/net/ip_addr/tests.rs new file mode 100644 index 000000000..7c3430b2b --- /dev/null +++ b/library/std/src/net/ip_addr/tests.rs @@ -0,0 +1,1039 @@ +use crate::net::test::{sa4, sa6, tsa}; +use crate::net::*; +use crate::str::FromStr; + +#[test] +fn test_from_str_ipv4() { + assert_eq!(Ok(Ipv4Addr::new(127, 0, 0, 1)), "127.0.0.1".parse()); + assert_eq!(Ok(Ipv4Addr::new(255, 255, 255, 255)), "255.255.255.255".parse()); + assert_eq!(Ok(Ipv4Addr::new(0, 0, 0, 0)), "0.0.0.0".parse()); + + // out of range + let none: Option<Ipv4Addr> = "256.0.0.1".parse().ok(); + assert_eq!(None, none); + // too short + let none: Option<Ipv4Addr> = "255.0.0".parse().ok(); + assert_eq!(None, none); + // too long + let none: Option<Ipv4Addr> = "255.0.0.1.2".parse().ok(); + assert_eq!(None, none); + // no number between dots + let none: Option<Ipv4Addr> = "255.0..1".parse().ok(); + assert_eq!(None, none); + // octal + let none: Option<Ipv4Addr> = "255.0.0.01".parse().ok(); + assert_eq!(None, none); + // octal zero + let none: Option<Ipv4Addr> = "255.0.0.00".parse().ok(); + assert_eq!(None, none); + let none: Option<Ipv4Addr> = "255.0.00.0".parse().ok(); + assert_eq!(None, none); +} + +#[test] +fn test_from_str_ipv6() { + assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)), "0:0:0:0:0:0:0:0".parse()); + assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)), "0:0:0:0:0:0:0:1".parse()); + + assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)), "::1".parse()); + assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)), "::".parse()); + + assert_eq!(Ok(Ipv6Addr::new(0x2a02, 0x6b8, 0, 0, 0, 0, 0x11, 0x11)), "2a02:6b8::11:11".parse()); + + // too long group + let none: Option<Ipv6Addr> = "::00000".parse().ok(); + assert_eq!(None, none); + // too short + let none: Option<Ipv6Addr> = "1:2:3:4:5:6:7".parse().ok(); + assert_eq!(None, none); + // too long + let none: Option<Ipv6Addr> = "1:2:3:4:5:6:7:8:9".parse().ok(); + assert_eq!(None, none); + // triple colon + let none: Option<Ipv6Addr> = "1:2:::6:7:8".parse().ok(); + assert_eq!(None, none); + // two double colons + let none: Option<Ipv6Addr> = "1:2::6::8".parse().ok(); + assert_eq!(None, none); + // `::` indicating zero groups of zeros + let none: Option<Ipv6Addr> = "1:2:3:4::5:6:7:8".parse().ok(); + assert_eq!(None, none); +} + +#[test] +fn test_from_str_ipv4_in_ipv6() { + assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 49152, 545)), "::192.0.2.33".parse()); + assert_eq!(Ok(Ipv6Addr::new(0, 0, 0, 0, 0, 0xFFFF, 49152, 545)), "::FFFF:192.0.2.33".parse()); + assert_eq!( + Ok(Ipv6Addr::new(0x64, 0xff9b, 0, 0, 0, 0, 49152, 545)), + "64:ff9b::192.0.2.33".parse() + ); + assert_eq!( + Ok(Ipv6Addr::new(0x2001, 0xdb8, 0x122, 0xc000, 0x2, 0x2100, 49152, 545)), + "2001:db8:122:c000:2:2100:192.0.2.33".parse() + ); + + // colon after v4 + let none: Option<Ipv4Addr> = "::127.0.0.1:".parse().ok(); + assert_eq!(None, none); + // not enough groups + let none: Option<Ipv6Addr> = "1:2:3:4:5:127.0.0.1".parse().ok(); + assert_eq!(None, none); + // too many groups + let none: Option<Ipv6Addr> = "1:2:3:4:5:6:7:127.0.0.1".parse().ok(); + assert_eq!(None, none); +} + +#[test] +fn test_from_str_socket_addr() { + assert_eq!(Ok(sa4(Ipv4Addr::new(77, 88, 21, 11), 80)), "77.88.21.11:80".parse()); + assert_eq!(Ok(SocketAddrV4::new(Ipv4Addr::new(77, 88, 21, 11), 80)), "77.88.21.11:80".parse()); + assert_eq!( + Ok(sa6(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53)), + "[2a02:6b8:0:1::1]:53".parse() + ); + assert_eq!( + Ok(SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53, 0, 0)), + "[2a02:6b8:0:1::1]:53".parse() + ); + assert_eq!(Ok(sa6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x7F00, 1), 22)), "[::127.0.0.1]:22".parse()); + assert_eq!( + Ok(SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x7F00, 1), 22, 0, 0)), + "[::127.0.0.1]:22".parse() + ); + + // without port + let none: Option<SocketAddr> = "127.0.0.1".parse().ok(); + assert_eq!(None, none); + // without port + let none: Option<SocketAddr> = "127.0.0.1:".parse().ok(); + assert_eq!(None, none); + // wrong brackets around v4 + let none: Option<SocketAddr> = "[127.0.0.1]:22".parse().ok(); + assert_eq!(None, none); + // port out of range + let none: Option<SocketAddr> = "127.0.0.1:123456".parse().ok(); + assert_eq!(None, none); +} + +#[test] +fn ipv4_addr_to_string() { + assert_eq!(Ipv4Addr::new(127, 0, 0, 1).to_string(), "127.0.0.1"); + // Short address + assert_eq!(Ipv4Addr::new(1, 1, 1, 1).to_string(), "1.1.1.1"); + // Long address + assert_eq!(Ipv4Addr::new(127, 127, 127, 127).to_string(), "127.127.127.127"); + + // Test padding + assert_eq!(&format!("{:16}", Ipv4Addr::new(1, 1, 1, 1)), "1.1.1.1 "); + assert_eq!(&format!("{:>16}", Ipv4Addr::new(1, 1, 1, 1)), " 1.1.1.1"); +} + +#[test] +fn ipv6_addr_to_string() { + // ipv4-mapped address + let a1 = Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc000, 0x280); + assert_eq!(a1.to_string(), "::ffff:192.0.2.128"); + + // ipv4-compatible address + let a1 = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0xc000, 0x280); + assert_eq!(a1.to_string(), "::192.0.2.128"); + + // v6 address with no zero segments + assert_eq!(Ipv6Addr::new(8, 9, 10, 11, 12, 13, 14, 15).to_string(), "8:9:a:b:c:d:e:f"); + + // longest possible IPv6 length + assert_eq!( + Ipv6Addr::new(0x1111, 0x2222, 0x3333, 0x4444, 0x5555, 0x6666, 0x7777, 0x8888).to_string(), + "1111:2222:3333:4444:5555:6666:7777:8888" + ); + // padding + assert_eq!(&format!("{:20}", Ipv6Addr::new(1, 2, 3, 4, 5, 6, 7, 8)), "1:2:3:4:5:6:7:8 "); + assert_eq!(&format!("{:>20}", Ipv6Addr::new(1, 2, 3, 4, 5, 6, 7, 8)), " 1:2:3:4:5:6:7:8"); + + // reduce a single run of zeros + assert_eq!( + "ae::ffff:102:304", + Ipv6Addr::new(0xae, 0, 0, 0, 0, 0xffff, 0x0102, 0x0304).to_string() + ); + + // don't reduce just a single zero segment + assert_eq!("1:2:3:4:5:6:0:8", Ipv6Addr::new(1, 2, 3, 4, 5, 6, 0, 8).to_string()); + + // 'any' address + assert_eq!("::", Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0).to_string()); + + // loopback address + assert_eq!("::1", Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1).to_string()); + + // ends in zeros + assert_eq!("1::", Ipv6Addr::new(1, 0, 0, 0, 0, 0, 0, 0).to_string()); + + // two runs of zeros, second one is longer + assert_eq!("1:0:0:4::8", Ipv6Addr::new(1, 0, 0, 4, 0, 0, 0, 8).to_string()); + + // two runs of zeros, equal length + assert_eq!("1::4:5:0:0:8", Ipv6Addr::new(1, 0, 0, 4, 5, 0, 0, 8).to_string()); + + // don't prefix `0x` to each segment in `dbg!`. + assert_eq!("1::4:5:0:0:8", &format!("{:#?}", Ipv6Addr::new(1, 0, 0, 4, 5, 0, 0, 8))); +} + +#[test] +fn ipv4_to_ipv6() { + assert_eq!( + Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678), + Ipv4Addr::new(0x12, 0x34, 0x56, 0x78).to_ipv6_mapped() + ); + assert_eq!( + Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x1234, 0x5678), + Ipv4Addr::new(0x12, 0x34, 0x56, 0x78).to_ipv6_compatible() + ); +} + +#[test] +fn ipv6_to_ipv4_mapped() { + assert_eq!( + Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678).to_ipv4_mapped(), + Some(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78)) + ); + assert_eq!(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x1234, 0x5678).to_ipv4_mapped(), None); +} + +#[test] +fn ipv6_to_ipv4() { + assert_eq!( + Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678).to_ipv4(), + Some(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78)) + ); + assert_eq!( + Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0x1234, 0x5678).to_ipv4(), + Some(Ipv4Addr::new(0x12, 0x34, 0x56, 0x78)) + ); + assert_eq!(Ipv6Addr::new(0, 0, 1, 0, 0, 0, 0x1234, 0x5678).to_ipv4(), None); +} + +#[test] +fn ip_properties() { + macro_rules! ip { + ($s:expr) => { + IpAddr::from_str($s).unwrap() + }; + } + + macro_rules! check { + ($s:expr) => { + check!($s, 0); + }; + + ($s:expr, $mask:expr) => {{ + let unspec: u8 = 1 << 0; + let loopback: u8 = 1 << 1; + let global: u8 = 1 << 2; + let multicast: u8 = 1 << 3; + let doc: u8 = 1 << 4; + let benchmarking: u8 = 1 << 5; + + if ($mask & unspec) == unspec { + assert!(ip!($s).is_unspecified()); + } else { + assert!(!ip!($s).is_unspecified()); + } + + if ($mask & loopback) == loopback { + assert!(ip!($s).is_loopback()); + } else { + assert!(!ip!($s).is_loopback()); + } + + if ($mask & global) == global { + assert!(ip!($s).is_global()); + } else { + assert!(!ip!($s).is_global()); + } + + if ($mask & multicast) == multicast { + assert!(ip!($s).is_multicast()); + } else { + assert!(!ip!($s).is_multicast()); + } + + if ($mask & doc) == doc { + assert!(ip!($s).is_documentation()); + } else { + assert!(!ip!($s).is_documentation()); + } + + if ($mask & benchmarking) == benchmarking { + assert!(ip!($s).is_benchmarking()); + } else { + assert!(!ip!($s).is_benchmarking()); + } + }}; + } + + let unspec: u8 = 1 << 0; + let loopback: u8 = 1 << 1; + let global: u8 = 1 << 2; + let multicast: u8 = 1 << 3; + let doc: u8 = 1 << 4; + let benchmarking: u8 = 1 << 5; + + check!("0.0.0.0", unspec); + check!("0.0.0.1"); + check!("0.1.0.0"); + check!("10.9.8.7"); + check!("127.1.2.3", loopback); + check!("172.31.254.253"); + check!("169.254.253.242"); + check!("192.0.2.183", doc); + check!("192.1.2.183", global); + check!("192.168.254.253"); + check!("198.51.100.0", doc); + check!("203.0.113.0", doc); + check!("203.2.113.0", global); + check!("224.0.0.0", global | multicast); + check!("239.255.255.255", global | multicast); + check!("255.255.255.255"); + // make sure benchmarking addresses are not global + check!("198.18.0.0", benchmarking); + check!("198.18.54.2", benchmarking); + check!("198.19.255.255", benchmarking); + // make sure addresses reserved for protocol assignment are not global + check!("192.0.0.0"); + check!("192.0.0.255"); + check!("192.0.0.100"); + // make sure reserved addresses are not global + check!("240.0.0.0"); + check!("251.54.1.76"); + check!("254.255.255.255"); + // make sure shared addresses are not global + check!("100.64.0.0"); + check!("100.127.255.255"); + check!("100.100.100.0"); + + check!("::", unspec); + check!("::1", loopback); + check!("::0.0.0.2", global); + check!("1::", global); + check!("fc00::"); + check!("fdff:ffff::"); + check!("fe80:ffff::"); + check!("febf:ffff::"); + check!("fec0::", global); + check!("ff01::", global | multicast); + check!("ff02::", global | multicast); + check!("ff03::", global | multicast); + check!("ff04::", global | multicast); + check!("ff05::", global | multicast); + check!("ff08::", global | multicast); + check!("ff0e::", global | multicast); + check!("2001:db8:85a3::8a2e:370:7334", doc); + check!("2001:2::ac32:23ff:21", benchmarking); + check!("102:304:506:708:90a:b0c:d0e:f10", global); +} + +#[test] +fn ipv4_properties() { + macro_rules! ip { + ($s:expr) => { + Ipv4Addr::from_str($s).unwrap() + }; + } + + macro_rules! check { + ($s:expr) => { + check!($s, 0); + }; + + ($s:expr, $mask:expr) => {{ + let unspec: u16 = 1 << 0; + let loopback: u16 = 1 << 1; + let private: u16 = 1 << 2; + let link_local: u16 = 1 << 3; + let global: u16 = 1 << 4; + let multicast: u16 = 1 << 5; + let broadcast: u16 = 1 << 6; + let documentation: u16 = 1 << 7; + let benchmarking: u16 = 1 << 8; + let reserved: u16 = 1 << 10; + let shared: u16 = 1 << 11; + + if ($mask & unspec) == unspec { + assert!(ip!($s).is_unspecified()); + } else { + assert!(!ip!($s).is_unspecified()); + } + + if ($mask & loopback) == loopback { + assert!(ip!($s).is_loopback()); + } else { + assert!(!ip!($s).is_loopback()); + } + + if ($mask & private) == private { + assert!(ip!($s).is_private()); + } else { + assert!(!ip!($s).is_private()); + } + + if ($mask & link_local) == link_local { + assert!(ip!($s).is_link_local()); + } else { + assert!(!ip!($s).is_link_local()); + } + + if ($mask & global) == global { + assert!(ip!($s).is_global()); + } else { + assert!(!ip!($s).is_global()); + } + + if ($mask & multicast) == multicast { + assert!(ip!($s).is_multicast()); + } else { + assert!(!ip!($s).is_multicast()); + } + + if ($mask & broadcast) == broadcast { + assert!(ip!($s).is_broadcast()); + } else { + assert!(!ip!($s).is_broadcast()); + } + + if ($mask & documentation) == documentation { + assert!(ip!($s).is_documentation()); + } else { + assert!(!ip!($s).is_documentation()); + } + + if ($mask & benchmarking) == benchmarking { + assert!(ip!($s).is_benchmarking()); + } else { + assert!(!ip!($s).is_benchmarking()); + } + + if ($mask & reserved) == reserved { + assert!(ip!($s).is_reserved()); + } else { + assert!(!ip!($s).is_reserved()); + } + + if ($mask & shared) == shared { + assert!(ip!($s).is_shared()); + } else { + assert!(!ip!($s).is_shared()); + } + }}; + } + + let unspec: u16 = 1 << 0; + let loopback: u16 = 1 << 1; + let private: u16 = 1 << 2; + let link_local: u16 = 1 << 3; + let global: u16 = 1 << 4; + let multicast: u16 = 1 << 5; + let broadcast: u16 = 1 << 6; + let documentation: u16 = 1 << 7; + let benchmarking: u16 = 1 << 8; + let reserved: u16 = 1 << 10; + let shared: u16 = 1 << 11; + + check!("0.0.0.0", unspec); + check!("0.0.0.1"); + check!("0.1.0.0"); + check!("10.9.8.7", private); + check!("127.1.2.3", loopback); + check!("172.31.254.253", private); + check!("169.254.253.242", link_local); + check!("192.0.2.183", documentation); + check!("192.1.2.183", global); + check!("192.168.254.253", private); + check!("198.51.100.0", documentation); + check!("203.0.113.0", documentation); + check!("203.2.113.0", global); + check!("224.0.0.0", global | multicast); + check!("239.255.255.255", global | multicast); + check!("255.255.255.255", broadcast); + check!("198.18.0.0", benchmarking); + check!("198.18.54.2", benchmarking); + check!("198.19.255.255", benchmarking); + check!("192.0.0.0"); + check!("192.0.0.255"); + check!("192.0.0.100"); + check!("240.0.0.0", reserved); + check!("251.54.1.76", reserved); + check!("254.255.255.255", reserved); + check!("100.64.0.0", shared); + check!("100.127.255.255", shared); + check!("100.100.100.0", shared); +} + +#[test] +fn ipv6_properties() { + macro_rules! ip { + ($s:expr) => { + Ipv6Addr::from_str($s).unwrap() + }; + } + + macro_rules! check { + ($s:expr, &[$($octet:expr),*], $mask:expr) => { + assert_eq!($s, ip!($s).to_string()); + let octets = &[$($octet),*]; + assert_eq!(&ip!($s).octets(), octets); + assert_eq!(Ipv6Addr::from(*octets), ip!($s)); + + let unspecified: u32 = 1 << 0; + let loopback: u32 = 1 << 1; + let unique_local: u32 = 1 << 2; + let global: u32 = 1 << 3; + let unicast_link_local: u32 = 1 << 4; + let unicast_global: u32 = 1 << 7; + let documentation: u32 = 1 << 8; + let benchmarking: u32 = 1 << 16; + let multicast_interface_local: u32 = 1 << 9; + let multicast_link_local: u32 = 1 << 10; + let multicast_realm_local: u32 = 1 << 11; + let multicast_admin_local: u32 = 1 << 12; + let multicast_site_local: u32 = 1 << 13; + let multicast_organization_local: u32 = 1 << 14; + let multicast_global: u32 = 1 << 15; + let multicast: u32 = multicast_interface_local + | multicast_admin_local + | multicast_global + | multicast_link_local + | multicast_realm_local + | multicast_site_local + | multicast_organization_local; + + if ($mask & unspecified) == unspecified { + assert!(ip!($s).is_unspecified()); + } else { + assert!(!ip!($s).is_unspecified()); + } + if ($mask & loopback) == loopback { + assert!(ip!($s).is_loopback()); + } else { + assert!(!ip!($s).is_loopback()); + } + if ($mask & unique_local) == unique_local { + assert!(ip!($s).is_unique_local()); + } else { + assert!(!ip!($s).is_unique_local()); + } + if ($mask & global) == global { + assert!(ip!($s).is_global()); + } else { + assert!(!ip!($s).is_global()); + } + if ($mask & unicast_link_local) == unicast_link_local { + assert!(ip!($s).is_unicast_link_local()); + } else { + assert!(!ip!($s).is_unicast_link_local()); + } + if ($mask & unicast_global) == unicast_global { + assert!(ip!($s).is_unicast_global()); + } else { + assert!(!ip!($s).is_unicast_global()); + } + if ($mask & documentation) == documentation { + assert!(ip!($s).is_documentation()); + } else { + assert!(!ip!($s).is_documentation()); + } + if ($mask & benchmarking) == benchmarking { + assert!(ip!($s).is_benchmarking()); + } else { + assert!(!ip!($s).is_benchmarking()); + } + if ($mask & multicast) != 0 { + assert!(ip!($s).multicast_scope().is_some()); + assert!(ip!($s).is_multicast()); + } else { + assert!(ip!($s).multicast_scope().is_none()); + assert!(!ip!($s).is_multicast()); + } + if ($mask & multicast_interface_local) == multicast_interface_local { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::InterfaceLocal); + } + if ($mask & multicast_link_local) == multicast_link_local { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::LinkLocal); + } + if ($mask & multicast_realm_local) == multicast_realm_local { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::RealmLocal); + } + if ($mask & multicast_admin_local) == multicast_admin_local { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::AdminLocal); + } + if ($mask & multicast_site_local) == multicast_site_local { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::SiteLocal); + } + if ($mask & multicast_organization_local) == multicast_organization_local { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::OrganizationLocal); + } + if ($mask & multicast_global) == multicast_global { + assert_eq!(ip!($s).multicast_scope().unwrap(), + Ipv6MulticastScope::Global); + } + } + } + + let unspecified: u32 = 1 << 0; + let loopback: u32 = 1 << 1; + let unique_local: u32 = 1 << 2; + let global: u32 = 1 << 3; + let unicast_link_local: u32 = 1 << 4; + let unicast_global: u32 = 1 << 7; + let documentation: u32 = 1 << 8; + let benchmarking: u32 = 1 << 16; + let multicast_interface_local: u32 = 1 << 9; + let multicast_link_local: u32 = 1 << 10; + let multicast_realm_local: u32 = 1 << 11; + let multicast_admin_local: u32 = 1 << 12; + let multicast_site_local: u32 = 1 << 13; + let multicast_organization_local: u32 = 1 << 14; + let multicast_global: u32 = 1 << 15; + + check!("::", &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unspecified); + + check!("::1", &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1], loopback); + + check!("::0.0.0.2", &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2], global | unicast_global); + + check!("1::", &[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], global | unicast_global); + + check!( + "::ffff:127.0.0.1", + &[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0xff, 0xff, 0x7f, 0, 0, 1], + unicast_global + ); + + check!( + "64:ff9b:1::", + &[0, 0x64, 0xff, 0x9b, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + unicast_global + ); + + check!("100::", &[0x01, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_global); + + check!("2001::", &[0x20, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_global); + + check!( + "2001:1::1", + &[0x20, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1], + global | unicast_global + ); + + check!( + "2001:1::2", + &[0x20, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2], + global | unicast_global + ); + + check!( + "2001:3::", + &[0x20, 1, 0, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + global | unicast_global + ); + + check!( + "2001:4:112::", + &[0x20, 1, 0, 4, 1, 0x12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + global | unicast_global + ); + + check!( + "2001:20::", + &[0x20, 1, 0, 0x20, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + global | unicast_global + ); + + check!("2001:30::", &[0x20, 1, 0, 0x30, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_global); + + check!( + "2001:200::", + &[0x20, 1, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + global | unicast_global + ); + + check!("fc00::", &[0xfc, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unique_local); + + check!( + "fdff:ffff::", + &[0xfd, 0xff, 0xff, 0xff, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + unique_local + ); + + check!( + "fe80:ffff::", + &[0xfe, 0x80, 0xff, 0xff, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + unicast_link_local + ); + + check!("fe80::", &[0xfe, 0x80, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_link_local); + + check!( + "febf:ffff::", + &[0xfe, 0xbf, 0xff, 0xff, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + unicast_link_local + ); + + check!("febf::", &[0xfe, 0xbf, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], unicast_link_local); + + check!( + "febf:ffff:ffff:ffff:ffff:ffff:ffff:ffff", + &[ + 0xfe, 0xbf, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff + ], + unicast_link_local + ); + + check!( + "fe80::ffff:ffff:ffff:ffff", + &[ + 0xfe, 0x80, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, + 0xff, 0xff + ], + unicast_link_local + ); + + check!( + "fe80:0:0:1::", + &[0xfe, 0x80, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0], + unicast_link_local + ); + + check!( + "fec0::", + &[0xfe, 0xc0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + unicast_global | global + ); + + check!( + "ff01::", + &[0xff, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_interface_local | global + ); + + check!( + "ff02::", + &[0xff, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_link_local | global + ); + + check!( + "ff03::", + &[0xff, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_realm_local | global + ); + + check!( + "ff04::", + &[0xff, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_admin_local | global + ); + + check!( + "ff05::", + &[0xff, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_site_local | global + ); + + check!( + "ff08::", + &[0xff, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_organization_local | global + ); + + check!( + "ff0e::", + &[0xff, 0xe, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], + multicast_global | global + ); + + check!( + "2001:db8:85a3::8a2e:370:7334", + &[0x20, 1, 0xd, 0xb8, 0x85, 0xa3, 0, 0, 0, 0, 0x8a, 0x2e, 3, 0x70, 0x73, 0x34], + documentation + ); + + check!( + "2001:2::ac32:23ff:21", + &[0x20, 1, 0, 2, 0, 0, 0, 0, 0, 0, 0xac, 0x32, 0x23, 0xff, 0, 0x21], + benchmarking + ); + + check!( + "102:304:506:708:90a:b0c:d0e:f10", + &[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], + global | unicast_global + ); +} + +#[test] +fn to_socket_addr_socketaddr() { + let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 12345); + assert_eq!(Ok(vec![a]), tsa(a)); +} + +#[test] +fn test_ipv4_to_int() { + let a = Ipv4Addr::new(0x11, 0x22, 0x33, 0x44); + assert_eq!(u32::from(a), 0x11223344); +} + +#[test] +fn test_int_to_ipv4() { + let a = Ipv4Addr::new(0x11, 0x22, 0x33, 0x44); + assert_eq!(Ipv4Addr::from(0x11223344), a); +} + +#[test] +fn test_ipv6_to_int() { + let a = Ipv6Addr::new(0x1122, 0x3344, 0x5566, 0x7788, 0x99aa, 0xbbcc, 0xddee, 0xff11); + assert_eq!(u128::from(a), 0x112233445566778899aabbccddeeff11u128); +} + +#[test] +fn test_int_to_ipv6() { + let a = Ipv6Addr::new(0x1122, 0x3344, 0x5566, 0x7788, 0x99aa, 0xbbcc, 0xddee, 0xff11); + assert_eq!(Ipv6Addr::from(0x112233445566778899aabbccddeeff11u128), a); +} + +#[test] +fn ipv4_from_constructors() { + assert_eq!(Ipv4Addr::LOCALHOST, Ipv4Addr::new(127, 0, 0, 1)); + assert!(Ipv4Addr::LOCALHOST.is_loopback()); + assert_eq!(Ipv4Addr::UNSPECIFIED, Ipv4Addr::new(0, 0, 0, 0)); + assert!(Ipv4Addr::UNSPECIFIED.is_unspecified()); + assert_eq!(Ipv4Addr::BROADCAST, Ipv4Addr::new(255, 255, 255, 255)); + assert!(Ipv4Addr::BROADCAST.is_broadcast()); +} + +#[test] +fn ipv6_from_constructors() { + assert_eq!(Ipv6Addr::LOCALHOST, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); + assert!(Ipv6Addr::LOCALHOST.is_loopback()); + assert_eq!(Ipv6Addr::UNSPECIFIED, Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 0)); + assert!(Ipv6Addr::UNSPECIFIED.is_unspecified()); +} + +#[test] +fn ipv4_from_octets() { + assert_eq!(Ipv4Addr::from([127, 0, 0, 1]), Ipv4Addr::new(127, 0, 0, 1)) +} + +#[test] +fn ipv6_from_segments() { + let from_u16s = + Ipv6Addr::from([0x0011, 0x2233, 0x4455, 0x6677, 0x8899, 0xaabb, 0xccdd, 0xeeff]); + let new = Ipv6Addr::new(0x0011, 0x2233, 0x4455, 0x6677, 0x8899, 0xaabb, 0xccdd, 0xeeff); + assert_eq!(new, from_u16s); +} + +#[test] +fn ipv6_from_octets() { + let from_u16s = + Ipv6Addr::from([0x0011, 0x2233, 0x4455, 0x6677, 0x8899, 0xaabb, 0xccdd, 0xeeff]); + let from_u8s = Ipv6Addr::from([ + 0x00, 0x11, 0x22, 0x33, 0x44, 0x55, 0x66, 0x77, 0x88, 0x99, 0xaa, 0xbb, 0xcc, 0xdd, 0xee, + 0xff, + ]); + assert_eq!(from_u16s, from_u8s); +} + +#[test] +fn cmp() { + let v41 = Ipv4Addr::new(100, 64, 3, 3); + let v42 = Ipv4Addr::new(192, 0, 2, 2); + let v61 = "2001:db8:f00::1002".parse::<Ipv6Addr>().unwrap(); + let v62 = "2001:db8:f00::2001".parse::<Ipv6Addr>().unwrap(); + assert!(v41 < v42); + assert!(v61 < v62); + + assert_eq!(v41, IpAddr::V4(v41)); + assert_eq!(v61, IpAddr::V6(v61)); + assert!(v41 != IpAddr::V4(v42)); + assert!(v61 != IpAddr::V6(v62)); + + assert!(v41 < IpAddr::V4(v42)); + assert!(v61 < IpAddr::V6(v62)); + assert!(IpAddr::V4(v41) < v42); + assert!(IpAddr::V6(v61) < v62); + + assert!(v41 < IpAddr::V6(v61)); + assert!(IpAddr::V4(v41) < v61); +} + +#[test] +fn is_v4() { + let ip = IpAddr::V4(Ipv4Addr::new(100, 64, 3, 3)); + assert!(ip.is_ipv4()); + assert!(!ip.is_ipv6()); +} + +#[test] +fn is_v6() { + let ip = IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0x1234, 0x5678)); + assert!(!ip.is_ipv4()); + assert!(ip.is_ipv6()); +} + +#[test] +fn ipv4_const() { + // test that the methods of `Ipv4Addr` are usable in a const context + + const IP_ADDRESS: Ipv4Addr = Ipv4Addr::new(127, 0, 0, 1); + assert_eq!(IP_ADDRESS, Ipv4Addr::LOCALHOST); + + const OCTETS: [u8; 4] = IP_ADDRESS.octets(); + assert_eq!(OCTETS, [127, 0, 0, 1]); + + const IS_UNSPECIFIED: bool = IP_ADDRESS.is_unspecified(); + assert!(!IS_UNSPECIFIED); + + const IS_LOOPBACK: bool = IP_ADDRESS.is_loopback(); + assert!(IS_LOOPBACK); + + const IS_PRIVATE: bool = IP_ADDRESS.is_private(); + assert!(!IS_PRIVATE); + + const IS_LINK_LOCAL: bool = IP_ADDRESS.is_link_local(); + assert!(!IS_LINK_LOCAL); + + const IS_GLOBAL: bool = IP_ADDRESS.is_global(); + assert!(!IS_GLOBAL); + + const IS_SHARED: bool = IP_ADDRESS.is_shared(); + assert!(!IS_SHARED); + + const IS_BENCHMARKING: bool = IP_ADDRESS.is_benchmarking(); + assert!(!IS_BENCHMARKING); + + const IS_RESERVED: bool = IP_ADDRESS.is_reserved(); + assert!(!IS_RESERVED); + + const IS_MULTICAST: bool = IP_ADDRESS.is_multicast(); + assert!(!IS_MULTICAST); + + const IS_BROADCAST: bool = IP_ADDRESS.is_broadcast(); + assert!(!IS_BROADCAST); + + const IS_DOCUMENTATION: bool = IP_ADDRESS.is_documentation(); + assert!(!IS_DOCUMENTATION); + + const IP_V6_COMPATIBLE: Ipv6Addr = IP_ADDRESS.to_ipv6_compatible(); + assert_eq!( + IP_V6_COMPATIBLE, + Ipv6Addr::from([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 127, 0, 0, 1]) + ); + + const IP_V6_MAPPED: Ipv6Addr = IP_ADDRESS.to_ipv6_mapped(); + assert_eq!( + IP_V6_MAPPED, + Ipv6Addr::from([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 255, 255, 127, 0, 0, 1]) + ); +} + +#[test] +fn ipv6_const() { + // test that the methods of `Ipv6Addr` are usable in a const context + + const IP_ADDRESS: Ipv6Addr = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); + assert_eq!(IP_ADDRESS, Ipv6Addr::LOCALHOST); + + const SEGMENTS: [u16; 8] = IP_ADDRESS.segments(); + assert_eq!(SEGMENTS, [0, 0, 0, 0, 0, 0, 0, 1]); + + const OCTETS: [u8; 16] = IP_ADDRESS.octets(); + assert_eq!(OCTETS, [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]); + + const IS_UNSPECIFIED: bool = IP_ADDRESS.is_unspecified(); + assert!(!IS_UNSPECIFIED); + + const IS_LOOPBACK: bool = IP_ADDRESS.is_loopback(); + assert!(IS_LOOPBACK); + + const IS_GLOBAL: bool = IP_ADDRESS.is_global(); + assert!(!IS_GLOBAL); + + const IS_UNIQUE_LOCAL: bool = IP_ADDRESS.is_unique_local(); + assert!(!IS_UNIQUE_LOCAL); + + const IS_UNICAST_LINK_LOCAL: bool = IP_ADDRESS.is_unicast_link_local(); + assert!(!IS_UNICAST_LINK_LOCAL); + + const IS_DOCUMENTATION: bool = IP_ADDRESS.is_documentation(); + assert!(!IS_DOCUMENTATION); + + const IS_BENCHMARKING: bool = IP_ADDRESS.is_benchmarking(); + assert!(!IS_BENCHMARKING); + + const IS_UNICAST_GLOBAL: bool = IP_ADDRESS.is_unicast_global(); + assert!(!IS_UNICAST_GLOBAL); + + const MULTICAST_SCOPE: Option<Ipv6MulticastScope> = IP_ADDRESS.multicast_scope(); + assert_eq!(MULTICAST_SCOPE, None); + + const IS_MULTICAST: bool = IP_ADDRESS.is_multicast(); + assert!(!IS_MULTICAST); + + const IP_V4: Option<Ipv4Addr> = IP_ADDRESS.to_ipv4(); + assert_eq!(IP_V4.unwrap(), Ipv4Addr::new(0, 0, 0, 1)); +} + +#[test] +fn ip_const() { + // test that the methods of `IpAddr` are usable in a const context + + const IP_ADDRESS: IpAddr = IpAddr::V4(Ipv4Addr::LOCALHOST); + + const IS_UNSPECIFIED: bool = IP_ADDRESS.is_unspecified(); + assert!(!IS_UNSPECIFIED); + + const IS_LOOPBACK: bool = IP_ADDRESS.is_loopback(); + assert!(IS_LOOPBACK); + + const IS_GLOBAL: bool = IP_ADDRESS.is_global(); + assert!(!IS_GLOBAL); + + const IS_MULTICAST: bool = IP_ADDRESS.is_multicast(); + assert!(!IS_MULTICAST); + + const IS_IP_V4: bool = IP_ADDRESS.is_ipv4(); + assert!(IS_IP_V4); + + const IS_IP_V6: bool = IP_ADDRESS.is_ipv6(); + assert!(!IS_IP_V6); +} + +#[test] +fn structural_match() { + // test that all IP types can be structurally matched upon + + const IPV4: Ipv4Addr = Ipv4Addr::LOCALHOST; + match IPV4 { + Ipv4Addr::LOCALHOST => {} + _ => unreachable!(), + } + + const IPV6: Ipv6Addr = Ipv6Addr::LOCALHOST; + match IPV6 { + Ipv6Addr::LOCALHOST => {} + _ => unreachable!(), + } + + const IP: IpAddr = IpAddr::V4(Ipv4Addr::LOCALHOST); + match IP { + IpAddr::V4(Ipv4Addr::LOCALHOST) => {} + _ => unreachable!(), + } +} diff --git a/library/std/src/net/mod.rs b/library/std/src/net/mod.rs index e7a40bdaf..01e3db9de 100644 --- a/library/std/src/net/mod.rs +++ b/library/std/src/net/mod.rs @@ -24,11 +24,11 @@ use crate::io::{self, ErrorKind}; #[stable(feature = "rust1", since = "1.0.0")] -pub use self::addr::{SocketAddr, SocketAddrV4, SocketAddrV6, ToSocketAddrs}; -#[stable(feature = "rust1", since = "1.0.0")] -pub use self::ip::{IpAddr, Ipv4Addr, Ipv6Addr, Ipv6MulticastScope}; +pub use self::ip_addr::{IpAddr, Ipv4Addr, Ipv6Addr, Ipv6MulticastScope}; #[stable(feature = "rust1", since = "1.0.0")] pub use self::parser::AddrParseError; +#[stable(feature = "rust1", since = "1.0.0")] +pub use self::socket_addr::{SocketAddr, SocketAddrV4, SocketAddrV6, ToSocketAddrs}; #[unstable(feature = "tcplistener_into_incoming", issue = "88339")] pub use self::tcp::IntoIncoming; #[stable(feature = "rust1", since = "1.0.0")] @@ -36,12 +36,13 @@ pub use self::tcp::{Incoming, TcpListener, TcpStream}; #[stable(feature = "rust1", since = "1.0.0")] pub use self::udp::UdpSocket; -mod addr; -mod ip; +mod display_buffer; +mod ip_addr; mod parser; +mod socket_addr; mod tcp; #[cfg(test)] -mod test; +pub(crate) mod test; mod udp; /// Possible values which can be passed to the [`TcpStream::shutdown`] method. diff --git a/library/std/src/net/parser.rs b/library/std/src/net/parser.rs index 069b66099..a38031c48 100644 --- a/library/std/src/net/parser.rs +++ b/library/std/src/net/parser.rs @@ -39,8 +39,8 @@ struct Parser<'a> { } impl<'a> Parser<'a> { - fn new(input: &'a str) -> Parser<'a> { - Parser { state: input.as_bytes() } + fn new(input: &'a [u8]) -> Parser<'a> { + Parser { state: input } } /// Run a parser, and restore the pre-parse state if it fails. @@ -273,32 +273,106 @@ impl<'a> Parser<'a> { } } +impl IpAddr { + /// Parse an IP address from a slice of bytes. + /// + /// ``` + /// #![feature(addr_parse_ascii)] + /// + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr}; + /// + /// let localhost_v4 = IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)); + /// let localhost_v6 = IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); + /// + /// assert_eq!(IpAddr::parse_ascii(b"127.0.0.1"), Ok(localhost_v4)); + /// assert_eq!(IpAddr::parse_ascii(b"::1"), Ok(localhost_v6)); + /// ``` + #[unstable(feature = "addr_parse_ascii", issue = "101035")] + pub fn parse_ascii(b: &[u8]) -> Result<Self, AddrParseError> { + Parser::new(b).parse_with(|p| p.read_ip_addr(), AddrKind::Ip) + } +} + #[stable(feature = "ip_addr", since = "1.7.0")] impl FromStr for IpAddr { type Err = AddrParseError; fn from_str(s: &str) -> Result<IpAddr, AddrParseError> { - Parser::new(s).parse_with(|p| p.read_ip_addr(), AddrKind::Ip) + Self::parse_ascii(s.as_bytes()) } } -#[stable(feature = "rust1", since = "1.0.0")] -impl FromStr for Ipv4Addr { - type Err = AddrParseError; - fn from_str(s: &str) -> Result<Ipv4Addr, AddrParseError> { +impl Ipv4Addr { + /// Parse an IPv4 address from a slice of bytes. + /// + /// ``` + /// #![feature(addr_parse_ascii)] + /// + /// use std::net::Ipv4Addr; + /// + /// let localhost = Ipv4Addr::new(127, 0, 0, 1); + /// + /// assert_eq!(Ipv4Addr::parse_ascii(b"127.0.0.1"), Ok(localhost)); + /// ``` + #[unstable(feature = "addr_parse_ascii", issue = "101035")] + pub fn parse_ascii(b: &[u8]) -> Result<Self, AddrParseError> { // don't try to parse if too long - if s.len() > 15 { + if b.len() > 15 { Err(AddrParseError(AddrKind::Ipv4)) } else { - Parser::new(s).parse_with(|p| p.read_ipv4_addr(), AddrKind::Ipv4) + Parser::new(b).parse_with(|p| p.read_ipv4_addr(), AddrKind::Ipv4) } } } #[stable(feature = "rust1", since = "1.0.0")] +impl FromStr for Ipv4Addr { + type Err = AddrParseError; + fn from_str(s: &str) -> Result<Ipv4Addr, AddrParseError> { + Self::parse_ascii(s.as_bytes()) + } +} + +impl Ipv6Addr { + /// Parse an IPv6 address from a slice of bytes. + /// + /// ``` + /// #![feature(addr_parse_ascii)] + /// + /// use std::net::Ipv6Addr; + /// + /// let localhost = Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1); + /// + /// assert_eq!(Ipv6Addr::parse_ascii(b"::1"), Ok(localhost)); + /// ``` + #[unstable(feature = "addr_parse_ascii", issue = "101035")] + pub fn parse_ascii(b: &[u8]) -> Result<Self, AddrParseError> { + Parser::new(b).parse_with(|p| p.read_ipv6_addr(), AddrKind::Ipv6) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] impl FromStr for Ipv6Addr { type Err = AddrParseError; fn from_str(s: &str) -> Result<Ipv6Addr, AddrParseError> { - Parser::new(s).parse_with(|p| p.read_ipv6_addr(), AddrKind::Ipv6) + Self::parse_ascii(s.as_bytes()) + } +} + +impl SocketAddrV4 { + /// Parse an IPv4 socket address from a slice of bytes. + /// + /// ``` + /// #![feature(addr_parse_ascii)] + /// + /// use std::net::{Ipv4Addr, SocketAddrV4}; + /// + /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); + /// + /// assert_eq!(SocketAddrV4::parse_ascii(b"127.0.0.1:8080"), Ok(socket)); + /// ``` + #[unstable(feature = "addr_parse_ascii", issue = "101035")] + pub fn parse_ascii(b: &[u8]) -> Result<Self, AddrParseError> { + Parser::new(b).parse_with(|p| p.read_socket_addr_v4(), AddrKind::SocketV4) } } @@ -306,7 +380,25 @@ impl FromStr for Ipv6Addr { impl FromStr for SocketAddrV4 { type Err = AddrParseError; fn from_str(s: &str) -> Result<SocketAddrV4, AddrParseError> { - Parser::new(s).parse_with(|p| p.read_socket_addr_v4(), AddrKind::SocketV4) + Self::parse_ascii(s.as_bytes()) + } +} + +impl SocketAddrV6 { + /// Parse an IPv6 socket address from a slice of bytes. + /// + /// ``` + /// #![feature(addr_parse_ascii)] + /// + /// use std::net::{Ipv6Addr, SocketAddrV6}; + /// + /// let socket = SocketAddrV6::new(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 1), 8080, 0, 0); + /// + /// assert_eq!(SocketAddrV6::parse_ascii(b"[2001:db8::1]:8080"), Ok(socket)); + /// ``` + #[unstable(feature = "addr_parse_ascii", issue = "101035")] + pub fn parse_ascii(b: &[u8]) -> Result<Self, AddrParseError> { + Parser::new(b).parse_with(|p| p.read_socket_addr_v6(), AddrKind::SocketV6) } } @@ -314,7 +406,27 @@ impl FromStr for SocketAddrV4 { impl FromStr for SocketAddrV6 { type Err = AddrParseError; fn from_str(s: &str) -> Result<SocketAddrV6, AddrParseError> { - Parser::new(s).parse_with(|p| p.read_socket_addr_v6(), AddrKind::SocketV6) + Self::parse_ascii(s.as_bytes()) + } +} + +impl SocketAddr { + /// Parse a socket address from a slice of bytes. + /// + /// ``` + /// #![feature(addr_parse_ascii)] + /// + /// use std::net::{IpAddr, Ipv4Addr, Ipv6Addr, SocketAddr}; + /// + /// let socket_v4 = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// let socket_v6 = SocketAddr::new(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)), 8080); + /// + /// assert_eq!(SocketAddr::parse_ascii(b"127.0.0.1:8080"), Ok(socket_v4)); + /// assert_eq!(SocketAddr::parse_ascii(b"[::1]:8080"), Ok(socket_v6)); + /// ``` + #[unstable(feature = "addr_parse_ascii", issue = "101035")] + pub fn parse_ascii(b: &[u8]) -> Result<Self, AddrParseError> { + Parser::new(b).parse_with(|p| p.read_socket_addr(), AddrKind::Socket) } } @@ -322,7 +434,7 @@ impl FromStr for SocketAddrV6 { impl FromStr for SocketAddr { type Err = AddrParseError; fn from_str(s: &str) -> Result<SocketAddr, AddrParseError> { - Parser::new(s).parse_with(|p| p.read_socket_addr(), AddrKind::Socket) + Self::parse_ascii(s.as_bytes()) } } diff --git a/library/std/src/net/socket_addr.rs b/library/std/src/net/socket_addr.rs new file mode 100644 index 000000000..33b0dfa03 --- /dev/null +++ b/library/std/src/net/socket_addr.rs @@ -0,0 +1,974 @@ +#[cfg(all(test, not(target_os = "emscripten")))] +mod tests; + +use crate::cmp::Ordering; +use crate::fmt::{self, Write}; +use crate::hash; +use crate::io; +use crate::iter; +use crate::mem; +use crate::net::{IpAddr, Ipv4Addr, Ipv6Addr}; +use crate::option; +use crate::slice; +use crate::sys::net::netc as c; +use crate::sys_common::net::LookupHost; +use crate::sys_common::{FromInner, IntoInner}; +use crate::vec; + +use super::display_buffer::DisplayBuffer; + +/// An internet socket address, either IPv4 or IPv6. +/// +/// Internet socket addresses consist of an [IP address], a 16-bit port number, as well +/// as possibly some version-dependent additional information. See [`SocketAddrV4`]'s and +/// [`SocketAddrV6`]'s respective documentation for more details. +/// +/// The size of a `SocketAddr` instance may vary depending on the target operating +/// system. +/// +/// [IP address]: IpAddr +/// +/// # Examples +/// +/// ``` +/// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; +/// +/// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); +/// +/// assert_eq!("127.0.0.1:8080".parse(), Ok(socket)); +/// assert_eq!(socket.port(), 8080); +/// assert_eq!(socket.is_ipv4(), true); +/// ``` +#[derive(Copy, Clone, PartialEq, Eq, Hash, PartialOrd, Ord)] +#[stable(feature = "rust1", since = "1.0.0")] +pub enum SocketAddr { + /// An IPv4 socket address. + #[stable(feature = "rust1", since = "1.0.0")] + V4(#[stable(feature = "rust1", since = "1.0.0")] SocketAddrV4), + /// An IPv6 socket address. + #[stable(feature = "rust1", since = "1.0.0")] + V6(#[stable(feature = "rust1", since = "1.0.0")] SocketAddrV6), +} + +/// An IPv4 socket address. +/// +/// IPv4 socket addresses consist of an [`IPv4` address] and a 16-bit port number, as +/// stated in [IETF RFC 793]. +/// +/// See [`SocketAddr`] for a type encompassing both IPv4 and IPv6 socket addresses. +/// +/// The size of a `SocketAddrV4` struct may vary depending on the target operating +/// system. Do not assume that this type has the same memory layout as the underlying +/// system representation. +/// +/// [IETF RFC 793]: https://tools.ietf.org/html/rfc793 +/// [`IPv4` address]: Ipv4Addr +/// +/// # Examples +/// +/// ``` +/// use std::net::{Ipv4Addr, SocketAddrV4}; +/// +/// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); +/// +/// assert_eq!("127.0.0.1:8080".parse(), Ok(socket)); +/// assert_eq!(socket.ip(), &Ipv4Addr::new(127, 0, 0, 1)); +/// assert_eq!(socket.port(), 8080); +/// ``` +#[derive(Copy, Clone, Eq, PartialEq)] +#[stable(feature = "rust1", since = "1.0.0")] +pub struct SocketAddrV4 { + ip: Ipv4Addr, + port: u16, +} + +/// An IPv6 socket address. +/// +/// IPv6 socket addresses consist of an [`IPv6` address], a 16-bit port number, as well +/// as fields containing the traffic class, the flow label, and a scope identifier +/// (see [IETF RFC 2553, Section 3.3] for more details). +/// +/// See [`SocketAddr`] for a type encompassing both IPv4 and IPv6 socket addresses. +/// +/// The size of a `SocketAddrV6` struct may vary depending on the target operating +/// system. Do not assume that this type has the same memory layout as the underlying +/// system representation. +/// +/// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 +/// [`IPv6` address]: Ipv6Addr +/// +/// # Examples +/// +/// ``` +/// use std::net::{Ipv6Addr, SocketAddrV6}; +/// +/// let socket = SocketAddrV6::new(Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 1), 8080, 0, 0); +/// +/// assert_eq!("[2001:db8::1]:8080".parse(), Ok(socket)); +/// assert_eq!(socket.ip(), &Ipv6Addr::new(0x2001, 0xdb8, 0, 0, 0, 0, 0, 1)); +/// assert_eq!(socket.port(), 8080); +/// ``` +#[derive(Copy, Clone, Eq, PartialEq)] +#[stable(feature = "rust1", since = "1.0.0")] +pub struct SocketAddrV6 { + ip: Ipv6Addr, + port: u16, + flowinfo: u32, + scope_id: u32, +} + +impl SocketAddr { + /// Creates a new socket address from an [IP address] and a port number. + /// + /// [IP address]: IpAddr + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; + /// + /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// assert_eq!(socket.ip(), IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1))); + /// assert_eq!(socket.port(), 8080); + /// ``` + #[stable(feature = "ip_addr", since = "1.7.0")] + #[must_use] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn new(ip: IpAddr, port: u16) -> SocketAddr { + match ip { + IpAddr::V4(a) => SocketAddr::V4(SocketAddrV4::new(a, port)), + IpAddr::V6(a) => SocketAddr::V6(SocketAddrV6::new(a, port, 0, 0)), + } + } + + /// Returns the IP address associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; + /// + /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// assert_eq!(socket.ip(), IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1))); + /// ``` + #[must_use] + #[stable(feature = "ip_addr", since = "1.7.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn ip(&self) -> IpAddr { + match *self { + SocketAddr::V4(ref a) => IpAddr::V4(*a.ip()), + SocketAddr::V6(ref a) => IpAddr::V6(*a.ip()), + } + } + + /// Changes the IP address associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; + /// + /// let mut socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// socket.set_ip(IpAddr::V4(Ipv4Addr::new(10, 10, 0, 1))); + /// assert_eq!(socket.ip(), IpAddr::V4(Ipv4Addr::new(10, 10, 0, 1))); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_ip(&mut self, new_ip: IpAddr) { + // `match (*self, new_ip)` would have us mutate a copy of self only to throw it away. + match (self, new_ip) { + (&mut SocketAddr::V4(ref mut a), IpAddr::V4(new_ip)) => a.set_ip(new_ip), + (&mut SocketAddr::V6(ref mut a), IpAddr::V6(new_ip)) => a.set_ip(new_ip), + (self_, new_ip) => *self_ = Self::new(new_ip, self_.port()), + } + } + + /// Returns the port number associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; + /// + /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// assert_eq!(socket.port(), 8080); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn port(&self) -> u16 { + match *self { + SocketAddr::V4(ref a) => a.port(), + SocketAddr::V6(ref a) => a.port(), + } + } + + /// Changes the port number associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; + /// + /// let mut socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// socket.set_port(1025); + /// assert_eq!(socket.port(), 1025); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_port(&mut self, new_port: u16) { + match *self { + SocketAddr::V4(ref mut a) => a.set_port(new_port), + SocketAddr::V6(ref mut a) => a.set_port(new_port), + } + } + + /// Returns [`true`] if the [IP address] in this `SocketAddr` is an + /// [`IPv4` address], and [`false`] otherwise. + /// + /// [IP address]: IpAddr + /// [`IPv4` address]: IpAddr::V4 + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv4Addr, SocketAddr}; + /// + /// let socket = SocketAddr::new(IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1)), 8080); + /// assert_eq!(socket.is_ipv4(), true); + /// assert_eq!(socket.is_ipv6(), false); + /// ``` + #[must_use] + #[stable(feature = "sockaddr_checker", since = "1.16.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn is_ipv4(&self) -> bool { + matches!(*self, SocketAddr::V4(_)) + } + + /// Returns [`true`] if the [IP address] in this `SocketAddr` is an + /// [`IPv6` address], and [`false`] otherwise. + /// + /// [IP address]: IpAddr + /// [`IPv6` address]: IpAddr::V6 + /// + /// # Examples + /// + /// ``` + /// use std::net::{IpAddr, Ipv6Addr, SocketAddr}; + /// + /// let socket = SocketAddr::new(IpAddr::V6(Ipv6Addr::new(0, 0, 0, 0, 0, 65535, 0, 1)), 8080); + /// assert_eq!(socket.is_ipv4(), false); + /// assert_eq!(socket.is_ipv6(), true); + /// ``` + #[must_use] + #[stable(feature = "sockaddr_checker", since = "1.16.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn is_ipv6(&self) -> bool { + matches!(*self, SocketAddr::V6(_)) + } +} + +impl SocketAddrV4 { + /// Creates a new socket address from an [`IPv4` address] and a port number. + /// + /// [`IPv4` address]: Ipv4Addr + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV4, Ipv4Addr}; + /// + /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); + /// ``` + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn new(ip: Ipv4Addr, port: u16) -> SocketAddrV4 { + SocketAddrV4 { ip, port } + } + + /// Returns the IP address associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV4, Ipv4Addr}; + /// + /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); + /// assert_eq!(socket.ip(), &Ipv4Addr::new(127, 0, 0, 1)); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn ip(&self) -> &Ipv4Addr { + &self.ip + } + + /// Changes the IP address associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV4, Ipv4Addr}; + /// + /// let mut socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); + /// socket.set_ip(Ipv4Addr::new(192, 168, 0, 1)); + /// assert_eq!(socket.ip(), &Ipv4Addr::new(192, 168, 0, 1)); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_ip(&mut self, new_ip: Ipv4Addr) { + self.ip = new_ip; + } + + /// Returns the port number associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV4, Ipv4Addr}; + /// + /// let socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); + /// assert_eq!(socket.port(), 8080); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn port(&self) -> u16 { + self.port + } + + /// Changes the port number associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV4, Ipv4Addr}; + /// + /// let mut socket = SocketAddrV4::new(Ipv4Addr::new(127, 0, 0, 1), 8080); + /// socket.set_port(4242); + /// assert_eq!(socket.port(), 4242); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_port(&mut self, new_port: u16) { + self.port = new_port; + } +} + +impl SocketAddrV6 { + /// Creates a new socket address from an [`IPv6` address], a 16-bit port number, + /// and the `flowinfo` and `scope_id` fields. + /// + /// For more information on the meaning and layout of the `flowinfo` and `scope_id` + /// parameters, see [IETF RFC 2553, Section 3.3]. + /// + /// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 + /// [`IPv6` address]: Ipv6Addr + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); + /// ``` + #[stable(feature = "rust1", since = "1.0.0")] + #[must_use] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn new(ip: Ipv6Addr, port: u16, flowinfo: u32, scope_id: u32) -> SocketAddrV6 { + SocketAddrV6 { ip, port, flowinfo, scope_id } + } + + /// Returns the IP address associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); + /// assert_eq!(socket.ip(), &Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1)); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn ip(&self) -> &Ipv6Addr { + &self.ip + } + + /// Changes the IP address associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); + /// socket.set_ip(Ipv6Addr::new(76, 45, 0, 0, 0, 0, 0, 0)); + /// assert_eq!(socket.ip(), &Ipv6Addr::new(76, 45, 0, 0, 0, 0, 0, 0)); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_ip(&mut self, new_ip: Ipv6Addr) { + self.ip = new_ip; + } + + /// Returns the port number associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); + /// assert_eq!(socket.port(), 8080); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn port(&self) -> u16 { + self.port + } + + /// Changes the port number associated with this socket address. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 0); + /// socket.set_port(4242); + /// assert_eq!(socket.port(), 4242); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_port(&mut self, new_port: u16) { + self.port = new_port; + } + + /// Returns the flow information associated with this address. + /// + /// This information corresponds to the `sin6_flowinfo` field in C's `netinet/in.h`, + /// as specified in [IETF RFC 2553, Section 3.3]. + /// It combines information about the flow label and the traffic class as specified + /// in [IETF RFC 2460], respectively [Section 6] and [Section 7]. + /// + /// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 + /// [IETF RFC 2460]: https://tools.ietf.org/html/rfc2460 + /// [Section 6]: https://tools.ietf.org/html/rfc2460#section-6 + /// [Section 7]: https://tools.ietf.org/html/rfc2460#section-7 + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 10, 0); + /// assert_eq!(socket.flowinfo(), 10); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn flowinfo(&self) -> u32 { + self.flowinfo + } + + /// Changes the flow information associated with this socket address. + /// + /// See [`SocketAddrV6::flowinfo`]'s documentation for more details. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 10, 0); + /// socket.set_flowinfo(56); + /// assert_eq!(socket.flowinfo(), 56); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_flowinfo(&mut self, new_flowinfo: u32) { + self.flowinfo = new_flowinfo; + } + + /// Returns the scope ID associated with this address. + /// + /// This information corresponds to the `sin6_scope_id` field in C's `netinet/in.h`, + /// as specified in [IETF RFC 2553, Section 3.3]. + /// + /// [IETF RFC 2553, Section 3.3]: https://tools.ietf.org/html/rfc2553#section-3.3 + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 78); + /// assert_eq!(socket.scope_id(), 78); + /// ``` + #[must_use] + #[stable(feature = "rust1", since = "1.0.0")] + #[rustc_const_unstable(feature = "const_socketaddr", issue = "82485")] + pub const fn scope_id(&self) -> u32 { + self.scope_id + } + + /// Changes the scope ID associated with this socket address. + /// + /// See [`SocketAddrV6::scope_id`]'s documentation for more details. + /// + /// # Examples + /// + /// ``` + /// use std::net::{SocketAddrV6, Ipv6Addr}; + /// + /// let mut socket = SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0, 1), 8080, 0, 78); + /// socket.set_scope_id(42); + /// assert_eq!(socket.scope_id(), 42); + /// ``` + #[stable(feature = "sockaddr_setters", since = "1.9.0")] + pub fn set_scope_id(&mut self, new_scope_id: u32) { + self.scope_id = new_scope_id; + } +} + +impl FromInner<c::sockaddr_in> for SocketAddrV4 { + fn from_inner(addr: c::sockaddr_in) -> SocketAddrV4 { + SocketAddrV4 { ip: Ipv4Addr::from_inner(addr.sin_addr), port: u16::from_be(addr.sin_port) } + } +} + +impl FromInner<c::sockaddr_in6> for SocketAddrV6 { + fn from_inner(addr: c::sockaddr_in6) -> SocketAddrV6 { + SocketAddrV6 { + ip: Ipv6Addr::from_inner(addr.sin6_addr), + port: u16::from_be(addr.sin6_port), + flowinfo: addr.sin6_flowinfo, + scope_id: addr.sin6_scope_id, + } + } +} + +impl IntoInner<c::sockaddr_in> for SocketAddrV4 { + fn into_inner(self) -> c::sockaddr_in { + c::sockaddr_in { + sin_family: c::AF_INET as c::sa_family_t, + sin_port: self.port.to_be(), + sin_addr: self.ip.into_inner(), + ..unsafe { mem::zeroed() } + } + } +} + +impl IntoInner<c::sockaddr_in6> for SocketAddrV6 { + fn into_inner(self) -> c::sockaddr_in6 { + c::sockaddr_in6 { + sin6_family: c::AF_INET6 as c::sa_family_t, + sin6_port: self.port.to_be(), + sin6_addr: self.ip.into_inner(), + sin6_flowinfo: self.flowinfo, + sin6_scope_id: self.scope_id, + ..unsafe { mem::zeroed() } + } + } +} + +#[stable(feature = "ip_from_ip", since = "1.16.0")] +impl From<SocketAddrV4> for SocketAddr { + /// Converts a [`SocketAddrV4`] into a [`SocketAddr::V4`]. + fn from(sock4: SocketAddrV4) -> SocketAddr { + SocketAddr::V4(sock4) + } +} + +#[stable(feature = "ip_from_ip", since = "1.16.0")] +impl From<SocketAddrV6> for SocketAddr { + /// Converts a [`SocketAddrV6`] into a [`SocketAddr::V6`]. + fn from(sock6: SocketAddrV6) -> SocketAddr { + SocketAddr::V6(sock6) + } +} + +#[stable(feature = "addr_from_into_ip", since = "1.17.0")] +impl<I: Into<IpAddr>> From<(I, u16)> for SocketAddr { + /// Converts a tuple struct (Into<[`IpAddr`]>, `u16`) into a [`SocketAddr`]. + /// + /// This conversion creates a [`SocketAddr::V4`] for an [`IpAddr::V4`] + /// and creates a [`SocketAddr::V6`] for an [`IpAddr::V6`]. + /// + /// `u16` is treated as port of the newly created [`SocketAddr`]. + fn from(pieces: (I, u16)) -> SocketAddr { + SocketAddr::new(pieces.0.into(), pieces.1) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Display for SocketAddr { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + match *self { + SocketAddr::V4(ref a) => a.fmt(f), + SocketAddr::V6(ref a) => a.fmt(f), + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Debug for SocketAddr { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self, fmt) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Display for SocketAddrV4 { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + // If there are no alignment requirements, write the socket address directly to `f`. + // Otherwise, write it to a local buffer and then use `f.pad`. + if f.precision().is_none() && f.width().is_none() { + write!(f, "{}:{}", self.ip(), self.port()) + } else { + const LONGEST_IPV4_SOCKET_ADDR: &str = "255.255.255.255:65536"; + + let mut buf = DisplayBuffer::<{ LONGEST_IPV4_SOCKET_ADDR.len() }>::new(); + // Buffer is long enough for the longest possible IPv4 socket address, so this should never fail. + write!(buf, "{}:{}", self.ip(), self.port()).unwrap(); + + f.pad(buf.as_str()) + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Debug for SocketAddrV4 { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self, fmt) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Display for SocketAddrV6 { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + // If there are no alignment requirements, write the socket address directly to `f`. + // Otherwise, write it to a local buffer and then use `f.pad`. + if f.precision().is_none() && f.width().is_none() { + match self.scope_id() { + 0 => write!(f, "[{}]:{}", self.ip(), self.port()), + scope_id => write!(f, "[{}%{}]:{}", self.ip(), scope_id, self.port()), + } + } else { + const LONGEST_IPV6_SOCKET_ADDR: &str = + "[ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff%4294967296]:65536"; + + let mut buf = DisplayBuffer::<{ LONGEST_IPV6_SOCKET_ADDR.len() }>::new(); + match self.scope_id() { + 0 => write!(buf, "[{}]:{}", self.ip(), self.port()), + scope_id => write!(buf, "[{}%{}]:{}", self.ip(), scope_id, self.port()), + } + // Buffer is long enough for the longest possible IPv6 socket address, so this should never fail. + .unwrap(); + + f.pad(buf.as_str()) + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl fmt::Debug for SocketAddrV6 { + fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Display::fmt(self, fmt) + } +} + +#[stable(feature = "socketaddr_ordering", since = "1.45.0")] +impl PartialOrd for SocketAddrV4 { + fn partial_cmp(&self, other: &SocketAddrV4) -> Option<Ordering> { + Some(self.cmp(other)) + } +} + +#[stable(feature = "socketaddr_ordering", since = "1.45.0")] +impl PartialOrd for SocketAddrV6 { + fn partial_cmp(&self, other: &SocketAddrV6) -> Option<Ordering> { + Some(self.cmp(other)) + } +} + +#[stable(feature = "socketaddr_ordering", since = "1.45.0")] +impl Ord for SocketAddrV4 { + fn cmp(&self, other: &SocketAddrV4) -> Ordering { + self.ip().cmp(other.ip()).then(self.port().cmp(&other.port())) + } +} + +#[stable(feature = "socketaddr_ordering", since = "1.45.0")] +impl Ord for SocketAddrV6 { + fn cmp(&self, other: &SocketAddrV6) -> Ordering { + self.ip().cmp(other.ip()).then(self.port().cmp(&other.port())) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl hash::Hash for SocketAddrV4 { + fn hash<H: hash::Hasher>(&self, s: &mut H) { + (self.port, self.ip).hash(s) + } +} +#[stable(feature = "rust1", since = "1.0.0")] +impl hash::Hash for SocketAddrV6 { + fn hash<H: hash::Hasher>(&self, s: &mut H) { + (self.port, &self.ip, self.flowinfo, self.scope_id).hash(s) + } +} + +/// A trait for objects which can be converted or resolved to one or more +/// [`SocketAddr`] values. +/// +/// This trait is used for generic address resolution when constructing network +/// objects. By default it is implemented for the following types: +/// +/// * [`SocketAddr`]: [`to_socket_addrs`] is the identity function. +/// +/// * [`SocketAddrV4`], [`SocketAddrV6`], <code>([IpAddr], [u16])</code>, +/// <code>([Ipv4Addr], [u16])</code>, <code>([Ipv6Addr], [u16])</code>: +/// [`to_socket_addrs`] constructs a [`SocketAddr`] trivially. +/// +/// * <code>(&[str], [u16])</code>: <code>&[str]</code> should be either a string representation +/// of an [`IpAddr`] address as expected by [`FromStr`] implementation or a host +/// name. [`u16`] is the port number. +/// +/// * <code>&[str]</code>: the string should be either a string representation of a +/// [`SocketAddr`] as expected by its [`FromStr`] implementation or a string like +/// `<host_name>:<port>` pair where `<port>` is a [`u16`] value. +/// +/// This trait allows constructing network objects like [`TcpStream`] or +/// [`UdpSocket`] easily with values of various types for the bind/connection +/// address. It is needed because sometimes one type is more appropriate than +/// the other: for simple uses a string like `"localhost:12345"` is much nicer +/// than manual construction of the corresponding [`SocketAddr`], but sometimes +/// [`SocketAddr`] value is *the* main source of the address, and converting it to +/// some other type (e.g., a string) just for it to be converted back to +/// [`SocketAddr`] in constructor methods is pointless. +/// +/// Addresses returned by the operating system that are not IP addresses are +/// silently ignored. +/// +/// [`FromStr`]: crate::str::FromStr "std::str::FromStr" +/// [`TcpStream`]: crate::net::TcpStream "net::TcpStream" +/// [`to_socket_addrs`]: ToSocketAddrs::to_socket_addrs +/// [`UdpSocket`]: crate::net::UdpSocket "net::UdpSocket" +/// +/// # Examples +/// +/// Creating a [`SocketAddr`] iterator that yields one item: +/// +/// ``` +/// use std::net::{ToSocketAddrs, SocketAddr}; +/// +/// let addr = SocketAddr::from(([127, 0, 0, 1], 443)); +/// let mut addrs_iter = addr.to_socket_addrs().unwrap(); +/// +/// assert_eq!(Some(addr), addrs_iter.next()); +/// assert!(addrs_iter.next().is_none()); +/// ``` +/// +/// Creating a [`SocketAddr`] iterator from a hostname: +/// +/// ```no_run +/// use std::net::{SocketAddr, ToSocketAddrs}; +/// +/// // assuming 'localhost' resolves to 127.0.0.1 +/// let mut addrs_iter = "localhost:443".to_socket_addrs().unwrap(); +/// assert_eq!(addrs_iter.next(), Some(SocketAddr::from(([127, 0, 0, 1], 443)))); +/// assert!(addrs_iter.next().is_none()); +/// +/// // assuming 'foo' does not resolve +/// assert!("foo:443".to_socket_addrs().is_err()); +/// ``` +/// +/// Creating a [`SocketAddr`] iterator that yields multiple items: +/// +/// ``` +/// use std::net::{SocketAddr, ToSocketAddrs}; +/// +/// let addr1 = SocketAddr::from(([0, 0, 0, 0], 80)); +/// let addr2 = SocketAddr::from(([127, 0, 0, 1], 443)); +/// let addrs = vec![addr1, addr2]; +/// +/// let mut addrs_iter = (&addrs[..]).to_socket_addrs().unwrap(); +/// +/// assert_eq!(Some(addr1), addrs_iter.next()); +/// assert_eq!(Some(addr2), addrs_iter.next()); +/// assert!(addrs_iter.next().is_none()); +/// ``` +/// +/// Attempting to create a [`SocketAddr`] iterator from an improperly formatted +/// socket address `&str` (missing the port): +/// +/// ``` +/// use std::io; +/// use std::net::ToSocketAddrs; +/// +/// let err = "127.0.0.1".to_socket_addrs().unwrap_err(); +/// assert_eq!(err.kind(), io::ErrorKind::InvalidInput); +/// ``` +/// +/// [`TcpStream::connect`] is an example of an function that utilizes +/// `ToSocketAddrs` as a trait bound on its parameter in order to accept +/// different types: +/// +/// ```no_run +/// use std::net::{TcpStream, Ipv4Addr}; +/// +/// let stream = TcpStream::connect(("127.0.0.1", 443)); +/// // or +/// let stream = TcpStream::connect("127.0.0.1:443"); +/// // or +/// let stream = TcpStream::connect((Ipv4Addr::new(127, 0, 0, 1), 443)); +/// ``` +/// +/// [`TcpStream::connect`]: crate::net::TcpStream::connect +#[stable(feature = "rust1", since = "1.0.0")] +pub trait ToSocketAddrs { + /// Returned iterator over socket addresses which this type may correspond + /// to. + #[stable(feature = "rust1", since = "1.0.0")] + type Iter: Iterator<Item = SocketAddr>; + + /// Converts this object to an iterator of resolved [`SocketAddr`]s. + /// + /// The returned iterator might not actually yield any values depending on the + /// outcome of any resolution performed. + /// + /// Note that this function may block the current thread while resolution is + /// performed. + #[stable(feature = "rust1", since = "1.0.0")] + fn to_socket_addrs(&self) -> io::Result<Self::Iter>; +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for SocketAddr { + type Iter = option::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { + Ok(Some(*self).into_iter()) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for SocketAddrV4 { + type Iter = option::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { + SocketAddr::V4(*self).to_socket_addrs() + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for SocketAddrV6 { + type Iter = option::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { + SocketAddr::V6(*self).to_socket_addrs() + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for (IpAddr, u16) { + type Iter = option::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { + let (ip, port) = *self; + match ip { + IpAddr::V4(ref a) => (*a, port).to_socket_addrs(), + IpAddr::V6(ref a) => (*a, port).to_socket_addrs(), + } + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for (Ipv4Addr, u16) { + type Iter = option::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { + let (ip, port) = *self; + SocketAddrV4::new(ip, port).to_socket_addrs() + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for (Ipv6Addr, u16) { + type Iter = option::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<option::IntoIter<SocketAddr>> { + let (ip, port) = *self; + SocketAddrV6::new(ip, port, 0, 0).to_socket_addrs() + } +} + +fn resolve_socket_addr(lh: LookupHost) -> io::Result<vec::IntoIter<SocketAddr>> { + let p = lh.port(); + let v: Vec<_> = lh + .map(|mut a| { + a.set_port(p); + a + }) + .collect(); + Ok(v.into_iter()) +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for (&str, u16) { + type Iter = vec::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { + let (host, port) = *self; + + // try to parse the host as a regular IP address first + if let Ok(addr) = host.parse::<Ipv4Addr>() { + let addr = SocketAddrV4::new(addr, port); + return Ok(vec![SocketAddr::V4(addr)].into_iter()); + } + if let Ok(addr) = host.parse::<Ipv6Addr>() { + let addr = SocketAddrV6::new(addr, port, 0, 0); + return Ok(vec![SocketAddr::V6(addr)].into_iter()); + } + + resolve_socket_addr((host, port).try_into()?) + } +} + +#[stable(feature = "string_u16_to_socket_addrs", since = "1.46.0")] +impl ToSocketAddrs for (String, u16) { + type Iter = vec::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { + (&*self.0, self.1).to_socket_addrs() + } +} + +// accepts strings like 'localhost:12345' +#[stable(feature = "rust1", since = "1.0.0")] +impl ToSocketAddrs for str { + type Iter = vec::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { + // try to parse as a regular SocketAddr first + if let Ok(addr) = self.parse() { + return Ok(vec![addr].into_iter()); + } + + resolve_socket_addr(self.try_into()?) + } +} + +#[stable(feature = "slice_to_socket_addrs", since = "1.8.0")] +impl<'a> ToSocketAddrs for &'a [SocketAddr] { + type Iter = iter::Cloned<slice::Iter<'a, SocketAddr>>; + + fn to_socket_addrs(&self) -> io::Result<Self::Iter> { + Ok(self.iter().cloned()) + } +} + +#[stable(feature = "rust1", since = "1.0.0")] +impl<T: ToSocketAddrs + ?Sized> ToSocketAddrs for &T { + type Iter = T::Iter; + fn to_socket_addrs(&self) -> io::Result<T::Iter> { + (**self).to_socket_addrs() + } +} + +#[stable(feature = "string_to_socket_addrs", since = "1.16.0")] +impl ToSocketAddrs for String { + type Iter = vec::IntoIter<SocketAddr>; + fn to_socket_addrs(&self) -> io::Result<vec::IntoIter<SocketAddr>> { + (&**self).to_socket_addrs() + } +} diff --git a/library/std/src/net/socket_addr/tests.rs b/library/std/src/net/socket_addr/tests.rs new file mode 100644 index 000000000..15211f819 --- /dev/null +++ b/library/std/src/net/socket_addr/tests.rs @@ -0,0 +1,306 @@ +use crate::net::test::{sa4, sa6, tsa}; +use crate::net::*; + +#[test] +fn to_socket_addr_ipaddr_u16() { + let a = Ipv4Addr::new(77, 88, 21, 11); + let p = 12345; + let e = SocketAddr::V4(SocketAddrV4::new(a, p)); + assert_eq!(Ok(vec![e]), tsa((a, p))); +} + +#[test] +fn to_socket_addr_str_u16() { + let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 24352); + assert_eq!(Ok(vec![a]), tsa(("77.88.21.11", 24352))); + + let a = sa6(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53); + assert_eq!(Ok(vec![a]), tsa(("2a02:6b8:0:1::1", 53))); + + let a = sa4(Ipv4Addr::new(127, 0, 0, 1), 23924); + #[cfg(not(target_env = "sgx"))] + assert!(tsa(("localhost", 23924)).unwrap().contains(&a)); + #[cfg(target_env = "sgx")] + let _ = a; +} + +#[test] +fn to_socket_addr_str() { + let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 24352); + assert_eq!(Ok(vec![a]), tsa("77.88.21.11:24352")); + + let a = sa6(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53); + assert_eq!(Ok(vec![a]), tsa("[2a02:6b8:0:1::1]:53")); + + let a = sa4(Ipv4Addr::new(127, 0, 0, 1), 23924); + #[cfg(not(target_env = "sgx"))] + assert!(tsa("localhost:23924").unwrap().contains(&a)); + #[cfg(target_env = "sgx")] + let _ = a; +} + +#[test] +fn to_socket_addr_string() { + let a = sa4(Ipv4Addr::new(77, 88, 21, 11), 24352); + assert_eq!(Ok(vec![a]), tsa(&*format!("{}:{}", "77.88.21.11", "24352"))); + assert_eq!(Ok(vec![a]), tsa(&format!("{}:{}", "77.88.21.11", "24352"))); + assert_eq!(Ok(vec![a]), tsa(format!("{}:{}", "77.88.21.11", "24352"))); + + let s = format!("{}:{}", "77.88.21.11", "24352"); + assert_eq!(Ok(vec![a]), tsa(s)); + // s has been moved into the tsa call +} + +#[test] +fn ipv4_socket_addr_to_string() { + // Shortest possible IPv4 length. + assert_eq!(SocketAddrV4::new(Ipv4Addr::new(0, 0, 0, 0), 0).to_string(), "0.0.0.0:0"); + + // Longest possible IPv4 length. + assert_eq!( + SocketAddrV4::new(Ipv4Addr::new(255, 255, 255, 255), u16::MAX).to_string(), + "255.255.255.255:65535" + ); + + // Test padding. + assert_eq!( + &format!("{:16}", SocketAddrV4::new(Ipv4Addr::new(1, 1, 1, 1), 53)), + "1.1.1.1:53 " + ); + assert_eq!( + &format!("{:>16}", SocketAddrV4::new(Ipv4Addr::new(1, 1, 1, 1), 53)), + " 1.1.1.1:53" + ); +} + +#[test] +fn ipv6_socket_addr_to_string() { + // IPv4-mapped address. + assert_eq!( + SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0xffff, 0xc000, 0x280), 8080, 0, 0) + .to_string(), + "[::ffff:192.0.2.128]:8080" + ); + + // IPv4-compatible address. + assert_eq!( + SocketAddrV6::new(Ipv6Addr::new(0, 0, 0, 0, 0, 0, 0xc000, 0x280), 8080, 0, 0).to_string(), + "[::192.0.2.128]:8080" + ); + + // IPv6 address with no zero segments. + assert_eq!( + SocketAddrV6::new(Ipv6Addr::new(8, 9, 10, 11, 12, 13, 14, 15), 80, 0, 0).to_string(), + "[8:9:a:b:c:d:e:f]:80" + ); + + // Shortest possible IPv6 length. + assert_eq!(SocketAddrV6::new(Ipv6Addr::UNSPECIFIED, 0, 0, 0).to_string(), "[::]:0"); + + // Longest possible IPv6 length. + assert_eq!( + SocketAddrV6::new( + Ipv6Addr::new(0x1111, 0x2222, 0x3333, 0x4444, 0x5555, 0x6666, 0x7777, 0x8888), + u16::MAX, + u32::MAX, + u32::MAX, + ) + .to_string(), + "[1111:2222:3333:4444:5555:6666:7777:8888%4294967295]:65535" + ); + + // Test padding. + assert_eq!( + &format!("{:22}", SocketAddrV6::new(Ipv6Addr::new(1, 2, 3, 4, 5, 6, 7, 8), 9, 0, 0)), + "[1:2:3:4:5:6:7:8]:9 " + ); + assert_eq!( + &format!("{:>22}", SocketAddrV6::new(Ipv6Addr::new(1, 2, 3, 4, 5, 6, 7, 8), 9, 0, 0)), + " [1:2:3:4:5:6:7:8]:9" + ); +} + +#[test] +fn bind_udp_socket_bad() { + // rust-lang/rust#53957: This is a regression test for a parsing problem + // discovered as part of issue rust-lang/rust#23076, where we were + // incorrectly parsing invalid input and then that would result in a + // successful `UdpSocket` binding when we would expect failure. + // + // At one time, this test was written as a call to `tsa` with + // INPUT_23076. However, that structure yields an unreliable test, + // because it ends up passing junk input to the DNS server, and some DNS + // servers will respond with `Ok` to such input, with the ip address of + // the DNS server itself. + // + // This form of the test is more robust: even when the DNS server + // returns its own address, it is still an error to bind a UDP socket to + // a non-local address, and so we still get an error here in that case. + + const INPUT_23076: &str = "1200::AB00:1234::2552:7777:1313:34300"; + + assert!(crate::net::UdpSocket::bind(INPUT_23076).is_err()) +} + +#[test] +fn set_ip() { + fn ip4(low: u8) -> Ipv4Addr { + Ipv4Addr::new(77, 88, 21, low) + } + fn ip6(low: u16) -> Ipv6Addr { + Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, low) + } + + let mut v4 = SocketAddrV4::new(ip4(11), 80); + assert_eq!(v4.ip(), &ip4(11)); + v4.set_ip(ip4(12)); + assert_eq!(v4.ip(), &ip4(12)); + + let mut addr = SocketAddr::V4(v4); + assert_eq!(addr.ip(), IpAddr::V4(ip4(12))); + addr.set_ip(IpAddr::V4(ip4(13))); + assert_eq!(addr.ip(), IpAddr::V4(ip4(13))); + addr.set_ip(IpAddr::V6(ip6(14))); + assert_eq!(addr.ip(), IpAddr::V6(ip6(14))); + + let mut v6 = SocketAddrV6::new(ip6(1), 80, 0, 0); + assert_eq!(v6.ip(), &ip6(1)); + v6.set_ip(ip6(2)); + assert_eq!(v6.ip(), &ip6(2)); + + let mut addr = SocketAddr::V6(v6); + assert_eq!(addr.ip(), IpAddr::V6(ip6(2))); + addr.set_ip(IpAddr::V6(ip6(3))); + assert_eq!(addr.ip(), IpAddr::V6(ip6(3))); + addr.set_ip(IpAddr::V4(ip4(4))); + assert_eq!(addr.ip(), IpAddr::V4(ip4(4))); +} + +#[test] +fn set_port() { + let mut v4 = SocketAddrV4::new(Ipv4Addr::new(77, 88, 21, 11), 80); + assert_eq!(v4.port(), 80); + v4.set_port(443); + assert_eq!(v4.port(), 443); + + let mut addr = SocketAddr::V4(v4); + assert_eq!(addr.port(), 443); + addr.set_port(8080); + assert_eq!(addr.port(), 8080); + + let mut v6 = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 80, 0, 0); + assert_eq!(v6.port(), 80); + v6.set_port(443); + assert_eq!(v6.port(), 443); + + let mut addr = SocketAddr::V6(v6); + assert_eq!(addr.port(), 443); + addr.set_port(8080); + assert_eq!(addr.port(), 8080); +} + +#[test] +fn set_flowinfo() { + let mut v6 = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 80, 10, 0); + assert_eq!(v6.flowinfo(), 10); + v6.set_flowinfo(20); + assert_eq!(v6.flowinfo(), 20); +} + +#[test] +fn set_scope_id() { + let mut v6 = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 80, 0, 10); + assert_eq!(v6.scope_id(), 10); + v6.set_scope_id(20); + assert_eq!(v6.scope_id(), 20); +} + +#[test] +fn is_v4() { + let v4 = SocketAddr::V4(SocketAddrV4::new(Ipv4Addr::new(77, 88, 21, 11), 80)); + assert!(v4.is_ipv4()); + assert!(!v4.is_ipv6()); +} + +#[test] +fn is_v6() { + let v6 = SocketAddr::V6(SocketAddrV6::new( + Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), + 80, + 10, + 0, + )); + assert!(!v6.is_ipv4()); + assert!(v6.is_ipv6()); +} + +#[test] +fn socket_v4_to_str() { + let socket = SocketAddrV4::new(Ipv4Addr::new(192, 168, 0, 1), 8080); + + assert_eq!(format!("{socket}"), "192.168.0.1:8080"); + assert_eq!(format!("{socket:<20}"), "192.168.0.1:8080 "); + assert_eq!(format!("{socket:>20}"), " 192.168.0.1:8080"); + assert_eq!(format!("{socket:^20}"), " 192.168.0.1:8080 "); + assert_eq!(format!("{socket:.10}"), "192.168.0."); +} + +#[test] +fn socket_v6_to_str() { + let mut socket = SocketAddrV6::new(Ipv6Addr::new(0x2a02, 0x6b8, 0, 1, 0, 0, 0, 1), 53, 0, 0); + + assert_eq!(format!("{socket}"), "[2a02:6b8:0:1::1]:53"); + assert_eq!(format!("{socket:<24}"), "[2a02:6b8:0:1::1]:53 "); + assert_eq!(format!("{socket:>24}"), " [2a02:6b8:0:1::1]:53"); + assert_eq!(format!("{socket:^24}"), " [2a02:6b8:0:1::1]:53 "); + assert_eq!(format!("{socket:.15}"), "[2a02:6b8:0:1::"); + + socket.set_scope_id(5); + + assert_eq!(format!("{socket}"), "[2a02:6b8:0:1::1%5]:53"); + assert_eq!(format!("{socket:<24}"), "[2a02:6b8:0:1::1%5]:53 "); + assert_eq!(format!("{socket:>24}"), " [2a02:6b8:0:1::1%5]:53"); + assert_eq!(format!("{socket:^24}"), " [2a02:6b8:0:1::1%5]:53 "); + assert_eq!(format!("{socket:.18}"), "[2a02:6b8:0:1::1%5"); +} + +#[test] +fn compare() { + let v4_1 = "224.120.45.1:23456".parse::<SocketAddrV4>().unwrap(); + let v4_2 = "224.210.103.5:12345".parse::<SocketAddrV4>().unwrap(); + let v4_3 = "224.210.103.5:23456".parse::<SocketAddrV4>().unwrap(); + let v6_1 = "[2001:db8:f00::1002]:23456".parse::<SocketAddrV6>().unwrap(); + let v6_2 = "[2001:db8:f00::2001]:12345".parse::<SocketAddrV6>().unwrap(); + let v6_3 = "[2001:db8:f00::2001]:23456".parse::<SocketAddrV6>().unwrap(); + + // equality + assert_eq!(v4_1, v4_1); + assert_eq!(v6_1, v6_1); + assert_eq!(SocketAddr::V4(v4_1), SocketAddr::V4(v4_1)); + assert_eq!(SocketAddr::V6(v6_1), SocketAddr::V6(v6_1)); + assert!(v4_1 != v4_2); + assert!(v6_1 != v6_2); + + // compare different addresses + assert!(v4_1 < v4_2); + assert!(v6_1 < v6_2); + assert!(v4_2 > v4_1); + assert!(v6_2 > v6_1); + + // compare the same address with different ports + assert!(v4_2 < v4_3); + assert!(v6_2 < v6_3); + assert!(v4_3 > v4_2); + assert!(v6_3 > v6_2); + + // compare different addresses with the same port + assert!(v4_1 < v4_3); + assert!(v6_1 < v6_3); + assert!(v4_3 > v4_1); + assert!(v6_3 > v6_1); + + // compare with an inferred right-hand side + assert_eq!(v4_1, "224.120.45.1:23456".parse().unwrap()); + assert_eq!(v6_1, "[2001:db8:f00::1002]:23456".parse().unwrap()); + assert_eq!(SocketAddr::V4(v4_1), "224.120.45.1:23456".parse().unwrap()); +} diff --git a/library/std/src/os/android/mod.rs b/library/std/src/os/android/mod.rs index dbb0127f3..5adcb82b6 100644 --- a/library/std/src/os/android/mod.rs +++ b/library/std/src/os/android/mod.rs @@ -3,4 +3,5 @@ #![stable(feature = "raw_ext", since = "1.1.0")] pub mod fs; +pub mod net; pub mod raw; diff --git a/library/std/src/os/android/net.rs b/library/std/src/os/android/net.rs new file mode 100644 index 000000000..ff96125c3 --- /dev/null +++ b/library/std/src/os/android/net.rs @@ -0,0 +1,4 @@ +//! Linux and Android-specific definitions for socket options. + +#![unstable(feature = "tcp_quickack", issue = "96256")] +pub use crate::os::net::tcp::TcpStreamExt; diff --git a/library/std/src/os/fd/mod.rs b/library/std/src/os/fd/mod.rs index a45694753..c6aa7c77d 100644 --- a/library/std/src/os/fd/mod.rs +++ b/library/std/src/os/fd/mod.rs @@ -1,16 +1,25 @@ //! Owned and borrowed Unix-like file descriptors. +//! +//! This module is supported on Unix platforms and WASI, which both use a +//! similar file descriptor system for referencing OS resources. #![stable(feature = "io_safety", since = "1.63.0")] #![deny(unsafe_op_in_unsafe_fn)] // `RawFd`, `AsRawFd`, etc. -pub mod raw; +mod raw; // `OwnedFd`, `AsFd`, etc. -pub mod owned; +mod owned; // Implementations for `AsRawFd` etc. for network types. mod net; #[cfg(test)] mod tests; + +// Export the types and traits for the public API. +#[unstable(feature = "os_fd", issue = "98699")] +pub use owned::*; +#[unstable(feature = "os_fd", issue = "98699")] +pub use raw::*; diff --git a/library/std/src/os/fd/owned.rs b/library/std/src/os/fd/owned.rs index a463bc41d..c16518577 100644 --- a/library/std/src/os/fd/owned.rs +++ b/library/std/src/os/fd/owned.rs @@ -6,6 +6,7 @@ use super::raw::{AsRawFd, FromRawFd, IntoRawFd, RawFd}; use crate::fmt; use crate::fs; +use crate::io; use crate::marker::PhantomData; use crate::mem::forget; #[cfg(not(any(target_arch = "wasm32", target_env = "sgx")))] @@ -104,7 +105,8 @@ impl BorrowedFd<'_> { #[cfg(target_os = "espidf")] let cmd = libc::F_DUPFD; - let fd = cvt(unsafe { libc::fcntl(self.as_raw_fd(), cmd, 0) })?; + // Avoid using file descriptors below 3 as they are used for stdio + let fd = cvt(unsafe { libc::fcntl(self.as_raw_fd(), cmd, 3) })?; Ok(unsafe { OwnedFd::from_raw_fd(fd) }) } @@ -191,6 +193,23 @@ impl fmt::Debug for OwnedFd { } } +macro_rules! impl_is_terminal { + ($($t:ty),*$(,)?) => {$( + #[unstable(feature = "sealed", issue = "none")] + impl crate::sealed::Sealed for $t {} + + #[unstable(feature = "is_terminal", issue = "98070")] + impl crate::io::IsTerminal for $t { + #[inline] + fn is_terminal(&self) -> bool { + crate::sys::io::is_terminal(self) + } + } + )*} +} + +impl_is_terminal!(BorrowedFd<'_>, OwnedFd); + /// A trait to borrow the file descriptor from an underlying object. /// /// This is only available on unix platforms and must be imported in order to @@ -205,10 +224,8 @@ pub trait AsFd { /// ```rust,no_run /// use std::fs::File; /// # use std::io; - /// # #[cfg(target_os = "wasi")] - /// # use std::os::wasi::io::{AsFd, BorrowedFd}; - /// # #[cfg(unix)] - /// # use std::os::unix::io::{AsFd, BorrowedFd}; + /// # #[cfg(any(unix, target_os = "wasi"))] + /// # use std::os::fd::{AsFd, BorrowedFd}; /// /// let mut f = File::open("foo.txt")?; /// # #[cfg(any(unix, target_os = "wasi"))] @@ -386,3 +403,54 @@ impl<T: AsFd> AsFd for Box<T> { (**self).as_fd() } } + +#[stable(feature = "io_safety", since = "1.63.0")] +impl AsFd for io::Stdin { + #[inline] + fn as_fd(&self) -> BorrowedFd<'_> { + unsafe { BorrowedFd::borrow_raw(0) } + } +} + +#[stable(feature = "io_safety", since = "1.63.0")] +impl<'a> AsFd for io::StdinLock<'a> { + #[inline] + fn as_fd(&self) -> BorrowedFd<'_> { + // SAFETY: user code should not close stdin out from under the standard library + unsafe { BorrowedFd::borrow_raw(0) } + } +} + +#[stable(feature = "io_safety", since = "1.63.0")] +impl AsFd for io::Stdout { + #[inline] + fn as_fd(&self) -> BorrowedFd<'_> { + unsafe { BorrowedFd::borrow_raw(1) } + } +} + +#[stable(feature = "io_safety", since = "1.63.0")] +impl<'a> AsFd for io::StdoutLock<'a> { + #[inline] + fn as_fd(&self) -> BorrowedFd<'_> { + // SAFETY: user code should not close stdout out from under the standard library + unsafe { BorrowedFd::borrow_raw(1) } + } +} + +#[stable(feature = "io_safety", since = "1.63.0")] +impl AsFd for io::Stderr { + #[inline] + fn as_fd(&self) -> BorrowedFd<'_> { + unsafe { BorrowedFd::borrow_raw(2) } + } +} + +#[stable(feature = "io_safety", since = "1.63.0")] +impl<'a> AsFd for io::StderrLock<'a> { + #[inline] + fn as_fd(&self) -> BorrowedFd<'_> { + // SAFETY: user code should not close stderr out from under the standard library + unsafe { BorrowedFd::borrow_raw(2) } + } +} diff --git a/library/std/src/os/fd/raw.rs b/library/std/src/os/fd/raw.rs index 081915ed1..f92a05066 100644 --- a/library/std/src/os/fd/raw.rs +++ b/library/std/src/os/fd/raw.rs @@ -14,7 +14,7 @@ use crate::os::wasi::io::OwnedFd; use crate::sys_common::{AsInner, IntoInner}; /// Raw file descriptors. -#[cfg_attr(not(bootstrap), rustc_allowed_through_unstable_modules)] +#[rustc_allowed_through_unstable_modules] #[stable(feature = "rust1", since = "1.0.0")] pub type RawFd = raw::c_int; @@ -23,7 +23,7 @@ pub type RawFd = raw::c_int; /// This is only available on unix and WASI platforms and must be imported in /// order to call the method. Windows platforms have a corresponding /// `AsRawHandle` and `AsRawSocket` set of traits. -#[cfg_attr(not(bootstrap), rustc_allowed_through_unstable_modules)] +#[rustc_allowed_through_unstable_modules] #[stable(feature = "rust1", since = "1.0.0")] pub trait AsRawFd { /// Extracts the raw file descriptor. @@ -42,10 +42,8 @@ pub trait AsRawFd { /// ```no_run /// use std::fs::File; /// # use std::io; - /// #[cfg(unix)] - /// use std::os::unix::io::{AsRawFd, RawFd}; - /// #[cfg(target_os = "wasi")] - /// use std::os::wasi::io::{AsRawFd, RawFd}; + /// #[cfg(any(unix, target_os = "wasi"))] + /// use std::os::fd::{AsRawFd, RawFd}; /// /// let mut f = File::open("foo.txt")?; /// // Note that `raw_fd` is only valid as long as `f` exists. @@ -59,7 +57,7 @@ pub trait AsRawFd { /// A trait to express the ability to construct an object from a raw file /// descriptor. -#[cfg_attr(not(bootstrap), rustc_allowed_through_unstable_modules)] +#[rustc_allowed_through_unstable_modules] #[stable(feature = "from_raw_os", since = "1.1.0")] pub trait FromRawFd { /// Constructs a new instance of `Self` from the given raw file @@ -83,10 +81,8 @@ pub trait FromRawFd { /// ```no_run /// use std::fs::File; /// # use std::io; - /// #[cfg(unix)] - /// use std::os::unix::io::{FromRawFd, IntoRawFd, RawFd}; - /// #[cfg(target_os = "wasi")] - /// use std::os::wasi::io::{FromRawFd, IntoRawFd, RawFd}; + /// #[cfg(any(unix, target_os = "wasi"))] + /// use std::os::fd::{FromRawFd, IntoRawFd, RawFd}; /// /// let f = File::open("foo.txt")?; /// # #[cfg(any(unix, target_os = "wasi"))] @@ -103,7 +99,7 @@ pub trait FromRawFd { /// A trait to express the ability to consume an object and acquire ownership of /// its raw file descriptor. -#[cfg_attr(not(bootstrap), rustc_allowed_through_unstable_modules)] +#[rustc_allowed_through_unstable_modules] #[stable(feature = "into_raw_os", since = "1.4.0")] pub trait IntoRawFd { /// Consumes this object, returning the raw underlying file descriptor. @@ -121,10 +117,8 @@ pub trait IntoRawFd { /// ```no_run /// use std::fs::File; /// # use std::io; - /// #[cfg(unix)] - /// use std::os::unix::io::{IntoRawFd, RawFd}; - /// #[cfg(target_os = "wasi")] - /// use std::os::wasi::io::{IntoRawFd, RawFd}; + /// #[cfg(any(unix, target_os = "wasi"))] + /// use std::os::fd::{IntoRawFd, RawFd}; /// /// let f = File::open("foo.txt")?; /// #[cfg(any(unix, target_os = "wasi"))] diff --git a/library/std/src/os/fortanix_sgx/mod.rs b/library/std/src/os/fortanix_sgx/mod.rs index a40dabe19..39a42f4e1 100644 --- a/library/std/src/os/fortanix_sgx/mod.rs +++ b/library/std/src/os/fortanix_sgx/mod.rs @@ -26,10 +26,13 @@ pub mod usercalls { free, insecure_time, launch_thread, read, read_alloc, send, wait, write, }; pub use crate::sys::abi::usercalls::raw::{do_usercall, Usercalls as UsercallNrs}; + pub use crate::sys::abi::usercalls::raw::{Register, RegisterArgument, ReturnValue}; // fortanix-sgx-abi re-exports pub use crate::sys::abi::usercalls::raw::Error; - pub use crate::sys::abi::usercalls::raw::{ByteBuffer, FifoDescriptor, Return, Usercall}; + pub use crate::sys::abi::usercalls::raw::{ + ByteBuffer, Cancel, FifoDescriptor, Return, Usercall, + }; pub use crate::sys::abi::usercalls::raw::{Fd, Result, Tcs}; pub use crate::sys::abi::usercalls::raw::{ EV_RETURNQ_NOT_EMPTY, EV_UNPARK, EV_USERCALLQ_NOT_FULL, FD_STDERR, FD_STDIN, FD_STDOUT, diff --git a/library/std/src/os/linux/mod.rs b/library/std/src/os/linux/mod.rs index 8e7776f66..c17053011 100644 --- a/library/std/src/os/linux/mod.rs +++ b/library/std/src/os/linux/mod.rs @@ -4,5 +4,6 @@ #![doc(cfg(target_os = "linux"))] pub mod fs; +pub mod net; pub mod process; pub mod raw; diff --git a/library/std/src/os/linux/net.rs b/library/std/src/os/linux/net.rs new file mode 100644 index 000000000..ff96125c3 --- /dev/null +++ b/library/std/src/os/linux/net.rs @@ -0,0 +1,4 @@ +//! Linux and Android-specific definitions for socket options. + +#![unstable(feature = "tcp_quickack", issue = "96256")] +pub use crate::os::net::tcp::TcpStreamExt; diff --git a/library/std/src/os/mod.rs b/library/std/src/os/mod.rs index 6fbaa42c7..42773805c 100644 --- a/library/std/src/os/mod.rs +++ b/library/std/src/os/mod.rs @@ -145,6 +145,11 @@ pub mod solaris; pub mod solid; #[cfg(target_os = "vxworks")] pub mod vxworks; +#[cfg(target_os = "watchos")] +pub(crate) mod watchos; #[cfg(any(unix, target_os = "wasi", doc))] -mod fd; +pub mod fd; + +#[cfg(any(target_os = "linux", target_os = "android", doc))] +mod net; diff --git a/library/std/src/os/net/mod.rs b/library/std/src/os/net/mod.rs new file mode 100644 index 000000000..d6d84d24e --- /dev/null +++ b/library/std/src/os/net/mod.rs @@ -0,0 +1,7 @@ +//! Linux and Android-specific definitions for socket options. + +#![unstable(feature = "tcp_quickack", issue = "96256")] +#![doc(cfg(any(target_os = "linux", target_os = "android",)))] +pub mod tcp; +#[cfg(test)] +mod tests; diff --git a/library/std/src/os/net/tcp.rs b/library/std/src/os/net/tcp.rs new file mode 100644 index 000000000..5e9ee65a4 --- /dev/null +++ b/library/std/src/os/net/tcp.rs @@ -0,0 +1,70 @@ +//! Linux and Android-specific tcp extensions to primitives in the [`std::net`] module. +//! +//! [`std::net`]: crate::net + +use crate::io; +use crate::net; +use crate::sealed::Sealed; +use crate::sys_common::AsInner; + +/// Os-specific extensions for [`TcpStream`] +/// +/// [`TcpStream`]: net::TcpStream +#[unstable(feature = "tcp_quickack", issue = "96256")] +pub trait TcpStreamExt: Sealed { + /// Enable or disable `TCP_QUICKACK`. + /// + /// This flag causes Linux to eagerly send ACKs rather than delaying them. + /// Linux may reset this flag after further operations on the socket. + /// + /// See [`man 7 tcp`](https://man7.org/linux/man-pages/man7/tcp.7.html) and + /// [TCP delayed acknowledgement](https://en.wikipedia.org/wiki/TCP_delayed_acknowledgment) + /// for more information. + /// + /// # Examples + /// + /// ```no_run + /// #![feature(tcp_quickack)] + /// use std::net::TcpStream; + /// use std::os::linux::net::TcpStreamExt; + /// + /// let stream = TcpStream::connect("127.0.0.1:8080") + /// .expect("Couldn't connect to the server..."); + /// stream.set_quickack(true).expect("set_quickack call failed"); + /// ``` + #[unstable(feature = "tcp_quickack", issue = "96256")] + fn set_quickack(&self, quickack: bool) -> io::Result<()>; + + /// Gets the value of the `TCP_QUICKACK` option on this socket. + /// + /// For more information about this option, see [`TcpStreamExt::set_quickack`]. + /// + /// # Examples + /// + /// ```no_run + /// #![feature(tcp_quickack)] + /// use std::net::TcpStream; + /// use std::os::linux::net::TcpStreamExt; + /// + /// let stream = TcpStream::connect("127.0.0.1:8080") + /// .expect("Couldn't connect to the server..."); + /// stream.set_quickack(true).expect("set_quickack call failed"); + /// assert_eq!(stream.quickack().unwrap_or(false), true); + /// ``` + #[unstable(feature = "tcp_quickack", issue = "96256")] + fn quickack(&self) -> io::Result<bool>; +} + +#[unstable(feature = "tcp_quickack", issue = "96256")] +impl Sealed for net::TcpStream {} + +#[unstable(feature = "tcp_quickack", issue = "96256")] +impl TcpStreamExt for net::TcpStream { + fn set_quickack(&self, quickack: bool) -> io::Result<()> { + self.as_inner().as_inner().set_quickack(quickack) + } + + fn quickack(&self) -> io::Result<bool> { + self.as_inner().as_inner().quickack() + } +} diff --git a/library/std/src/os/net/tests.rs b/library/std/src/os/net/tests.rs new file mode 100644 index 000000000..4704e3156 --- /dev/null +++ b/library/std/src/os/net/tests.rs @@ -0,0 +1,29 @@ +#[cfg(any(target_os = "android", target_os = "linux",))] +#[test] +fn quickack() { + use crate::{ + net::{test::next_test_ip4, TcpListener, TcpStream}, + os::net::tcp::TcpStreamExt, + }; + + macro_rules! t { + ($e:expr) => { + match $e { + Ok(t) => t, + Err(e) => panic!("received error for `{}`: {}", stringify!($e), e), + } + }; + } + + let addr = next_test_ip4(); + let _listener = t!(TcpListener::bind(&addr)); + + let stream = t!(TcpStream::connect(&("localhost", addr.port()))); + + t!(stream.set_quickack(false)); + assert_eq!(false, t!(stream.quickack())); + t!(stream.set_quickack(true)); + assert_eq!(true, t!(stream.quickack())); + t!(stream.set_quickack(false)); + assert_eq!(false, t!(stream.quickack())); +} diff --git a/library/std/src/os/unix/io/fd.rs b/library/std/src/os/unix/io/fd.rs deleted file mode 100644 index d4cb69645..000000000 --- a/library/std/src/os/unix/io/fd.rs +++ /dev/null @@ -1,8 +0,0 @@ -//! Owned and borrowed file descriptors. - -// Tests for this module -#[cfg(test)] -mod tests; - -#[stable(feature = "io_safety", since = "1.63.0")] -pub use crate::os::fd::owned::*; diff --git a/library/std/src/os/unix/io/mod.rs b/library/std/src/os/unix/io/mod.rs index 3ab5606f8..25b5dbff1 100644 --- a/library/std/src/os/unix/io/mod.rs +++ b/library/std/src/os/unix/io/mod.rs @@ -77,10 +77,9 @@ #![stable(feature = "rust1", since = "1.0.0")] -mod fd; -mod raw; - -#[stable(feature = "io_safety", since = "1.63.0")] -pub use fd::*; #[stable(feature = "rust1", since = "1.0.0")] -pub use raw::*; +pub use crate::os::fd::*; + +// Tests for this module +#[cfg(test)] +mod tests; diff --git a/library/std/src/os/unix/io/raw.rs b/library/std/src/os/unix/io/raw.rs deleted file mode 100644 index a4d2ba797..000000000 --- a/library/std/src/os/unix/io/raw.rs +++ /dev/null @@ -1,6 +0,0 @@ -//! Unix-specific extensions to general I/O primitives. - -#![stable(feature = "rust1", since = "1.0.0")] - -#[stable(feature = "rust1", since = "1.0.0")] -pub use crate::os::fd::raw::*; diff --git a/library/std/src/os/unix/io/fd/tests.rs b/library/std/src/os/unix/io/tests.rs index 84d2a7a1a..84d2a7a1a 100644 --- a/library/std/src/os/unix/io/fd/tests.rs +++ b/library/std/src/os/unix/io/tests.rs diff --git a/library/std/src/os/unix/mod.rs b/library/std/src/os/unix/mod.rs index 411cc0925..f97fa0fb0 100644 --- a/library/std/src/os/unix/mod.rs +++ b/library/std/src/os/unix/mod.rs @@ -73,6 +73,8 @@ mod platform { pub use crate::os::solaris::*; #[cfg(target_os = "vxworks")] pub use crate::os::vxworks::*; + #[cfg(target_os = "watchos")] + pub use crate::os::watchos::*; } pub mod ffi; diff --git a/library/std/src/os/unix/net/addr.rs b/library/std/src/os/unix/net/addr.rs index 9aeae4b2c..094085e19 100644 --- a/library/std/src/os/unix/net/addr.rs +++ b/library/std/src/os/unix/net/addr.rs @@ -2,7 +2,7 @@ use crate::ffi::OsStr; use crate::os::unix::ffi::OsStrExt; use crate::path::Path; use crate::sys::cvt; -use crate::{ascii, fmt, io, mem, ptr}; +use crate::{fmt, io, mem, ptr}; // FIXME(#43348): Make libc adapt #[doc(cfg(...))] so we don't need these fake definitions here? #[cfg(not(unix))] @@ -64,18 +64,6 @@ enum AddressKind<'a> { Abstract(&'a [u8]), } -struct AsciiEscaped<'a>(&'a [u8]); - -impl<'a> fmt::Display for AsciiEscaped<'a> { - fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { - write!(fmt, "\"")?; - for byte in self.0.iter().cloned().flat_map(ascii::escape_default) { - write!(fmt, "{}", byte as char)?; - } - write!(fmt, "\"") - } -} - /// An address associated with a Unix socket. /// /// # Examples @@ -329,7 +317,7 @@ impl SocketAddr { crate::ptr::copy_nonoverlapping( namespace.as_ptr(), - addr.sun_path.as_mut_ptr().offset(1) as *mut u8, + addr.sun_path.as_mut_ptr().add(1) as *mut u8, namespace.len(), ); let len = (sun_path_offset(&addr) + 1 + namespace.len()) as libc::socklen_t; @@ -343,7 +331,7 @@ impl fmt::Debug for SocketAddr { fn fmt(&self, fmt: &mut fmt::Formatter<'_>) -> fmt::Result { match self.address() { AddressKind::Unnamed => write!(fmt, "(unnamed)"), - AddressKind::Abstract(name) => write!(fmt, "{} (abstract)", AsciiEscaped(name)), + AddressKind::Abstract(name) => write!(fmt, "\"{}\" (abstract)", name.escape_ascii()), AddressKind::Pathname(path) => write!(fmt, "{path:?} (pathname)"), } } diff --git a/library/std/src/os/unix/net/datagram.rs b/library/std/src/os/unix/net/datagram.rs index 8008acfd1..f758f88d0 100644 --- a/library/std/src/os/unix/net/datagram.rs +++ b/library/std/src/os/unix/net/datagram.rs @@ -838,6 +838,31 @@ impl UnixDatagram { self.0.passcred() } + /// Set the id of the socket for network filtering purpose + /// + #[cfg_attr( + any(target_os = "linux", target_os = "freebsd", target_os = "openbsd"), + doc = "```no_run" + )] + #[cfg_attr( + not(any(target_os = "linux", target_os = "freebsd", target_os = "openbsd")), + doc = "```ignore" + )] + /// #![feature(unix_set_mark)] + /// use std::os::unix::net::UnixDatagram; + /// + /// fn main() -> std::io::Result<()> { + /// let sock = UnixDatagram::unbound()?; + /// sock.set_mark(32)?; + /// Ok(()) + /// } + /// ``` + #[cfg(any(doc, target_os = "linux", target_os = "freebsd", target_os = "openbsd",))] + #[unstable(feature = "unix_set_mark", issue = "96467")] + pub fn set_mark(&self, mark: u32) -> io::Result<()> { + self.0.set_mark(mark) + } + /// Returns the value of the `SO_ERROR` option. /// /// # Examples diff --git a/library/std/src/os/unix/net/listener.rs b/library/std/src/os/unix/net/listener.rs index 7c0d53950..02090afc8 100644 --- a/library/std/src/os/unix/net/listener.rs +++ b/library/std/src/os/unix/net/listener.rs @@ -73,9 +73,11 @@ impl UnixListener { unsafe { let inner = Socket::new_raw(libc::AF_UNIX, libc::SOCK_STREAM)?; let (addr, len) = sockaddr_un(path.as_ref())?; + const backlog: libc::c_int = + if cfg!(any(target_os = "linux", target_os = "freebsd")) { -1 } else { 128 }; cvt(libc::bind(inner.as_inner().as_raw_fd(), &addr as *const _ as *const _, len as _))?; - cvt(libc::listen(inner.as_inner().as_raw_fd(), 128))?; + cvt(libc::listen(inner.as_inner().as_raw_fd(), backlog))?; Ok(UnixListener(inner)) } @@ -109,12 +111,16 @@ impl UnixListener { pub fn bind_addr(socket_addr: &SocketAddr) -> io::Result<UnixListener> { unsafe { let inner = Socket::new_raw(libc::AF_UNIX, libc::SOCK_STREAM)?; + #[cfg(target_os = "linux")] + const backlog: libc::c_int = -1; + #[cfg(not(target_os = "linux"))] + const backlog: libc::c_int = 128; cvt(libc::bind( inner.as_raw_fd(), &socket_addr.addr as *const _ as *const _, socket_addr.len as _, ))?; - cvt(libc::listen(inner.as_raw_fd(), 128))?; + cvt(libc::listen(inner.as_raw_fd(), backlog))?; Ok(UnixListener(inner)) } } diff --git a/library/std/src/os/unix/net/stream.rs b/library/std/src/os/unix/net/stream.rs index cc3a88587..dff8f6e85 100644 --- a/library/std/src/os/unix/net/stream.rs +++ b/library/std/src/os/unix/net/stream.rs @@ -427,6 +427,31 @@ impl UnixStream { self.0.passcred() } + /// Set the id of the socket for network filtering purpose + /// + #[cfg_attr( + any(target_os = "linux", target_os = "freebsd", target_os = "openbsd"), + doc = "```no_run" + )] + #[cfg_attr( + not(any(target_os = "linux", target_os = "freebsd", target_os = "openbsd")), + doc = "```ignore" + )] + /// #![feature(unix_set_mark)] + /// use std::os::unix::net::UnixStream; + /// + /// fn main() -> std::io::Result<()> { + /// let sock = UnixStream::connect("/tmp/sock")?; + /// sock.set_mark(32)?; + /// Ok(()) + /// } + /// ``` + #[cfg(any(doc, target_os = "linux", target_os = "freebsd", target_os = "openbsd",))] + #[unstable(feature = "unix_set_mark", issue = "96467")] + pub fn set_mark(&self, mark: u32) -> io::Result<()> { + self.0.set_mark(mark) + } + /// Returns the value of the `SO_ERROR` option. /// /// # Examples diff --git a/library/std/src/os/wasi/io/mod.rs b/library/std/src/os/wasi/io/mod.rs index 6c884e2ea..57bd842a5 100644 --- a/library/std/src/os/wasi/io/mod.rs +++ b/library/std/src/os/wasi/io/mod.rs @@ -1,12 +1,6 @@ //! WASI-specific extensions to general I/O primitives. -#![deny(unsafe_op_in_unsafe_fn)] -#![unstable(feature = "wasi_ext", issue = "71213")] +#![stable(feature = "io_safety", since = "1.63.0")] -mod fd; -mod raw; - -#[unstable(feature = "wasi_ext", issue = "71213")] -pub use fd::*; -#[unstable(feature = "wasi_ext", issue = "71213")] -pub use raw::*; +#[stable(feature = "io_safety", since = "1.63.0")] +pub use crate::os::fd::*; diff --git a/library/std/src/os/watchos/fs.rs b/library/std/src/os/watchos/fs.rs new file mode 100644 index 000000000..a14fe35a7 --- /dev/null +++ b/library/std/src/os/watchos/fs.rs @@ -0,0 +1,142 @@ +#![stable(feature = "metadata_ext", since = "1.1.0")] + +use crate::fs::Metadata; +use crate::sys_common::AsInner; + +#[allow(deprecated)] +use crate::os::watchos::raw; + +/// OS-specific extensions to [`fs::Metadata`]. +/// +/// [`fs::Metadata`]: crate::fs::Metadata +#[stable(feature = "metadata_ext", since = "1.1.0")] +pub trait MetadataExt { + /// Gain a reference to the underlying `stat` structure which contains + /// the raw information returned by the OS. + /// + /// The contents of the returned `stat` are **not** consistent across + /// Unix platforms. The `os::unix::fs::MetadataExt` trait contains the + /// cross-Unix abstractions contained within the raw stat. + #[stable(feature = "metadata_ext", since = "1.1.0")] + #[deprecated( + since = "1.8.0", + note = "deprecated in favor of the accessor \ + methods of this trait" + )] + #[allow(deprecated)] + fn as_raw_stat(&self) -> &raw::stat; + + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_dev(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_ino(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_mode(&self) -> u32; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_nlink(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_uid(&self) -> u32; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_gid(&self) -> u32; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_rdev(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_size(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_atime(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_atime_nsec(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_mtime(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_mtime_nsec(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_ctime(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_ctime_nsec(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_birthtime(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_birthtime_nsec(&self) -> i64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_blksize(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_blocks(&self) -> u64; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_flags(&self) -> u32; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_gen(&self) -> u32; + #[stable(feature = "metadata_ext2", since = "1.8.0")] + fn st_lspare(&self) -> u32; +} + +#[stable(feature = "metadata_ext", since = "1.1.0")] +impl MetadataExt for Metadata { + #[allow(deprecated)] + fn as_raw_stat(&self) -> &raw::stat { + unsafe { &*(self.as_inner().as_inner() as *const libc::stat as *const raw::stat) } + } + fn st_dev(&self) -> u64 { + self.as_inner().as_inner().st_dev as u64 + } + fn st_ino(&self) -> u64 { + self.as_inner().as_inner().st_ino as u64 + } + fn st_mode(&self) -> u32 { + self.as_inner().as_inner().st_mode as u32 + } + fn st_nlink(&self) -> u64 { + self.as_inner().as_inner().st_nlink as u64 + } + fn st_uid(&self) -> u32 { + self.as_inner().as_inner().st_uid as u32 + } + fn st_gid(&self) -> u32 { + self.as_inner().as_inner().st_gid as u32 + } + fn st_rdev(&self) -> u64 { + self.as_inner().as_inner().st_rdev as u64 + } + fn st_size(&self) -> u64 { + self.as_inner().as_inner().st_size as u64 + } + fn st_atime(&self) -> i64 { + self.as_inner().as_inner().st_atime as i64 + } + fn st_atime_nsec(&self) -> i64 { + self.as_inner().as_inner().st_atime_nsec as i64 + } + fn st_mtime(&self) -> i64 { + self.as_inner().as_inner().st_mtime as i64 + } + fn st_mtime_nsec(&self) -> i64 { + self.as_inner().as_inner().st_mtime_nsec as i64 + } + fn st_ctime(&self) -> i64 { + self.as_inner().as_inner().st_ctime as i64 + } + fn st_ctime_nsec(&self) -> i64 { + self.as_inner().as_inner().st_ctime_nsec as i64 + } + fn st_birthtime(&self) -> i64 { + self.as_inner().as_inner().st_birthtime as i64 + } + fn st_birthtime_nsec(&self) -> i64 { + self.as_inner().as_inner().st_birthtime_nsec as i64 + } + fn st_blksize(&self) -> u64 { + self.as_inner().as_inner().st_blksize as u64 + } + fn st_blocks(&self) -> u64 { + self.as_inner().as_inner().st_blocks as u64 + } + fn st_gen(&self) -> u32 { + self.as_inner().as_inner().st_gen as u32 + } + fn st_flags(&self) -> u32 { + self.as_inner().as_inner().st_flags as u32 + } + fn st_lspare(&self) -> u32 { + self.as_inner().as_inner().st_lspare as u32 + } +} diff --git a/library/std/src/os/watchos/mod.rs b/library/std/src/os/watchos/mod.rs new file mode 100644 index 000000000..cd6454ebb --- /dev/null +++ b/library/std/src/os/watchos/mod.rs @@ -0,0 +1,6 @@ +//! watchOS-specific definitions + +#![stable(feature = "raw_ext", since = "1.1.0")] + +pub mod fs; +pub mod raw; diff --git a/library/std/src/os/watchos/raw.rs b/library/std/src/os/watchos/raw.rs new file mode 100644 index 000000000..630a533d9 --- /dev/null +++ b/library/std/src/os/watchos/raw.rs @@ -0,0 +1,83 @@ +//! watchOS-specific raw type definitions + +#![stable(feature = "raw_ext", since = "1.1.0")] +#![deprecated( + since = "1.8.0", + note = "these type aliases are no longer supported by \ + the standard library, the `libc` crate on \ + crates.io should be used instead for the correct \ + definitions" +)] +#![allow(deprecated)] + +use crate::os::raw::c_long; + +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type blkcnt_t = u64; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type blksize_t = u64; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type dev_t = u64; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type ino_t = u64; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type mode_t = u32; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type nlink_t = u64; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type off_t = u64; +#[stable(feature = "raw_ext", since = "1.1.0")] +pub type time_t = i64; + +#[stable(feature = "pthread_t", since = "1.8.0")] +pub type pthread_t = usize; + +#[repr(C)] +#[derive(Clone)] +#[stable(feature = "raw_ext", since = "1.1.0")] +pub struct stat { + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_dev: i32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_mode: u16, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_nlink: u16, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_ino: u64, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_uid: u32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_gid: u32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_rdev: i32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_atime: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_atime_nsec: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_mtime: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_mtime_nsec: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_ctime: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_ctime_nsec: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_birthtime: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_birthtime_nsec: c_long, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_size: i64, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_blocks: i64, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_blksize: i32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_flags: u32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_gen: u32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_lspare: i32, + #[stable(feature = "raw_ext", since = "1.1.0")] + pub st_qspare: [i64; 2], +} diff --git a/library/std/src/os/windows/io/handle.rs b/library/std/src/os/windows/io/handle.rs index 16cc8fa27..1dfecc573 100644 --- a/library/std/src/os/windows/io/handle.rs +++ b/library/std/src/os/windows/io/handle.rs @@ -384,6 +384,23 @@ impl fmt::Debug for OwnedHandle { } } +macro_rules! impl_is_terminal { + ($($t:ty),*$(,)?) => {$( + #[unstable(feature = "sealed", issue = "none")] + impl crate::sealed::Sealed for $t {} + + #[unstable(feature = "is_terminal", issue = "98070")] + impl crate::io::IsTerminal for $t { + #[inline] + fn is_terminal(&self) -> bool { + crate::sys::io::is_terminal(self) + } + } + )*} +} + +impl_is_terminal!(BorrowedHandle<'_>, OwnedHandle); + /// A trait to borrow the handle from an underlying object. #[stable(feature = "io_safety", since = "1.63.0")] pub trait AsHandle { diff --git a/library/std/src/panic.rs b/library/std/src/panic.rs index 45bc56efb..c4f022de0 100644 --- a/library/std/src/panic.rs +++ b/library/std/src/panic.rs @@ -295,23 +295,22 @@ pub fn get_backtrace_style() -> Option<BacktraceStyle> { return Some(style); } - // Setting environment variables for Fuchsia components isn't a standard - // or easily supported workflow. For now, display backtraces by default. - let format = if cfg!(target_os = "fuchsia") { - BacktraceStyle::Full - } else { - crate::env::var_os("RUST_BACKTRACE") - .map(|x| { - if &x == "0" { - BacktraceStyle::Off - } else if &x == "full" { - BacktraceStyle::Full - } else { - BacktraceStyle::Short - } - }) - .unwrap_or(BacktraceStyle::Off) - }; + let format = crate::env::var_os("RUST_BACKTRACE") + .map(|x| { + if &x == "0" { + BacktraceStyle::Off + } else if &x == "full" { + BacktraceStyle::Full + } else { + BacktraceStyle::Short + } + }) + .unwrap_or(if cfg!(target_os = "fuchsia") { + // Fuchsia components default to full backtrace. + BacktraceStyle::Full + } else { + BacktraceStyle::Off + }); set_backtrace_style(format); Some(format) } diff --git a/library/std/src/panicking.rs b/library/std/src/panicking.rs index 25c9201f2..d4976a469 100644 --- a/library/std/src/panicking.rs +++ b/library/std/src/panicking.rs @@ -18,9 +18,9 @@ use crate::intrinsics; use crate::mem::{self, ManuallyDrop}; use crate::process; use crate::sync::atomic::{AtomicBool, Ordering}; +use crate::sync::{PoisonError, RwLock}; use crate::sys::stdio::panic_output; use crate::sys_common::backtrace; -use crate::sys_common::rwlock::StaticRwLock; use crate::sys_common::thread_info; use crate::thread; @@ -71,20 +71,29 @@ extern "C" fn __rust_foreign_exception() -> ! { rtabort!("Rust cannot catch foreign exceptions"); } -#[derive(Copy, Clone)] enum Hook { Default, - Custom(*mut (dyn Fn(&PanicInfo<'_>) + 'static + Sync + Send)), + Custom(Box<dyn Fn(&PanicInfo<'_>) + 'static + Sync + Send>), } impl Hook { - fn custom(f: impl Fn(&PanicInfo<'_>) + 'static + Sync + Send) -> Self { - Self::Custom(Box::into_raw(Box::new(f))) + #[inline] + fn into_box(self) -> Box<dyn Fn(&PanicInfo<'_>) + 'static + Sync + Send> { + match self { + Hook::Default => Box::new(default_hook), + Hook::Custom(hook) => hook, + } + } +} + +impl Default for Hook { + #[inline] + fn default() -> Hook { + Hook::Default } } -static HOOK_LOCK: StaticRwLock = StaticRwLock::new(); -static mut HOOK: Hook = Hook::Default; +static HOOK: RwLock<Hook> = RwLock::new(Hook::Default); /// Registers a custom panic hook, replacing any that was previously registered. /// @@ -125,24 +134,13 @@ pub fn set_hook(hook: Box<dyn Fn(&PanicInfo<'_>) + 'static + Sync + Send>) { panic!("cannot modify the panic hook from a panicking thread"); } - // SAFETY: - // - // - `HOOK` can only be modified while holding write access to `HOOK_LOCK`. - // - The argument of `Box::from_raw` is always a valid pointer that was created using - // `Box::into_raw`. - unsafe { - let guard = HOOK_LOCK.write(); - let old_hook = HOOK; - HOOK = Hook::Custom(Box::into_raw(hook)); - drop(guard); - - if let Hook::Custom(ptr) = old_hook { - #[allow(unused_must_use)] - { - Box::from_raw(ptr); - } - } - } + let new = Hook::Custom(hook); + let mut hook = HOOK.write().unwrap_or_else(PoisonError::into_inner); + let old = mem::replace(&mut *hook, new); + drop(hook); + // Only drop the old hook after releasing the lock to avoid deadlocking + // if its destructor panics. + drop(old); } /// Unregisters the current panic hook, returning it. @@ -179,22 +177,11 @@ pub fn take_hook() -> Box<dyn Fn(&PanicInfo<'_>) + 'static + Sync + Send> { panic!("cannot modify the panic hook from a panicking thread"); } - // SAFETY: - // - // - `HOOK` can only be modified while holding write access to `HOOK_LOCK`. - // - The argument of `Box::from_raw` is always a valid pointer that was created using - // `Box::into_raw`. - unsafe { - let guard = HOOK_LOCK.write(); - let hook = HOOK; - HOOK = Hook::Default; - drop(guard); + let mut hook = HOOK.write().unwrap_or_else(PoisonError::into_inner); + let old_hook = mem::take(&mut *hook); + drop(hook); - match hook { - Hook::Default => Box::new(default_hook), - Hook::Custom(ptr) => Box::from_raw(ptr), - } - } + old_hook.into_box() } /// Atomic combination of [`take_hook`] and [`set_hook`]. Use this to replace the panic handler with @@ -240,24 +227,9 @@ where panic!("cannot modify the panic hook from a panicking thread"); } - // SAFETY: - // - // - `HOOK` can only be modified while holding write access to `HOOK_LOCK`. - // - The argument of `Box::from_raw` is always a valid pointer that was created using - // `Box::into_raw`. - unsafe { - let guard = HOOK_LOCK.write(); - let old_hook = HOOK; - HOOK = Hook::Default; - - let prev = match old_hook { - Hook::Default => Box::new(default_hook), - Hook::Custom(ptr) => Box::from_raw(ptr), - }; - - HOOK = Hook::custom(move |info| hook_fn(&prev, info)); - drop(guard); - } + let mut hook = HOOK.write().unwrap_or_else(PoisonError::into_inner); + let prev = mem::take(&mut *hook).into_box(); + *hook = Hook::Custom(Box::new(move |info| hook_fn(&prev, info))); } fn default_hook(info: &PanicInfo<'_>) { @@ -328,7 +300,7 @@ pub mod panic_count { thread_local! { static LOCAL_PANIC_COUNT: Cell<usize> = const { Cell::new(0) } } // Sum of panic counts from all threads. The purpose of this is to have - // a fast path in `is_zero` (which is used by `panicking`). In any particular + // a fast path in `count_is_zero` (which is used by `panicking`). In any particular // thread, if that thread currently views `GLOBAL_PANIC_COUNT` as being zero, // then `LOCAL_PANIC_COUNT` in that thread is zero. This invariant holds before // and after increase and decrease, but not necessarily during their execution. @@ -336,6 +308,14 @@ pub mod panic_count { // Additionally, the top bit of GLOBAL_PANIC_COUNT (GLOBAL_ALWAYS_ABORT_FLAG) // records whether panic::always_abort() has been called. This can only be // set, never cleared. + // panic::always_abort() is usually called to prevent memory allocations done by + // the panic handling in the child created by `libc::fork`. + // Memory allocations performed in a child created with `libc::fork` are undefined + // behavior in most operating systems. + // Accessing LOCAL_PANIC_COUNT in a child created by `libc::fork` would lead to a memory + // allocation. Only GLOBAL_PANIC_COUNT can be accessed in this situation. This is + // sufficient because a child process will always have exactly one thread only. + // See also #85261 for details. // // This could be viewed as a struct containing a single bit and an n-1-bit // value, but if we wrote it like that it would be more than a single word, @@ -346,15 +326,26 @@ pub mod panic_count { // panicking thread consumes at least 2 bytes of address space. static GLOBAL_PANIC_COUNT: AtomicUsize = AtomicUsize::new(0); + // Return the state of the ALWAYS_ABORT_FLAG and number of panics. + // + // If ALWAYS_ABORT_FLAG is not set, the number is determined on a per-thread + // base (stored in LOCAL_PANIC_COUNT), i.e. it is the amount of recursive calls + // of the calling thread. + // If ALWAYS_ABORT_FLAG is set, the number equals the *global* number of panic + // calls. See above why LOCAL_PANIC_COUNT is not used. pub fn increase() -> (bool, usize) { - ( - GLOBAL_PANIC_COUNT.fetch_add(1, Ordering::Relaxed) & ALWAYS_ABORT_FLAG != 0, + let global_count = GLOBAL_PANIC_COUNT.fetch_add(1, Ordering::Relaxed); + let must_abort = global_count & ALWAYS_ABORT_FLAG != 0; + let panics = if must_abort { + global_count & !ALWAYS_ABORT_FLAG + } else { LOCAL_PANIC_COUNT.with(|c| { let next = c.get() + 1; c.set(next); next - }), - ) + }) + }; + (must_abort, panics) } pub fn decrease() { @@ -397,7 +388,7 @@ pub mod panic_count { } // Slow path is in a separate function to reduce the amount of code - // inlined from `is_zero`. + // inlined from `count_is_zero`. #[inline(never)] #[cold] fn is_zero_slow_path() -> bool { @@ -682,27 +673,26 @@ fn rust_panic_with_hook( crate::sys::abort_internal(); } - unsafe { - let mut info = PanicInfo::internal_constructor(message, location, can_unwind); - let _guard = HOOK_LOCK.read(); - match HOOK { - // Some platforms (like wasm) know that printing to stderr won't ever actually - // print anything, and if that's the case we can skip the default - // hook. Since string formatting happens lazily when calling `payload` - // methods, this means we avoid formatting the string at all! - // (The panic runtime might still call `payload.take_box()` though and trigger - // formatting.) - Hook::Default if panic_output().is_none() => {} - Hook::Default => { - info.set_payload(payload.get()); - default_hook(&info); - } - Hook::Custom(ptr) => { - info.set_payload(payload.get()); - (*ptr)(&info); - } - }; - } + let mut info = PanicInfo::internal_constructor(message, location, can_unwind); + let hook = HOOK.read().unwrap_or_else(PoisonError::into_inner); + match *hook { + // Some platforms (like wasm) know that printing to stderr won't ever actually + // print anything, and if that's the case we can skip the default + // hook. Since string formatting happens lazily when calling `payload` + // methods, this means we avoid formatting the string at all! + // (The panic runtime might still call `payload.take_box()` though and trigger + // formatting.) + Hook::Default if panic_output().is_none() => {} + Hook::Default => { + info.set_payload(payload.get()); + default_hook(&info); + } + Hook::Custom(ref hook) => { + info.set_payload(payload.get()); + hook(&info); + } + }; + drop(hook); if panics > 1 || !can_unwind { // If a thread panics while it's already unwinding then we diff --git a/library/std/src/path.rs b/library/std/src/path.rs index 5dfeb517a..9d6328162 100644 --- a/library/std/src/path.rs +++ b/library/std/src/path.rs @@ -2158,6 +2158,7 @@ impl Path { /// assert_eq!(grand_parent.parent(), None); /// ``` #[stable(feature = "rust1", since = "1.0.0")] + #[doc(alias = "dirname")] #[must_use] pub fn parent(&self) -> Option<&Path> { let mut comps = self.components(); @@ -2225,6 +2226,7 @@ impl Path { /// assert_eq!(None, Path::new("/").file_name()); /// ``` #[stable(feature = "rust1", since = "1.0.0")] + #[doc(alias = "basename")] #[must_use] pub fn file_name(&self) -> Option<&OsStr> { self.components().next_back().and_then(|p| match p { @@ -2401,7 +2403,7 @@ impl Path { self.file_name().map(split_file_at_dot).and_then(|(before, _after)| Some(before)) } - /// Extracts the extension of [`self.file_name`], if possible. + /// Extracts the extension (without the leading dot) of [`self.file_name`], if possible. /// /// The extension is: /// diff --git a/library/std/src/path/tests.rs b/library/std/src/path/tests.rs index 351cf6988..dd307022c 100644 --- a/library/std/src/path/tests.rs +++ b/library/std/src/path/tests.rs @@ -1768,6 +1768,7 @@ fn test_windows_absolute() { } #[bench] +#[cfg_attr(miri, ignore)] // Miri isn't fast... fn bench_path_cmp_fast_path_buf_sort(b: &mut test::Bencher) { let prefix = "my/home"; let mut paths: Vec<_> = @@ -1781,6 +1782,7 @@ fn bench_path_cmp_fast_path_buf_sort(b: &mut test::Bencher) { } #[bench] +#[cfg_attr(miri, ignore)] // Miri isn't fast... fn bench_path_cmp_fast_path_long(b: &mut test::Bencher) { let prefix = "/my/home/is/my/castle/and/my/castle/has/a/rusty/workbench/"; let paths: Vec<_> = @@ -1799,6 +1801,7 @@ fn bench_path_cmp_fast_path_long(b: &mut test::Bencher) { } #[bench] +#[cfg_attr(miri, ignore)] // Miri isn't fast... fn bench_path_cmp_fast_path_short(b: &mut test::Bencher) { let prefix = "my/home"; let paths: Vec<_> = @@ -1817,6 +1820,7 @@ fn bench_path_cmp_fast_path_short(b: &mut test::Bencher) { } #[bench] +#[cfg_attr(miri, ignore)] // Miri isn't fast... fn bench_path_hashset(b: &mut test::Bencher) { let prefix = "/my/home/is/my/castle/and/my/castle/has/a/rusty/workbench/"; let paths: Vec<_> = @@ -1835,6 +1839,7 @@ fn bench_path_hashset(b: &mut test::Bencher) { } #[bench] +#[cfg_attr(miri, ignore)] // Miri isn't fast... fn bench_path_hashset_miss(b: &mut test::Bencher) { let prefix = "/my/home/is/my/castle/and/my/castle/has/a/rusty/workbench/"; let paths: Vec<_> = diff --git a/library/std/src/personality.rs b/library/std/src/personality.rs new file mode 100644 index 000000000..63f0ad4f1 --- /dev/null +++ b/library/std/src/personality.rs @@ -0,0 +1,46 @@ +//! This module contains the implementation of the `eh_personality` lang item. +//! +//! The actual implementation is heavily dependent on the target since Rust +//! tries to use the native stack unwinding mechanism whenever possible. +//! +//! This personality function is still required with `-C panic=abort` because +//! it is used to catch foreign exceptions from `extern "C-unwind"` and turn +//! them into aborts. +//! +//! Additionally, ARM EHABI uses the personality function when generating +//! backtraces. + +mod dwarf; + +#[cfg(not(test))] +cfg_if::cfg_if! { + if #[cfg(target_os = "emscripten")] { + mod emcc; + } else if #[cfg(target_env = "msvc")] { + // This is required by the compiler to exist (e.g., it's a lang item), + // but it's never actually called by the compiler because + // _CxxFrameHandler3 is the personality function that is always used. + // Hence this is just an aborting stub. + #[lang = "eh_personality"] + fn rust_eh_personality() { + core::intrinsics::abort() + } + } else if #[cfg(any( + all(target_family = "windows", target_env = "gnu"), + target_os = "psp", + target_os = "solid_asp3", + all(target_family = "unix", not(target_os = "espidf")), + all(target_vendor = "fortanix", target_env = "sgx"), + ))] { + mod gcc; + } else { + // Targets that don't support unwinding. + // - family=wasm + // - os=none ("bare metal" targets) + // - os=uefi + // - os=espidf + // - os=hermit + // - nvptx64-nvidia-cuda + // - arch=avr + } +} diff --git a/library/panic_unwind/src/dwarf/eh.rs b/library/std/src/personality/dwarf/eh.rs index 7394feab8..27b50c13b 100644 --- a/library/panic_unwind/src/dwarf/eh.rs +++ b/library/std/src/personality/dwarf/eh.rs @@ -11,7 +11,7 @@ #![allow(non_upper_case_globals)] #![allow(unused)] -use crate::dwarf::DwarfReader; +use super::DwarfReader; use core::mem; pub const DW_EH_PE_omit: u8 = 0xFF; @@ -75,7 +75,7 @@ pub unsafe fn find_eh_action(lsda: *const u8, context: &EHContext<'_>) -> Result let call_site_encoding = reader.read::<u8>(); let call_site_table_length = reader.read_uleb128(); - let action_table = reader.ptr.offset(call_site_table_length as isize); + let action_table = reader.ptr.add(call_site_table_length as usize); let ip = context.ip; if !USING_SJLJ_EXCEPTIONS { @@ -98,9 +98,8 @@ pub unsafe fn find_eh_action(lsda: *const u8, context: &EHContext<'_>) -> Result } } } - // Ip is not present in the table. This should not happen... but it does: issue #35011. - // So rather than returning EHAction::Terminate, we do this. - Ok(EHAction::None) + // Ip is not present in the table. This indicates a nounwind call. + Ok(EHAction::Terminate) } else { // SjLj version: // The "IP" is an index into the call-site table, with two exceptions: diff --git a/library/panic_unwind/src/dwarf/mod.rs b/library/std/src/personality/dwarf/mod.rs index 652fbe95a..652fbe95a 100644 --- a/library/panic_unwind/src/dwarf/mod.rs +++ b/library/std/src/personality/dwarf/mod.rs diff --git a/library/panic_unwind/src/dwarf/tests.rs b/library/std/src/personality/dwarf/tests.rs index 1644f3708..1644f3708 100644 --- a/library/panic_unwind/src/dwarf/tests.rs +++ b/library/std/src/personality/dwarf/tests.rs diff --git a/library/std/src/personality/emcc.rs b/library/std/src/personality/emcc.rs new file mode 100644 index 000000000..f942bdf18 --- /dev/null +++ b/library/std/src/personality/emcc.rs @@ -0,0 +1,20 @@ +//! On Emscripten Rust panics are wrapped in C++ exceptions, so we just forward +//! to `__gxx_personality_v0` which is provided by Emscripten. + +use libc::c_int; +use unwind as uw; + +// This is required by the compiler to exist (e.g., it's a lang item), but it's +// never actually called by the compiler. Emscripten EH doesn't use a +// personality function at all, it instead uses __cxa_find_matching_catch. +// Wasm error handling would use __gxx_personality_wasm0. +#[lang = "eh_personality"] +unsafe extern "C" fn rust_eh_personality( + _version: c_int, + _actions: uw::_Unwind_Action, + _exception_class: uw::_Unwind_Exception_Class, + _exception_object: *mut uw::_Unwind_Exception, + _context: *mut uw::_Unwind_Context, +) -> uw::_Unwind_Reason_Code { + core::intrinsics::abort() +} diff --git a/library/std/src/personality/gcc.rs b/library/std/src/personality/gcc.rs new file mode 100644 index 000000000..7f0b0439c --- /dev/null +++ b/library/std/src/personality/gcc.rs @@ -0,0 +1,279 @@ +//! Implementation of panics backed by libgcc/libunwind (in some form). +//! +//! For background on exception handling and stack unwinding please see +//! "Exception Handling in LLVM" (llvm.org/docs/ExceptionHandling.html) and +//! documents linked from it. +//! These are also good reads: +//! * <https://itanium-cxx-abi.github.io/cxx-abi/abi-eh.html> +//! * <https://monoinfinito.wordpress.com/series/exception-handling-in-c/> +//! * <https://www.airs.com/blog/index.php?s=exception+frames> +//! +//! ## A brief summary +//! +//! Exception handling happens in two phases: a search phase and a cleanup +//! phase. +//! +//! In both phases the unwinder walks stack frames from top to bottom using +//! information from the stack frame unwind sections of the current process's +//! modules ("module" here refers to an OS module, i.e., an executable or a +//! dynamic library). +//! +//! For each stack frame, it invokes the associated "personality routine", whose +//! address is also stored in the unwind info section. +//! +//! In the search phase, the job of a personality routine is to examine +//! exception object being thrown, and to decide whether it should be caught at +//! that stack frame. Once the handler frame has been identified, cleanup phase +//! begins. +//! +//! In the cleanup phase, the unwinder invokes each personality routine again. +//! This time it decides which (if any) cleanup code needs to be run for +//! the current stack frame. If so, the control is transferred to a special +//! branch in the function body, the "landing pad", which invokes destructors, +//! frees memory, etc. At the end of the landing pad, control is transferred +//! back to the unwinder and unwinding resumes. +//! +//! Once stack has been unwound down to the handler frame level, unwinding stops +//! and the last personality routine transfers control to the catch block. + +use super::dwarf::eh::{self, EHAction, EHContext}; +use libc::{c_int, uintptr_t}; +use unwind as uw; + +// Register ids were lifted from LLVM's TargetLowering::getExceptionPointerRegister() +// and TargetLowering::getExceptionSelectorRegister() for each architecture, +// then mapped to DWARF register numbers via register definition tables +// (typically <arch>RegisterInfo.td, search for "DwarfRegNum"). +// See also https://llvm.org/docs/WritingAnLLVMBackend.html#defining-a-register. + +#[cfg(target_arch = "x86")] +const UNWIND_DATA_REG: (i32, i32) = (0, 2); // EAX, EDX + +#[cfg(target_arch = "x86_64")] +const UNWIND_DATA_REG: (i32, i32) = (0, 1); // RAX, RDX + +#[cfg(any(target_arch = "arm", target_arch = "aarch64"))] +const UNWIND_DATA_REG: (i32, i32) = (0, 1); // R0, R1 / X0, X1 + +#[cfg(target_arch = "m68k")] +const UNWIND_DATA_REG: (i32, i32) = (0, 1); // D0, D1 + +#[cfg(any(target_arch = "mips", target_arch = "mips64"))] +const UNWIND_DATA_REG: (i32, i32) = (4, 5); // A0, A1 + +#[cfg(any(target_arch = "powerpc", target_arch = "powerpc64"))] +const UNWIND_DATA_REG: (i32, i32) = (3, 4); // R3, R4 / X3, X4 + +#[cfg(target_arch = "s390x")] +const UNWIND_DATA_REG: (i32, i32) = (6, 7); // R6, R7 + +#[cfg(any(target_arch = "sparc", target_arch = "sparc64"))] +const UNWIND_DATA_REG: (i32, i32) = (24, 25); // I0, I1 + +#[cfg(target_arch = "hexagon")] +const UNWIND_DATA_REG: (i32, i32) = (0, 1); // R0, R1 + +#[cfg(any(target_arch = "riscv64", target_arch = "riscv32"))] +const UNWIND_DATA_REG: (i32, i32) = (10, 11); // x10, x11 + +// The following code is based on GCC's C and C++ personality routines. For reference, see: +// https://github.com/gcc-mirror/gcc/blob/master/libstdc++-v3/libsupc++/eh_personality.cc +// https://github.com/gcc-mirror/gcc/blob/trunk/libgcc/unwind-c.c + +cfg_if::cfg_if! { + if #[cfg(all(target_arch = "arm", not(target_os = "ios"), not(target_os = "watchos"), not(target_os = "netbsd")))] { + // ARM EHABI personality routine. + // https://infocenter.arm.com/help/topic/com.arm.doc.ihi0038b/IHI0038B_ehabi.pdf + // + // iOS uses the default routine instead since it uses SjLj unwinding. + #[lang = "eh_personality"] + unsafe extern "C" fn rust_eh_personality( + state: uw::_Unwind_State, + exception_object: *mut uw::_Unwind_Exception, + context: *mut uw::_Unwind_Context, + ) -> uw::_Unwind_Reason_Code { + let state = state as c_int; + let action = state & uw::_US_ACTION_MASK as c_int; + let search_phase = if action == uw::_US_VIRTUAL_UNWIND_FRAME as c_int { + // Backtraces on ARM will call the personality routine with + // state == _US_VIRTUAL_UNWIND_FRAME | _US_FORCE_UNWIND. In those cases + // we want to continue unwinding the stack, otherwise all our backtraces + // would end at __rust_try + if state & uw::_US_FORCE_UNWIND as c_int != 0 { + return continue_unwind(exception_object, context); + } + true + } else if action == uw::_US_UNWIND_FRAME_STARTING as c_int { + false + } else if action == uw::_US_UNWIND_FRAME_RESUME as c_int { + return continue_unwind(exception_object, context); + } else { + return uw::_URC_FAILURE; + }; + + // The DWARF unwinder assumes that _Unwind_Context holds things like the function + // and LSDA pointers, however ARM EHABI places them into the exception object. + // To preserve signatures of functions like _Unwind_GetLanguageSpecificData(), which + // take only the context pointer, GCC personality routines stash a pointer to + // exception_object in the context, using location reserved for ARM's + // "scratch register" (r12). + uw::_Unwind_SetGR(context, uw::UNWIND_POINTER_REG, exception_object as uw::_Unwind_Ptr); + // ...A more principled approach would be to provide the full definition of ARM's + // _Unwind_Context in our libunwind bindings and fetch the required data from there + // directly, bypassing DWARF compatibility functions. + + let eh_action = match find_eh_action(context) { + Ok(action) => action, + Err(_) => return uw::_URC_FAILURE, + }; + if search_phase { + match eh_action { + EHAction::None | EHAction::Cleanup(_) => { + return continue_unwind(exception_object, context); + } + EHAction::Catch(_) => { + // EHABI requires the personality routine to update the + // SP value in the barrier cache of the exception object. + (*exception_object).private[5] = + uw::_Unwind_GetGR(context, uw::UNWIND_SP_REG); + return uw::_URC_HANDLER_FOUND; + } + EHAction::Terminate => return uw::_URC_FAILURE, + } + } else { + match eh_action { + EHAction::None => return continue_unwind(exception_object, context), + EHAction::Cleanup(lpad) | EHAction::Catch(lpad) => { + uw::_Unwind_SetGR( + context, + UNWIND_DATA_REG.0, + exception_object as uintptr_t, + ); + uw::_Unwind_SetGR(context, UNWIND_DATA_REG.1, 0); + uw::_Unwind_SetIP(context, lpad); + return uw::_URC_INSTALL_CONTEXT; + } + EHAction::Terminate => return uw::_URC_FAILURE, + } + } + + // On ARM EHABI the personality routine is responsible for actually + // unwinding a single stack frame before returning (ARM EHABI Sec. 6.1). + unsafe fn continue_unwind( + exception_object: *mut uw::_Unwind_Exception, + context: *mut uw::_Unwind_Context, + ) -> uw::_Unwind_Reason_Code { + if __gnu_unwind_frame(exception_object, context) == uw::_URC_NO_REASON { + uw::_URC_CONTINUE_UNWIND + } else { + uw::_URC_FAILURE + } + } + // defined in libgcc + extern "C" { + fn __gnu_unwind_frame( + exception_object: *mut uw::_Unwind_Exception, + context: *mut uw::_Unwind_Context, + ) -> uw::_Unwind_Reason_Code; + } + } + } else { + // Default personality routine, which is used directly on most targets + // and indirectly on Windows x86_64 via SEH. + unsafe extern "C" fn rust_eh_personality_impl( + version: c_int, + actions: uw::_Unwind_Action, + _exception_class: uw::_Unwind_Exception_Class, + exception_object: *mut uw::_Unwind_Exception, + context: *mut uw::_Unwind_Context, + ) -> uw::_Unwind_Reason_Code { + if version != 1 { + return uw::_URC_FATAL_PHASE1_ERROR; + } + let eh_action = match find_eh_action(context) { + Ok(action) => action, + Err(_) => return uw::_URC_FATAL_PHASE1_ERROR, + }; + if actions as i32 & uw::_UA_SEARCH_PHASE as i32 != 0 { + match eh_action { + EHAction::None | EHAction::Cleanup(_) => uw::_URC_CONTINUE_UNWIND, + EHAction::Catch(_) => uw::_URC_HANDLER_FOUND, + EHAction::Terminate => uw::_URC_FATAL_PHASE1_ERROR, + } + } else { + match eh_action { + EHAction::None => uw::_URC_CONTINUE_UNWIND, + EHAction::Cleanup(lpad) | EHAction::Catch(lpad) => { + uw::_Unwind_SetGR( + context, + UNWIND_DATA_REG.0, + exception_object as uintptr_t, + ); + uw::_Unwind_SetGR(context, UNWIND_DATA_REG.1, 0); + uw::_Unwind_SetIP(context, lpad); + uw::_URC_INSTALL_CONTEXT + } + EHAction::Terminate => uw::_URC_FATAL_PHASE2_ERROR, + } + } + } + + cfg_if::cfg_if! { + if #[cfg(all(windows, target_arch = "x86_64", target_env = "gnu"))] { + // On x86_64 MinGW targets, the unwinding mechanism is SEH however the unwind + // handler data (aka LSDA) uses GCC-compatible encoding. + #[lang = "eh_personality"] + #[allow(nonstandard_style)] + unsafe extern "C" fn rust_eh_personality( + exceptionRecord: *mut uw::EXCEPTION_RECORD, + establisherFrame: uw::LPVOID, + contextRecord: *mut uw::CONTEXT, + dispatcherContext: *mut uw::DISPATCHER_CONTEXT, + ) -> uw::EXCEPTION_DISPOSITION { + uw::_GCC_specific_handler( + exceptionRecord, + establisherFrame, + contextRecord, + dispatcherContext, + rust_eh_personality_impl, + ) + } + } else { + // The personality routine for most of our targets. + #[lang = "eh_personality"] + unsafe extern "C" fn rust_eh_personality( + version: c_int, + actions: uw::_Unwind_Action, + exception_class: uw::_Unwind_Exception_Class, + exception_object: *mut uw::_Unwind_Exception, + context: *mut uw::_Unwind_Context, + ) -> uw::_Unwind_Reason_Code { + rust_eh_personality_impl( + version, + actions, + exception_class, + exception_object, + context, + ) + } + } + } + } +} + +unsafe fn find_eh_action(context: *mut uw::_Unwind_Context) -> Result<EHAction, ()> { + let lsda = uw::_Unwind_GetLanguageSpecificData(context) as *const u8; + let mut ip_before_instr: c_int = 0; + let ip = uw::_Unwind_GetIPInfo(context, &mut ip_before_instr); + let eh_context = EHContext { + // The return address points 1 byte past the call instruction, + // which could be in the next IP range in LSDA range table. + // + // `ip = -1` has special meaning, so use wrapping sub to allow for that + ip: if ip_before_instr != 0 { ip } else { ip.wrapping_sub(1) }, + func_start: uw::_Unwind_GetRegionStart(context), + get_text_start: &|| uw::_Unwind_GetTextRelBase(context), + get_data_start: &|| uw::_Unwind_GetDataRelBase(context), + }; + eh::find_eh_action(lsda, &eh_context) +} diff --git a/library/std/src/primitive_docs.rs b/library/std/src/primitive_docs.rs index b8e546164..331714a99 100644 --- a/library/std/src/primitive_docs.rs +++ b/library/std/src/primitive_docs.rs @@ -611,7 +611,19 @@ mod prim_pointer {} /// /// Arrays coerce to [slices (`[T]`)][slice], so a slice method may be called on /// an array. Indeed, this provides most of the API for working with arrays. -/// Slices have a dynamic size and do not coerce to arrays. +/// +/// Slices have a dynamic size and do not coerce to arrays. Instead, use +/// `slice.try_into().unwrap()` or `<ArrayType>::try_from(slice).unwrap()`. +/// +/// Array's `try_from(slice)` implementations (and the corresponding `slice.try_into()` +/// array implementations) succeed if the input slice length is the same as the result +/// array length. They optimize especially well when the optimizer can easily determine +/// the slice length, e.g. `<[u8; 4]>::try_from(&slice[4..8]).unwrap()`. Array implements +/// [TryFrom](crate::convert::TryFrom) returning: +/// +/// - `[T; N]` copies from the slice's elements +/// - `&[T; N]` references the original slice's elements +/// - `&mut [T; N]` references the original slice's elements /// /// You can move elements out of an array with a [slice pattern]. If you want /// one element, see [`mem::replace`]. @@ -640,6 +652,15 @@ mod prim_pointer {} /// for x in &array { } /// ``` /// +/// You can use `<ArrayType>::try_from(slice)` or `slice.try_into()` to get an array from +/// a slice: +/// +/// ``` +/// let bytes: [u8; 3] = [1, 0, 2]; +/// assert_eq!(1, u16::from_le_bytes(<[u8; 2]>::try_from(&bytes[0..2]).unwrap())); +/// assert_eq!(512, u16::from_le_bytes(bytes[1..3].try_into().unwrap())); +/// ``` +/// /// You can use a [slice pattern] to move elements out of an array: /// /// ``` @@ -801,11 +822,53 @@ mod prim_array {} /// assert_eq!(2 * pointer_size, std::mem::size_of::<Box<[u8]>>()); /// assert_eq!(2 * pointer_size, std::mem::size_of::<Rc<[u8]>>()); /// ``` +/// +/// ## Trait Implementations +/// +/// Some traits are implemented for slices if the element type implements +/// that trait. This includes [`Eq`], [`Hash`] and [`Ord`]. +/// +/// ## Iteration +/// +/// The slices implement `IntoIterator`. The iterator yields references to the +/// slice elements. +/// +/// ``` +/// let numbers: &[i32] = &[0, 1, 2]; +/// for n in numbers { +/// println!("{n} is a number!"); +/// } +/// ``` +/// +/// The mutable slice yields mutable references to the elements: +/// +/// ``` +/// let mut scores: &mut [i32] = &mut [7, 8, 9]; +/// for score in scores { +/// *score += 1; +/// } +/// ``` +/// +/// This iterator yields mutable references to the slice's elements, so while +/// the element type of the slice is `i32`, the element type of the iterator is +/// `&mut i32`. +/// +/// * [`.iter`] and [`.iter_mut`] are the explicit methods to return the default +/// iterators. +/// * Further methods that return iterators are [`.split`], [`.splitn`], +/// [`.chunks`], [`.windows`] and more. +/// +/// [`Hash`]: core::hash::Hash +/// [`.iter`]: slice::iter +/// [`.iter_mut`]: slice::iter_mut +/// [`.split`]: slice::split +/// [`.splitn`]: slice::splitn +/// [`.chunks`]: slice::chunks +/// [`.windows`]: slice::windows #[stable(feature = "rust1", since = "1.0.0")] mod prim_slice {} #[doc(primitive = "str")] -// /// String slices. /// /// *[See also the `std::str` module](crate::str).* @@ -816,19 +879,22 @@ mod prim_slice {} /// /// String slices are always valid UTF-8. /// -/// # Examples +/// # Basic Usage /// /// String literals are string slices: /// /// ``` -/// let hello = "Hello, world!"; -/// -/// // with an explicit type annotation -/// let hello: &'static str = "Hello, world!"; +/// let hello_world = "Hello, World!"; /// ``` /// -/// They are `'static` because they're stored directly in the final binary, and -/// so will be valid for the `'static` duration. +/// Here we have declared a string slice initialized with a string literal. +/// String literals have a static lifetime, which means the string `hello_world` +/// is guaranteed to be valid for the duration of the entire program. +/// We can explicitly specify `hello_world`'s lifetime as well: +/// +/// ``` +/// let hello_world: &'static str = "Hello, world!"; +/// ``` /// /// # Representation /// @@ -996,7 +1062,7 @@ impl<T> (T,) {} // Fake impl that's only really used for docs. #[cfg(doc)] #[stable(feature = "rust1", since = "1.0.0")] -#[cfg_attr(not(bootstrap), doc(fake_variadic))] +#[doc(fake_variadic)] /// This trait is implemented on arbitrary-length tuples. impl<T: Clone> Clone for (T,) { fn clone(&self) -> Self { @@ -1007,7 +1073,7 @@ impl<T: Clone> Clone for (T,) { // Fake impl that's only really used for docs. #[cfg(doc)] #[stable(feature = "rust1", since = "1.0.0")] -#[cfg_attr(not(bootstrap), doc(fake_variadic))] +#[doc(fake_variadic)] /// This trait is implemented on arbitrary-length tuples. impl<T: Copy> Copy for (T,) { // empty @@ -1178,7 +1244,7 @@ mod prim_usize {} #[doc(alias = "&")] #[doc(alias = "&mut")] // -/// References, both shared and mutable. +/// References, `&T` and `&mut T`. /// /// A reference represents a borrow of some owned value. You can get one by using the `&` or `&mut` /// operators on a value, or by using a [`ref`](../std/keyword.ref.html) or @@ -1484,13 +1550,12 @@ mod prim_fn {} // Required to make auto trait impls render. // See src/librustdoc/passes/collect_trait_impls.rs:collect_trait_impls #[doc(hidden)] -#[cfg(not(bootstrap))] impl<Ret, T> fn(T) -> Ret {} // Fake impl that's only really used for docs. #[cfg(doc)] #[stable(feature = "rust1", since = "1.0.0")] -#[cfg_attr(not(bootstrap), doc(fake_variadic))] +#[doc(fake_variadic)] /// This trait is implemented on function pointers with any number of arguments. impl<Ret, T> Clone for fn(T) -> Ret { fn clone(&self) -> Self { @@ -1501,7 +1566,7 @@ impl<Ret, T> Clone for fn(T) -> Ret { // Fake impl that's only really used for docs. #[cfg(doc)] #[stable(feature = "rust1", since = "1.0.0")] -#[cfg_attr(not(bootstrap), doc(fake_variadic))] +#[doc(fake_variadic)] /// This trait is implemented on function pointers with any number of arguments. impl<Ret, T> Copy for fn(T) -> Ret { // empty diff --git a/library/std/src/process.rs b/library/std/src/process.rs index d6cba7e75..400d25beb 100644 --- a/library/std/src/process.rs +++ b/library/std/src/process.rs @@ -169,15 +169,15 @@ use crate::sys_common::{AsInner, AsInnerMut, FromInner, IntoInner}; pub struct Child { pub(crate) handle: imp::Process, - /// The handle for writing to the child's standard input (stdin), if it has - /// been captured. To avoid partially moving - /// the `child` and thus blocking yourself from calling - /// functions on `child` while using `stdin`, - /// you might find it helpful: + /// The handle for writing to the child's standard input (stdin), if it + /// has been captured. You might find it helpful to do /// /// ```compile_fail,E0425 /// let stdin = child.stdin.take().unwrap(); /// ``` + /// + /// to avoid partially moving the `child` and thus blocking yourself from calling + /// functions on `child` while using `stdin`. #[stable(feature = "process", since = "1.0.0")] pub stdin: Option<ChildStdin>, @@ -1629,7 +1629,7 @@ impl ExitStatusError { /// /// This is exactly like [`code()`](Self::code), except that it returns a `NonZeroI32`. /// - /// Plain `code`, returning a plain integer, is provided because is is often more convenient. + /// Plain `code`, returning a plain integer, is provided because it is often more convenient. /// The returned value from `code()` is indeed also nonzero; use `code_nonzero()` when you want /// a type-level guarantee of nonzeroness. /// @@ -2154,8 +2154,16 @@ pub fn id() -> u32 { #[cfg_attr(not(test), lang = "termination")] #[stable(feature = "termination_trait_lib", since = "1.61.0")] #[rustc_on_unimplemented( - message = "`main` has invalid return type `{Self}`", - label = "`main` can only return types that implement `{Termination}`" + on( + all(not(bootstrap), cause = "MainFunctionType"), + message = "`main` has invalid return type `{Self}`", + label = "`main` can only return types that implement `{Termination}`" + ), + on( + bootstrap, + message = "`main` has invalid return type `{Self}`", + label = "`main` can only return types that implement `{Termination}`" + ) )] pub trait Termination { /// Is called to get the representation of the value as status code. @@ -2200,9 +2208,7 @@ impl<T: Termination, E: fmt::Debug> Termination for Result<T, E> { match self { Ok(val) => val.report(), Err(err) => { - // Ignore error if the write fails, for example because stderr is - // already closed. There is not much point panicking at this point. - let _ = writeln!(io::stderr(), "Error: {err:?}"); + io::attempt_print_to_stderr(format_args_nl!("Error: {err:?}")); ExitCode::FAILURE } } diff --git a/library/std/src/rt.rs b/library/std/src/rt.rs index 663537a05..9c2f0c1dd 100644 --- a/library/std/src/rt.rs +++ b/library/std/src/rt.rs @@ -72,10 +72,29 @@ macro_rules! rtunwrap { // Runs before `main`. // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. +// +// # The `sigpipe` parameter +// +// Since 2014, the Rust runtime on Unix has set the `SIGPIPE` handler to +// `SIG_IGN`. Applications have good reasons to want a different behavior +// though, so there is a `#[unix_sigpipe = "..."]` attribute on `fn main()` that +// can be used to select how `SIGPIPE` shall be setup (if changed at all) before +// `fn main()` is called. See <https://github.com/rust-lang/rust/issues/97889> +// for more info. +// +// The `sigpipe` parameter to this function gets its value via the code that +// rustc generates to invoke `fn lang_start()`. The reason we have `sigpipe` for +// all platforms and not only Unix, is because std is not allowed to have `cfg` +// directives as this high level. See the module docs in +// `src/tools/tidy/src/pal.rs` for more info. On all other platforms, `sigpipe` +// has a value, but its value is ignored. +// +// Even though it is an `u8`, it only ever has 4 values. These are documented in +// `compiler/rustc_session/src/config/sigpipe.rs`. #[cfg_attr(test, allow(dead_code))] -unsafe fn init(argc: isize, argv: *const *const u8) { +unsafe fn init(argc: isize, argv: *const *const u8, sigpipe: u8) { unsafe { - sys::init(argc, argv); + sys::init(argc, argv, sigpipe); let main_guard = sys::thread::guard::init(); // Next, set up the current Thread with the guard information we just @@ -107,6 +126,7 @@ fn lang_start_internal( main: &(dyn Fn() -> i32 + Sync + crate::panic::RefUnwindSafe), argc: isize, argv: *const *const u8, + sigpipe: u8, ) -> Result<isize, !> { use crate::{mem, panic}; let rt_abort = move |e| { @@ -124,7 +144,7 @@ fn lang_start_internal( // prevent libstd from accidentally introducing a panic to these functions. Another is from // user code from `main` or, more nefariously, as described in e.g. issue #86030. // SAFETY: Only called once during runtime initialization. - panic::catch_unwind(move || unsafe { init(argc, argv) }).map_err(rt_abort)?; + panic::catch_unwind(move || unsafe { init(argc, argv, sigpipe) }).map_err(rt_abort)?; let ret_code = panic::catch_unwind(move || panic::catch_unwind(main).unwrap_or(101) as isize) .map_err(move |e| { mem::forget(e); @@ -140,11 +160,13 @@ fn lang_start<T: crate::process::Termination + 'static>( main: fn() -> T, argc: isize, argv: *const *const u8, + sigpipe: u8, ) -> isize { let Ok(v) = lang_start_internal( &move || crate::sys_common::backtrace::__rust_begin_short_backtrace(main).report().to_i32(), argc, argv, + sigpipe, ); v } diff --git a/library/std/src/sync/mpsc/mpsc_queue/tests.rs b/library/std/src/sync/mpsc/mpsc_queue/tests.rs index 9f4f31ed0..34b2a9a98 100644 --- a/library/std/src/sync/mpsc/mpsc_queue/tests.rs +++ b/library/std/src/sync/mpsc/mpsc_queue/tests.rs @@ -13,7 +13,7 @@ fn test_full() { #[test] fn test() { let nthreads = 8; - let nmsgs = 1000; + let nmsgs = if cfg!(miri) { 100 } else { 1000 }; let q = Queue::new(); match q.pop() { Empty => {} diff --git a/library/std/src/sync/mpsc/spsc_queue/tests.rs b/library/std/src/sync/mpsc/spsc_queue/tests.rs index 467ef3dbd..eb6d5c2cf 100644 --- a/library/std/src/sync/mpsc/spsc_queue/tests.rs +++ b/library/std/src/sync/mpsc/spsc_queue/tests.rs @@ -77,12 +77,13 @@ fn stress() { } unsafe fn stress_bound(bound: usize) { + let count = if cfg!(miri) { 1000 } else { 100000 }; let q = Arc::new(Queue::with_additions(bound, (), ())); let (tx, rx) = channel(); let q2 = q.clone(); let _t = thread::spawn(move || { - for _ in 0..100000 { + for _ in 0..count { loop { match q2.pop() { Some(1) => break, @@ -93,7 +94,7 @@ fn stress() { } tx.send(()).unwrap(); }); - for _ in 0..100000 { + for _ in 0..count { q.push(1); } rx.recv().unwrap(); diff --git a/library/std/src/sync/mpsc/stream.rs b/library/std/src/sync/mpsc/stream.rs index 4c3812c79..4592e9141 100644 --- a/library/std/src/sync/mpsc/stream.rs +++ b/library/std/src/sync/mpsc/stream.rs @@ -114,7 +114,7 @@ impl<T> Packet<T> { match self.queue.producer_addition().cnt.fetch_add(1, Ordering::SeqCst) { // As described in the mod's doc comment, -1 == wakeup -1 => UpWoke(self.take_to_wake()), - // As as described before, SPSC queues must be >= -2 + // As described before, SPSC queues must be >= -2 -2 => UpSuccess, // Be sure to preserve the disconnected state, and the return value diff --git a/library/std/src/sync/mpsc/sync_tests.rs b/library/std/src/sync/mpsc/sync_tests.rs index e58649bab..63c794369 100644 --- a/library/std/src/sync/mpsc/sync_tests.rs +++ b/library/std/src/sync/mpsc/sync_tests.rs @@ -113,23 +113,25 @@ fn chan_gone_concurrent() { #[test] fn stress() { + let count = if cfg!(miri) { 100 } else { 10000 }; let (tx, rx) = sync_channel::<i32>(0); thread::spawn(move || { - for _ in 0..10000 { + for _ in 0..count { tx.send(1).unwrap(); } }); - for _ in 0..10000 { + for _ in 0..count { assert_eq!(rx.recv().unwrap(), 1); } } #[test] fn stress_recv_timeout_two_threads() { + let count = if cfg!(miri) { 100 } else { 10000 }; let (tx, rx) = sync_channel::<i32>(0); thread::spawn(move || { - for _ in 0..10000 { + for _ in 0..count { tx.send(1).unwrap(); } }); @@ -146,12 +148,12 @@ fn stress_recv_timeout_two_threads() { } } - assert_eq!(recv_count, 10000); + assert_eq!(recv_count, count); } #[test] fn stress_recv_timeout_shared() { - const AMT: u32 = 1000; + const AMT: u32 = if cfg!(miri) { 100 } else { 1000 }; const NTHREADS: u32 = 8; let (tx, rx) = sync_channel::<i32>(0); let (dtx, drx) = sync_channel::<()>(0); @@ -191,7 +193,7 @@ fn stress_recv_timeout_shared() { #[test] fn stress_shared() { - const AMT: u32 = 1000; + const AMT: u32 = if cfg!(miri) { 100 } else { 1000 }; const NTHREADS: u32 = 8; let (tx, rx) = sync_channel::<i32>(0); let (dtx, drx) = sync_channel::<()>(0); @@ -438,12 +440,13 @@ fn stream_send_recv_stress() { #[test] fn recv_a_lot() { + let count = if cfg!(miri) { 1000 } else { 10000 }; // Regression test that we don't run out of stack in scheduler context - let (tx, rx) = sync_channel(10000); - for _ in 0..10000 { + let (tx, rx) = sync_channel(count); + for _ in 0..count { tx.send(()).unwrap(); } - for _ in 0..10000 { + for _ in 0..count { rx.recv().unwrap(); } } diff --git a/library/std/src/sync/mpsc/tests.rs b/library/std/src/sync/mpsc/tests.rs index 4deb3e596..f6d0796f6 100644 --- a/library/std/src/sync/mpsc/tests.rs +++ b/library/std/src/sync/mpsc/tests.rs @@ -120,13 +120,14 @@ fn chan_gone_concurrent() { #[test] fn stress() { + let count = if cfg!(miri) { 100 } else { 10000 }; let (tx, rx) = channel::<i32>(); let t = thread::spawn(move || { - for _ in 0..10000 { + for _ in 0..count { tx.send(1).unwrap(); } }); - for _ in 0..10000 { + for _ in 0..count { assert_eq!(rx.recv().unwrap(), 1); } t.join().ok().expect("thread panicked"); @@ -134,7 +135,7 @@ fn stress() { #[test] fn stress_shared() { - const AMT: u32 = 10000; + const AMT: u32 = if cfg!(miri) { 100 } else { 10000 }; const NTHREADS: u32 = 8; let (tx, rx) = channel::<i32>(); @@ -504,12 +505,13 @@ fn very_long_recv_timeout_wont_panic() { #[test] fn recv_a_lot() { + let count = if cfg!(miri) { 1000 } else { 10000 }; // Regression test that we don't run out of stack in scheduler context let (tx, rx) = channel(); - for _ in 0..10000 { + for _ in 0..count { tx.send(()).unwrap(); } - for _ in 0..10000 { + for _ in 0..count { rx.recv().unwrap(); } } diff --git a/library/std/src/sync/mutex.rs b/library/std/src/sync/mutex.rs index e0d13cd64..de851c8fb 100644 --- a/library/std/src/sync/mutex.rs +++ b/library/std/src/sync/mutex.rs @@ -192,6 +192,7 @@ unsafe impl<T: ?Sized + Send> Sync for Mutex<T> {} and cause Futures to not implement `Send`"] #[stable(feature = "rust1", since = "1.0.0")] #[clippy::has_significant_drop] +#[cfg_attr(not(test), rustc_diagnostic_item = "MutexGuard")] pub struct MutexGuard<'a, T: ?Sized + 'a> { lock: &'a Mutex<T>, poison: poison::Guard, diff --git a/library/std/src/sync/once.rs b/library/std/src/sync/once.rs index a7feea588..0f25417d6 100644 --- a/library/std/src/sync/once.rs +++ b/library/std/src/sync/once.rs @@ -3,99 +3,12 @@ //! This primitive is meant to be used to run one-time initialization. An //! example use case would be for initializing an FFI library. -// A "once" is a relatively simple primitive, and it's also typically provided -// by the OS as well (see `pthread_once` or `InitOnceExecuteOnce`). The OS -// primitives, however, tend to have surprising restrictions, such as the Unix -// one doesn't allow an argument to be passed to the function. -// -// As a result, we end up implementing it ourselves in the standard library. -// This also gives us the opportunity to optimize the implementation a bit which -// should help the fast path on call sites. Consequently, let's explain how this -// primitive works now! -// -// So to recap, the guarantees of a Once are that it will call the -// initialization closure at most once, and it will never return until the one -// that's running has finished running. This means that we need some form of -// blocking here while the custom callback is running at the very least. -// Additionally, we add on the restriction of **poisoning**. Whenever an -// initialization closure panics, the Once enters a "poisoned" state which means -// that all future calls will immediately panic as well. -// -// So to implement this, one might first reach for a `Mutex`, but those cannot -// be put into a `static`. It also gets a lot harder with poisoning to figure -// out when the mutex needs to be deallocated because it's not after the closure -// finishes, but after the first successful closure finishes. -// -// All in all, this is instead implemented with atomics and lock-free -// operations! Whee! Each `Once` has one word of atomic state, and this state is -// CAS'd on to determine what to do. There are four possible state of a `Once`: -// -// * Incomplete - no initialization has run yet, and no thread is currently -// using the Once. -// * Poisoned - some thread has previously attempted to initialize the Once, but -// it panicked, so the Once is now poisoned. There are no other -// threads currently accessing this Once. -// * Running - some thread is currently attempting to run initialization. It may -// succeed, so all future threads need to wait for it to finish. -// Note that this state is accompanied with a payload, described -// below. -// * Complete - initialization has completed and all future calls should finish -// immediately. -// -// With 4 states we need 2 bits to encode this, and we use the remaining bits -// in the word we have allocated as a queue of threads waiting for the thread -// responsible for entering the RUNNING state. This queue is just a linked list -// of Waiter nodes which is monotonically increasing in size. Each node is -// allocated on the stack, and whenever the running closure finishes it will -// consume the entire queue and notify all waiters they should try again. -// -// You'll find a few more details in the implementation, but that's the gist of -// it! -// -// Atomic orderings: -// When running `Once` we deal with multiple atomics: -// `Once.state_and_queue` and an unknown number of `Waiter.signaled`. -// * `state_and_queue` is used (1) as a state flag, (2) for synchronizing the -// result of the `Once`, and (3) for synchronizing `Waiter` nodes. -// - At the end of the `call_inner` function we have to make sure the result -// of the `Once` is acquired. So every load which can be the only one to -// load COMPLETED must have at least Acquire ordering, which means all -// three of them. -// - `WaiterQueue::Drop` is the only place that may store COMPLETED, and -// must do so with Release ordering to make the result available. -// - `wait` inserts `Waiter` nodes as a pointer in `state_and_queue`, and -// needs to make the nodes available with Release ordering. The load in -// its `compare_exchange` can be Relaxed because it only has to compare -// the atomic, not to read other data. -// - `WaiterQueue::Drop` must see the `Waiter` nodes, so it must load -// `state_and_queue` with Acquire ordering. -// - There is just one store where `state_and_queue` is used only as a -// state flag, without having to synchronize data: switching the state -// from INCOMPLETE to RUNNING in `call_inner`. This store can be Relaxed, -// but the read has to be Acquire because of the requirements mentioned -// above. -// * `Waiter.signaled` is both used as a flag, and to protect a field with -// interior mutability in `Waiter`. `Waiter.thread` is changed in -// `WaiterQueue::Drop` which then sets `signaled` with Release ordering. -// After `wait` loads `signaled` with Acquire and sees it is true, it needs to -// see the changes to drop the `Waiter` struct correctly. -// * There is one place where the two atomics `Once.state_and_queue` and -// `Waiter.signaled` come together, and might be reordered by the compiler or -// processor. Because both use Acquire ordering such a reordering is not -// allowed, so no need for SeqCst. - #[cfg(all(test, not(target_os = "emscripten")))] mod tests; -use crate::cell::Cell; use crate::fmt; -use crate::marker; use crate::panic::{RefUnwindSafe, UnwindSafe}; -use crate::ptr; -use crate::sync::atomic::{AtomicBool, AtomicPtr, Ordering}; -use crate::thread::{self, Thread}; - -type Masked = (); +use crate::sys_common::once as sys; /// A synchronization primitive which can be used to run a one-time global /// initialization. Useful for one-time initialization for FFI or related @@ -114,19 +27,9 @@ type Masked = (); /// ``` #[stable(feature = "rust1", since = "1.0.0")] pub struct Once { - // `state_and_queue` is actually a pointer to a `Waiter` with extra state - // bits, so we add the `PhantomData` appropriately. - state_and_queue: AtomicPtr<Masked>, - _marker: marker::PhantomData<*const Waiter>, + inner: sys::Once, } -// The `PhantomData` of a raw pointer removes these two auto traits, but we -// enforce both below in the implementation so this should be safe to add. -#[stable(feature = "rust1", since = "1.0.0")] -unsafe impl Sync for Once {} -#[stable(feature = "rust1", since = "1.0.0")] -unsafe impl Send for Once {} - #[stable(feature = "sync_once_unwind_safe", since = "1.59.0")] impl UnwindSafe for Once {} @@ -136,10 +39,8 @@ impl RefUnwindSafe for Once {} /// State yielded to [`Once::call_once_force()`]ās closure parameter. The state /// can be used to query the poison status of the [`Once`]. #[stable(feature = "once_poison", since = "1.51.0")] -#[derive(Debug)] pub struct OnceState { - poisoned: bool, - set_state_on_drop_to: Cell<*mut Masked>, + pub(crate) inner: sys::OnceState, } /// Initialization value for static [`Once`] values. @@ -159,38 +60,6 @@ pub struct OnceState { )] pub const ONCE_INIT: Once = Once::new(); -// Four states that a Once can be in, encoded into the lower bits of -// `state_and_queue` in the Once structure. -const INCOMPLETE: usize = 0x0; -const POISONED: usize = 0x1; -const RUNNING: usize = 0x2; -const COMPLETE: usize = 0x3; - -// Mask to learn about the state. All other bits are the queue of waiters if -// this is in the RUNNING state. -const STATE_MASK: usize = 0x3; - -// Representation of a node in the linked list of waiters, used while in the -// RUNNING state. -// Note: `Waiter` can't hold a mutable pointer to the next thread, because then -// `wait` would both hand out a mutable reference to its `Waiter` node, and keep -// a shared reference to check `signaled`. Instead we hold shared references and -// use interior mutability. -#[repr(align(4))] // Ensure the two lower bits are free to use as state bits. -struct Waiter { - thread: Cell<Option<Thread>>, - signaled: AtomicBool, - next: *const Waiter, -} - -// Head of a linked list of waiters. -// Every node is a struct on the stack of a waiting thread. -// Will wake up the waiters when it gets dropped, i.e. also on panic. -struct WaiterQueue<'a> { - state_and_queue: &'a AtomicPtr<Masked>, - set_state_on_drop_to: *mut Masked, -} - impl Once { /// Creates a new `Once` value. #[inline] @@ -198,10 +67,7 @@ impl Once { #[rustc_const_stable(feature = "const_once_new", since = "1.32.0")] #[must_use] pub const fn new() -> Once { - Once { - state_and_queue: AtomicPtr::new(ptr::invalid_mut(INCOMPLETE)), - _marker: marker::PhantomData, - } + Once { inner: sys::Once::new() } } /// Performs an initialization routine once and only once. The given closure @@ -261,6 +127,7 @@ impl Once { /// This is similar to [poisoning with mutexes][poison]. /// /// [poison]: struct.Mutex.html#poisoning + #[inline] #[stable(feature = "rust1", since = "1.0.0")] #[track_caller] pub fn call_once<F>(&self, f: F) @@ -268,12 +135,12 @@ impl Once { F: FnOnce(), { // Fast path check - if self.is_completed() { + if self.inner.is_completed() { return; } let mut f = Some(f); - self.call_inner(false, &mut |_| f.take().unwrap()()); + self.inner.call(false, &mut |_| f.take().unwrap()()); } /// Performs the same function as [`call_once()`] except ignores poisoning. @@ -320,18 +187,19 @@ impl Once { /// // once any success happens, we stop propagating the poison /// INIT.call_once(|| {}); /// ``` + #[inline] #[stable(feature = "once_poison", since = "1.51.0")] pub fn call_once_force<F>(&self, f: F) where F: FnOnce(&OnceState), { // Fast path check - if self.is_completed() { + if self.inner.is_completed() { return; } let mut f = Some(f); - self.call_inner(true, &mut |p| f.take().unwrap()(p)); + self.inner.call(true, &mut |p| f.take().unwrap()(p)); } /// Returns `true` if some [`call_once()`] call has completed @@ -378,119 +246,7 @@ impl Once { #[stable(feature = "once_is_completed", since = "1.43.0")] #[inline] pub fn is_completed(&self) -> bool { - // An `Acquire` load is enough because that makes all the initialization - // operations visible to us, and, this being a fast path, weaker - // ordering helps with performance. This `Acquire` synchronizes with - // `Release` operations on the slow path. - self.state_and_queue.load(Ordering::Acquire).addr() == COMPLETE - } - - // This is a non-generic function to reduce the monomorphization cost of - // using `call_once` (this isn't exactly a trivial or small implementation). - // - // Additionally, this is tagged with `#[cold]` as it should indeed be cold - // and it helps let LLVM know that calls to this function should be off the - // fast path. Essentially, this should help generate more straight line code - // in LLVM. - // - // Finally, this takes an `FnMut` instead of a `FnOnce` because there's - // currently no way to take an `FnOnce` and call it via virtual dispatch - // without some allocation overhead. - #[cold] - #[track_caller] - fn call_inner(&self, ignore_poisoning: bool, init: &mut dyn FnMut(&OnceState)) { - let mut state_and_queue = self.state_and_queue.load(Ordering::Acquire); - loop { - match state_and_queue.addr() { - COMPLETE => break, - POISONED if !ignore_poisoning => { - // Panic to propagate the poison. - panic!("Once instance has previously been poisoned"); - } - POISONED | INCOMPLETE => { - // Try to register this thread as the one RUNNING. - let exchange_result = self.state_and_queue.compare_exchange( - state_and_queue, - ptr::invalid_mut(RUNNING), - Ordering::Acquire, - Ordering::Acquire, - ); - if let Err(old) = exchange_result { - state_and_queue = old; - continue; - } - // `waiter_queue` will manage other waiting threads, and - // wake them up on drop. - let mut waiter_queue = WaiterQueue { - state_and_queue: &self.state_and_queue, - set_state_on_drop_to: ptr::invalid_mut(POISONED), - }; - // Run the initialization function, letting it know if we're - // poisoned or not. - let init_state = OnceState { - poisoned: state_and_queue.addr() == POISONED, - set_state_on_drop_to: Cell::new(ptr::invalid_mut(COMPLETE)), - }; - init(&init_state); - waiter_queue.set_state_on_drop_to = init_state.set_state_on_drop_to.get(); - break; - } - _ => { - // All other values must be RUNNING with possibly a - // pointer to the waiter queue in the more significant bits. - assert!(state_and_queue.addr() & STATE_MASK == RUNNING); - wait(&self.state_and_queue, state_and_queue); - state_and_queue = self.state_and_queue.load(Ordering::Acquire); - } - } - } - } -} - -fn wait(state_and_queue: &AtomicPtr<Masked>, mut current_state: *mut Masked) { - // Note: the following code was carefully written to avoid creating a - // mutable reference to `node` that gets aliased. - loop { - // Don't queue this thread if the status is no longer running, - // otherwise we will not be woken up. - if current_state.addr() & STATE_MASK != RUNNING { - return; - } - - // Create the node for our current thread. - let node = Waiter { - thread: Cell::new(Some(thread::current())), - signaled: AtomicBool::new(false), - next: current_state.with_addr(current_state.addr() & !STATE_MASK) as *const Waiter, - }; - let me = &node as *const Waiter as *const Masked as *mut Masked; - - // Try to slide in the node at the head of the linked list, making sure - // that another thread didn't just replace the head of the linked list. - let exchange_result = state_and_queue.compare_exchange( - current_state, - me.with_addr(me.addr() | RUNNING), - Ordering::Release, - Ordering::Relaxed, - ); - if let Err(old) = exchange_result { - current_state = old; - continue; - } - - // We have enqueued ourselves, now lets wait. - // It is important not to return before being signaled, otherwise we - // would drop our `Waiter` node and leave a hole in the linked list - // (and a dangling reference). Guard against spurious wakeups by - // reparking ourselves until we are signaled. - while !node.signaled.load(Ordering::Acquire) { - // If the managing thread happens to signal and unpark us before we - // can park ourselves, the result could be this thread never gets - // unparked. Luckily `park` comes with the guarantee that if it got - // an `unpark` just before on an unparked thread it does not park. - thread::park(); - } - break; + self.inner.is_completed() } } @@ -501,37 +257,6 @@ impl fmt::Debug for Once { } } -impl Drop for WaiterQueue<'_> { - fn drop(&mut self) { - // Swap out our state with however we finished. - let state_and_queue = - self.state_and_queue.swap(self.set_state_on_drop_to, Ordering::AcqRel); - - // We should only ever see an old state which was RUNNING. - assert_eq!(state_and_queue.addr() & STATE_MASK, RUNNING); - - // Walk the entire linked list of waiters and wake them up (in lifo - // order, last to register is first to wake up). - unsafe { - // Right after setting `node.signaled = true` the other thread may - // free `node` if there happens to be has a spurious wakeup. - // So we have to take out the `thread` field and copy the pointer to - // `next` first. - let mut queue = - state_and_queue.with_addr(state_and_queue.addr() & !STATE_MASK) as *const Waiter; - while !queue.is_null() { - let next = (*queue).next; - let thread = (*queue).thread.take().unwrap(); - (*queue).signaled.store(true, Ordering::Release); - // ^- FIXME (maybe): This is another case of issue #55005 - // `store()` has a potentially dangling ref to `signaled`. - queue = next; - thread.unpark(); - } - } - } -} - impl OnceState { /// Returns `true` if the associated [`Once`] was poisoned prior to the /// invocation of the closure passed to [`Once::call_once_force()`]. @@ -568,13 +293,22 @@ impl OnceState { /// assert!(!state.is_poisoned()); /// }); #[stable(feature = "once_poison", since = "1.51.0")] + #[inline] pub fn is_poisoned(&self) -> bool { - self.poisoned + self.inner.is_poisoned() } /// Poison the associated [`Once`] without explicitly panicking. - // NOTE: This is currently only exposed for the `lazy` module + // NOTE: This is currently only exposed for `OnceLock`. + #[inline] pub(crate) fn poison(&self) { - self.set_state_on_drop_to.set(ptr::invalid_mut(POISONED)); + self.inner.poison(); + } +} + +#[stable(feature = "std_debug", since = "1.16.0")] +impl fmt::Debug for OnceState { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("OnceState").field("poisoned", &self.is_poisoned()).finish() } } diff --git a/library/std/src/sync/once_lock.rs b/library/std/src/sync/once_lock.rs index 813516040..37413ec62 100644 --- a/library/std/src/sync/once_lock.rs +++ b/library/std/src/sync/once_lock.rs @@ -3,7 +3,6 @@ use crate::fmt; use crate::marker::PhantomData; use crate::mem::MaybeUninit; use crate::panic::{RefUnwindSafe, UnwindSafe}; -use crate::pin::Pin; use crate::sync::Once; /// A synchronization primitive which can be written to only once. @@ -223,60 +222,6 @@ impl<T> OnceLock<T> { Ok(unsafe { self.get_unchecked() }) } - /// Internal-only API that gets the contents of the cell, initializing it - /// in two steps with `f` and `g` if the cell was empty. - /// - /// `f` is called to construct the value, which is then moved into the cell - /// and given as a (pinned) mutable reference to `g` to finish - /// initialization. - /// - /// This allows `g` to inspect an manipulate the value after it has been - /// moved into its final place in the cell, but before the cell is - /// considered initialized. - /// - /// # Panics - /// - /// If `f` or `g` panics, the panic is propagated to the caller, and the - /// cell remains uninitialized. - /// - /// With the current implementation, if `g` panics, the value from `f` will - /// not be dropped. This should probably be fixed if this is ever used for - /// a type where this matters. - /// - /// It is an error to reentrantly initialize the cell from `f`. The exact - /// outcome is unspecified. Current implementation deadlocks, but this may - /// be changed to a panic in the future. - pub(crate) fn get_or_init_pin<F, G>(self: Pin<&Self>, f: F, g: G) -> Pin<&T> - where - F: FnOnce() -> T, - G: FnOnce(Pin<&mut T>), - { - if let Some(value) = self.get_ref().get() { - // SAFETY: The inner value was already initialized, and will not be - // moved anymore. - return unsafe { Pin::new_unchecked(value) }; - } - - let slot = &self.value; - - // Ignore poisoning from other threads - // If another thread panics, then we'll be able to run our closure - self.once.call_once_force(|_| { - let value = f(); - // SAFETY: We use the Once (self.once) to guarantee unique access - // to the UnsafeCell (slot). - let value: &mut T = unsafe { (&mut *slot.get()).write(value) }; - // SAFETY: The value has been written to its final place in - // self.value. We do not to move it anymore, which we promise here - // with a Pin<&mut T>. - g(unsafe { Pin::new_unchecked(value) }); - }); - - // SAFETY: The inner value has been initialized, and will not be moved - // anymore. - unsafe { Pin::new_unchecked(self.get_ref().get_unchecked()) } - } - /// Consumes the `OnceLock`, returning the wrapped value. Returns /// `None` if the cell was empty. /// diff --git a/library/std/src/sync/rwlock.rs b/library/std/src/sync/rwlock.rs index 6e4a2cfc8..8b3877607 100644 --- a/library/std/src/sync/rwlock.rs +++ b/library/std/src/sync/rwlock.rs @@ -76,6 +76,7 @@ use crate::sys_common::rwlock as sys; /// /// [`Mutex`]: super::Mutex #[stable(feature = "rust1", since = "1.0.0")] +#[cfg_attr(not(test), rustc_diagnostic_item = "RwLock")] pub struct RwLock<T: ?Sized> { inner: sys::MovableRwLock, poison: poison::Flag, @@ -101,6 +102,7 @@ unsafe impl<T: ?Sized + Send + Sync> Sync for RwLock<T> {} and cause Futures to not implement `Send`"] #[stable(feature = "rust1", since = "1.0.0")] #[clippy::has_significant_drop] +#[cfg_attr(not(test), rustc_diagnostic_item = "RwLockReadGuard")] pub struct RwLockReadGuard<'a, T: ?Sized + 'a> { // NB: we use a pointer instead of `&'a T` to avoid `noalias` violations, because a // `Ref` argument doesn't hold immutability for its whole scope, only until it drops. @@ -130,6 +132,7 @@ unsafe impl<T: ?Sized + Sync> Sync for RwLockReadGuard<'_, T> {} and cause Future's to not implement `Send`"] #[stable(feature = "rust1", since = "1.0.0")] #[clippy::has_significant_drop] +#[cfg_attr(not(test), rustc_diagnostic_item = "RwLockWriteGuard")] pub struct RwLockWriteGuard<'a, T: ?Sized + 'a> { lock: &'a RwLock<T>, poison: poison::Guard, @@ -164,7 +167,7 @@ impl<T> RwLock<T> { } impl<T: ?Sized> RwLock<T> { - /// Locks this rwlock with shared read access, blocking the current thread + /// Locks this `RwLock` with shared read access, blocking the current thread /// until it can be acquired. /// /// The calling thread will be blocked until there are no more writers which @@ -178,9 +181,10 @@ impl<T: ?Sized> RwLock<T> { /// /// # Errors /// - /// This function will return an error if the RwLock is poisoned. An RwLock - /// is poisoned whenever a writer panics while holding an exclusive lock. - /// The failure will occur immediately after the lock has been acquired. + /// This function will return an error if the `RwLock` is poisoned. An + /// `RwLock` is poisoned whenever a writer panics while holding an exclusive + /// lock. The failure will occur immediately after the lock has been + /// acquired. /// /// # Panics /// @@ -212,7 +216,7 @@ impl<T: ?Sized> RwLock<T> { } } - /// Attempts to acquire this rwlock with shared read access. + /// Attempts to acquire this `RwLock` with shared read access. /// /// If the access could not be granted at this time, then `Err` is returned. /// Otherwise, an RAII guard is returned which will release the shared access @@ -225,13 +229,13 @@ impl<T: ?Sized> RwLock<T> { /// /// # Errors /// - /// This function will return the [`Poisoned`] error if the RwLock is poisoned. - /// An RwLock is poisoned whenever a writer panics while holding an exclusive - /// lock. `Poisoned` will only be returned if the lock would have otherwise been - /// acquired. + /// This function will return the [`Poisoned`] error if the `RwLock` is + /// poisoned. An `RwLock` is poisoned whenever a writer panics while holding + /// an exclusive lock. `Poisoned` will only be returned if the lock would + /// have otherwise been acquired. /// - /// This function will return the [`WouldBlock`] error if the RwLock could not - /// be acquired because it was already locked exclusively. + /// This function will return the [`WouldBlock`] error if the `RwLock` could + /// not be acquired because it was already locked exclusively. /// /// [`Poisoned`]: TryLockError::Poisoned /// [`WouldBlock`]: TryLockError::WouldBlock @@ -260,20 +264,20 @@ impl<T: ?Sized> RwLock<T> { } } - /// Locks this rwlock with exclusive write access, blocking the current + /// Locks this `RwLock` with exclusive write access, blocking the current /// thread until it can be acquired. /// /// This function will not return while other writers or other readers /// currently have access to the lock. /// - /// Returns an RAII guard which will drop the write access of this rwlock + /// Returns an RAII guard which will drop the write access of this `RwLock` /// when dropped. /// /// # Errors /// - /// This function will return an error if the RwLock is poisoned. An RwLock - /// is poisoned whenever a writer panics while holding an exclusive lock. - /// An error will be returned when the lock is acquired. + /// This function will return an error if the `RwLock` is poisoned. An + /// `RwLock` is poisoned whenever a writer panics while holding an exclusive + /// lock. An error will be returned when the lock is acquired. /// /// # Panics /// @@ -300,7 +304,7 @@ impl<T: ?Sized> RwLock<T> { } } - /// Attempts to lock this rwlock with exclusive write access. + /// Attempts to lock this `RwLock` with exclusive write access. /// /// If the lock could not be acquired at this time, then `Err` is returned. /// Otherwise, an RAII guard is returned which will release the lock when @@ -313,13 +317,13 @@ impl<T: ?Sized> RwLock<T> { /// /// # Errors /// - /// This function will return the [`Poisoned`] error if the RwLock is - /// poisoned. An RwLock is poisoned whenever a writer panics while holding - /// an exclusive lock. `Poisoned` will only be returned if the lock would have - /// otherwise been acquired. + /// This function will return the [`Poisoned`] error if the `RwLock` is + /// poisoned. An `RwLock` is poisoned whenever a writer panics while holding + /// an exclusive lock. `Poisoned` will only be returned if the lock would + /// have otherwise been acquired. /// - /// This function will return the [`WouldBlock`] error if the RwLock could not - /// be acquired because it was already locked exclusively. + /// This function will return the [`WouldBlock`] error if the `RwLock` could + /// not be acquired because it was already locked exclusively. /// /// [`Poisoned`]: TryLockError::Poisoned /// [`WouldBlock`]: TryLockError::WouldBlock @@ -419,10 +423,10 @@ impl<T: ?Sized> RwLock<T> { /// /// # Errors /// - /// This function will return an error if the RwLock is poisoned. An RwLock - /// is poisoned whenever a writer panics while holding an exclusive lock. An - /// error will only be returned if the lock would have otherwise been - /// acquired. + /// This function will return an error if the `RwLock` is poisoned. An + /// `RwLock` is poisoned whenever a writer panics while holding an exclusive + /// lock. An error will only be returned if the lock would have otherwise + /// been acquired. /// /// # Examples /// @@ -452,10 +456,10 @@ impl<T: ?Sized> RwLock<T> { /// /// # Errors /// - /// This function will return an error if the RwLock is poisoned. An RwLock - /// is poisoned whenever a writer panics while holding an exclusive lock. An - /// error will only be returned if the lock would have otherwise been - /// acquired. + /// This function will return an error if the `RwLock` is poisoned. An + /// `RwLock` is poisoned whenever a writer panics while holding an exclusive + /// lock. An error will only be returned if the lock would have otherwise + /// been acquired. /// /// # Examples /// diff --git a/library/std/src/sync/rwlock/tests.rs b/library/std/src/sync/rwlock/tests.rs index 08255c985..b5b3ad989 100644 --- a/library/std/src/sync/rwlock/tests.rs +++ b/library/std/src/sync/rwlock/tests.rs @@ -19,7 +19,7 @@ fn smoke() { #[test] fn frob() { const N: u32 = 10; - const M: usize = 1000; + const M: usize = if cfg!(miri) { 100 } else { 1000 }; let r = Arc::new(RwLock::new(())); diff --git a/library/std/src/sys/common/mod.rs b/library/std/src/sys/common/mod.rs index ff64d2aa8..29fc0835d 100644 --- a/library/std/src/sys/common/mod.rs +++ b/library/std/src/sys/common/mod.rs @@ -11,3 +11,7 @@ #![allow(dead_code)] pub mod alloc; +pub mod small_c_string; + +#[cfg(test)] +mod tests; diff --git a/library/std/src/sys/common/small_c_string.rs b/library/std/src/sys/common/small_c_string.rs new file mode 100644 index 000000000..01acd5191 --- /dev/null +++ b/library/std/src/sys/common/small_c_string.rs @@ -0,0 +1,58 @@ +use crate::ffi::{CStr, CString}; +use crate::mem::MaybeUninit; +use crate::path::Path; +use crate::slice; +use crate::{io, ptr}; + +// Make sure to stay under 4096 so the compiler doesn't insert a probe frame: +// https://docs.rs/compiler_builtins/latest/compiler_builtins/probestack/index.html +#[cfg(not(target_os = "espidf"))] +const MAX_STACK_ALLOCATION: usize = 384; +#[cfg(target_os = "espidf")] +const MAX_STACK_ALLOCATION: usize = 32; + +const NUL_ERR: io::Error = + io::const_io_error!(io::ErrorKind::InvalidInput, "file name contained an unexpected NUL byte"); + +#[inline] +pub fn run_path_with_cstr<T, F>(path: &Path, f: F) -> io::Result<T> +where + F: FnOnce(&CStr) -> io::Result<T>, +{ + run_with_cstr(path.as_os_str().bytes(), f) +} + +#[inline] +pub fn run_with_cstr<T, F>(bytes: &[u8], f: F) -> io::Result<T> +where + F: FnOnce(&CStr) -> io::Result<T>, +{ + if bytes.len() >= MAX_STACK_ALLOCATION { + return run_with_cstr_allocating(bytes, f); + } + + let mut buf = MaybeUninit::<[u8; MAX_STACK_ALLOCATION]>::uninit(); + let buf_ptr = buf.as_mut_ptr() as *mut u8; + + unsafe { + ptr::copy_nonoverlapping(bytes.as_ptr(), buf_ptr, bytes.len()); + buf_ptr.add(bytes.len()).write(0); + } + + match CStr::from_bytes_with_nul(unsafe { slice::from_raw_parts(buf_ptr, bytes.len() + 1) }) { + Ok(s) => f(s), + Err(_) => Err(NUL_ERR), + } +} + +#[cold] +#[inline(never)] +fn run_with_cstr_allocating<T, F>(bytes: &[u8], f: F) -> io::Result<T> +where + F: FnOnce(&CStr) -> io::Result<T>, +{ + match CString::new(bytes) { + Ok(s) => f(&s), + Err(_) => Err(NUL_ERR), + } +} diff --git a/library/std/src/sys/common/tests.rs b/library/std/src/sys/common/tests.rs new file mode 100644 index 000000000..fb6f5d6af --- /dev/null +++ b/library/std/src/sys/common/tests.rs @@ -0,0 +1,66 @@ +use crate::ffi::CString; +use crate::hint::black_box; +use crate::path::Path; +use crate::sys::common::small_c_string::run_path_with_cstr; +use core::iter::repeat; + +#[test] +fn stack_allocation_works() { + let path = Path::new("abc"); + let result = run_path_with_cstr(path, |p| { + assert_eq!(p, &*CString::new(path.as_os_str().bytes()).unwrap()); + Ok(42) + }); + assert_eq!(result.unwrap(), 42); +} + +#[test] +fn stack_allocation_fails() { + let path = Path::new("ab\0"); + assert!(run_path_with_cstr::<(), _>(path, |_| unreachable!()).is_err()); +} + +#[test] +fn heap_allocation_works() { + let path = repeat("a").take(384).collect::<String>(); + let path = Path::new(&path); + let result = run_path_with_cstr(path, |p| { + assert_eq!(p, &*CString::new(path.as_os_str().bytes()).unwrap()); + Ok(42) + }); + assert_eq!(result.unwrap(), 42); +} + +#[test] +fn heap_allocation_fails() { + let mut path = repeat("a").take(384).collect::<String>(); + path.push('\0'); + let path = Path::new(&path); + assert!(run_path_with_cstr::<(), _>(path, |_| unreachable!()).is_err()); +} + +#[bench] +fn bench_stack_path_alloc(b: &mut test::Bencher) { + let path = repeat("a").take(383).collect::<String>(); + let p = Path::new(&path); + b.iter(|| { + run_path_with_cstr(p, |cstr| { + black_box(cstr); + Ok(()) + }) + .unwrap(); + }); +} + +#[bench] +fn bench_heap_path_alloc(b: &mut test::Bencher) { + let path = repeat("a").take(384).collect::<String>(); + let p = Path::new(&path); + b.iter(|| { + run_path_with_cstr(p, |cstr| { + black_box(cstr); + Ok(()) + }) + .unwrap(); + }); +} diff --git a/library/std/src/sys/hermit/args.rs b/library/std/src/sys/hermit/args.rs index 1c7e1dd8d..afcae6c90 100644 --- a/library/std/src/sys/hermit/args.rs +++ b/library/std/src/sys/hermit/args.rs @@ -1,20 +1,37 @@ -use crate::ffi::OsString; +use crate::ffi::{c_char, CStr, OsString}; use crate::fmt; +use crate::os::unix::ffi::OsStringExt; +use crate::ptr; +use crate::sync::atomic::{ + AtomicIsize, AtomicPtr, + Ordering::{Acquire, Relaxed, Release}, +}; use crate::vec; +static ARGC: AtomicIsize = AtomicIsize::new(0); +static ARGV: AtomicPtr<*const u8> = AtomicPtr::new(ptr::null_mut()); + /// One-time global initialization. pub unsafe fn init(argc: isize, argv: *const *const u8) { - imp::init(argc, argv) -} - -/// One-time global cleanup. -pub unsafe fn cleanup() { - imp::cleanup() + ARGC.store(argc, Relaxed); + // Use release ordering here to broadcast writes by the OS. + ARGV.store(argv as *mut *const u8, Release); } /// Returns the command line arguments pub fn args() -> Args { - imp::args() + // Synchronize with the store above. + let argv = ARGV.load(Acquire); + // If argv has not been initialized yet, do not return any arguments. + let argc = if argv.is_null() { 0 } else { ARGC.load(Relaxed) }; + let args: Vec<OsString> = (0..argc) + .map(|i| unsafe { + let cstr = CStr::from_ptr(*argv.offset(i) as *const c_char); + OsStringExt::from_vec(cstr.to_bytes().to_vec()) + }) + .collect(); + + Args { iter: args.into_iter() } } pub struct Args { @@ -51,44 +68,3 @@ impl DoubleEndedIterator for Args { self.iter.next_back() } } - -mod imp { - use super::Args; - use crate::ffi::{CStr, OsString}; - use crate::os::unix::ffi::OsStringExt; - use crate::ptr; - - use crate::sys_common::mutex::StaticMutex; - - static mut ARGC: isize = 0; - static mut ARGV: *const *const u8 = ptr::null(); - static LOCK: StaticMutex = StaticMutex::new(); - - pub unsafe fn init(argc: isize, argv: *const *const u8) { - let _guard = LOCK.lock(); - ARGC = argc; - ARGV = argv; - } - - pub unsafe fn cleanup() { - let _guard = LOCK.lock(); - ARGC = 0; - ARGV = ptr::null(); - } - - pub fn args() -> Args { - Args { iter: clone().into_iter() } - } - - fn clone() -> Vec<OsString> { - unsafe { - let _guard = LOCK.lock(); - (0..ARGC) - .map(|i| { - let cstr = CStr::from_ptr(*ARGV.offset(i) as *const i8); - OsStringExt::from_vec(cstr.to_bytes().to_vec()) - }) - .collect() - } - } -} diff --git a/library/std/src/sys/hermit/condvar.rs b/library/std/src/sys/hermit/condvar.rs deleted file mode 100644 index 22059ca0d..000000000 --- a/library/std/src/sys/hermit/condvar.rs +++ /dev/null @@ -1,90 +0,0 @@ -use crate::ffi::c_void; -use crate::ptr; -use crate::sync::atomic::{AtomicUsize, Ordering::SeqCst}; -use crate::sys::hermit::abi; -use crate::sys::locks::Mutex; -use crate::sys_common::lazy_box::{LazyBox, LazyInit}; -use crate::time::Duration; - -// The implementation is inspired by Andrew D. Birrell's paper -// "Implementing Condition Variables with Semaphores" - -pub struct Condvar { - counter: AtomicUsize, - sem1: *const c_void, - sem2: *const c_void, -} - -pub(crate) type MovableCondvar = LazyBox<Condvar>; - -impl LazyInit for Condvar { - fn init() -> Box<Self> { - Box::new(Self::new()) - } -} - -unsafe impl Send for Condvar {} -unsafe impl Sync for Condvar {} - -impl Condvar { - pub fn new() -> Self { - let mut condvar = - Self { counter: AtomicUsize::new(0), sem1: ptr::null(), sem2: ptr::null() }; - unsafe { - let _ = abi::sem_init(&mut condvar.sem1, 0); - let _ = abi::sem_init(&mut condvar.sem2, 0); - } - condvar - } - - pub unsafe fn notify_one(&self) { - if self.counter.load(SeqCst) > 0 { - self.counter.fetch_sub(1, SeqCst); - abi::sem_post(self.sem1); - abi::sem_timedwait(self.sem2, 0); - } - } - - pub unsafe fn notify_all(&self) { - let counter = self.counter.swap(0, SeqCst); - for _ in 0..counter { - abi::sem_post(self.sem1); - } - for _ in 0..counter { - abi::sem_timedwait(self.sem2, 0); - } - } - - pub unsafe fn wait(&self, mutex: &Mutex) { - self.counter.fetch_add(1, SeqCst); - mutex.unlock(); - abi::sem_timedwait(self.sem1, 0); - abi::sem_post(self.sem2); - mutex.lock(); - } - - pub unsafe fn wait_timeout(&self, mutex: &Mutex, dur: Duration) -> bool { - self.counter.fetch_add(1, SeqCst); - mutex.unlock(); - let millis = dur.as_millis().min(u32::MAX as u128) as u32; - - let res = if millis > 0 { - abi::sem_timedwait(self.sem1, millis) - } else { - abi::sem_trywait(self.sem1) - }; - - abi::sem_post(self.sem2); - mutex.lock(); - res == 0 - } -} - -impl Drop for Condvar { - fn drop(&mut self) { - unsafe { - let _ = abi::sem_destroy(self.sem1); - let _ = abi::sem_destroy(self.sem2); - } - } -} diff --git a/library/std/src/sys/hermit/fs.rs b/library/std/src/sys/hermit/fs.rs index fa9a7fb19..af297ff1e 100644 --- a/library/std/src/sys/hermit/fs.rs +++ b/library/std/src/sys/hermit/fs.rs @@ -1,10 +1,12 @@ +use crate::convert::TryFrom; use crate::ffi::{CStr, CString, OsString}; use crate::fmt; use crate::hash::{Hash, Hasher}; use crate::io::{self, Error, ErrorKind}; -use crate::io::{IoSlice, IoSliceMut, ReadBuf, SeekFrom}; +use crate::io::{BorrowedCursor, IoSlice, IoSliceMut, SeekFrom}; use crate::os::unix::ffi::OsStrExt; use crate::path::{Path, PathBuf}; +use crate::sys::common::small_c_string::run_path_with_cstr; use crate::sys::cvt; use crate::sys::hermit::abi; use crate::sys::hermit::abi::{O_APPEND, O_CREAT, O_EXCL, O_RDONLY, O_RDWR, O_TRUNC, O_WRONLY}; @@ -15,10 +17,6 @@ use crate::sys::unsupported; pub use crate::sys_common::fs::{copy, try_exists}; //pub use crate::sys_common::fs::remove_dir_all; -fn cstr(path: &Path) -> io::Result<CString> { - Ok(CString::new(path.as_os_str().as_bytes())?) -} - #[derive(Debug)] pub struct File(FileDesc); @@ -41,6 +39,9 @@ pub struct OpenOptions { mode: i32, } +#[derive(Copy, Clone, Debug, Default)] +pub struct FileTimes {} + pub struct FilePermissions(!); pub struct FileType(!); @@ -110,6 +111,11 @@ impl fmt::Debug for FilePermissions { } } +impl FileTimes { + pub fn set_accessed(&mut self, _t: SystemTime) {} + pub fn set_modified(&mut self, _t: SystemTime) {} +} + impl FileType { pub fn is_dir(&self) -> bool { self.0 @@ -264,8 +270,7 @@ impl OpenOptions { impl File { pub fn open(path: &Path, opts: &OpenOptions) -> io::Result<File> { - let path = cstr(path)?; - File::open_c(&path, opts) + run_path_with_cstr(path, |path| File::open_c(&path, opts)) } pub fn open_c(path: &CStr, opts: &OpenOptions) -> io::Result<File> { @@ -312,8 +317,8 @@ impl File { false } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - crate::io::default_read_buf(|buf| self.read(buf), buf) + pub fn read_buf(&self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + crate::io::default_read_buf(|buf| self.read(buf), cursor) } pub fn write(&self, buf: &[u8]) -> io::Result<usize> { @@ -344,6 +349,10 @@ impl File { pub fn set_permissions(&self, _perm: FilePermissions) -> io::Result<()> { Err(Error::from_raw_os_error(22)) } + + pub fn set_times(&self, _times: FileTimes) -> io::Result<()> { + Err(Error::from_raw_os_error(22)) + } } impl DirBuilder { @@ -361,9 +370,7 @@ pub fn readdir(_p: &Path) -> io::Result<ReadDir> { } pub fn unlink(path: &Path) -> io::Result<()> { - let name = cstr(path)?; - let _ = unsafe { cvt(abi::unlink(name.as_ptr()))? }; - Ok(()) + run_path_with_cstr(path, |path| cvt(unsafe { abi::unlink(path.as_ptr()) }).map(|_| ())) } pub fn rename(_old: &Path, _new: &Path) -> io::Result<()> { diff --git a/library/std/src/sys/hermit/futex.rs b/library/std/src/sys/hermit/futex.rs new file mode 100644 index 000000000..b64c174b0 --- /dev/null +++ b/library/std/src/sys/hermit/futex.rs @@ -0,0 +1,39 @@ +use super::abi; +use crate::ptr::null; +use crate::sync::atomic::AtomicU32; +use crate::time::Duration; + +pub fn futex_wait(futex: &AtomicU32, expected: u32, timeout: Option<Duration>) -> bool { + // Calculate the timeout as a relative timespec. + // + // Overflows are rounded up to an infinite timeout (None). + let timespec = timeout.and_then(|dur| { + Some(abi::timespec { + tv_sec: dur.as_secs().try_into().ok()?, + tv_nsec: dur.subsec_nanos().into(), + }) + }); + + let r = unsafe { + abi::futex_wait( + futex.as_mut_ptr(), + expected, + timespec.as_ref().map_or(null(), |t| t as *const abi::timespec), + abi::FUTEX_RELATIVE_TIMEOUT, + ) + }; + + r != -abi::errno::ETIMEDOUT +} + +#[inline] +pub fn futex_wake(futex: &AtomicU32) -> bool { + unsafe { abi::futex_wake(futex.as_mut_ptr(), 1) > 0 } +} + +#[inline] +pub fn futex_wake_all(futex: &AtomicU32) { + unsafe { + abi::futex_wake(futex.as_mut_ptr(), i32::MAX); + } +} diff --git a/library/std/src/sys/hermit/mod.rs b/library/std/src/sys/hermit/mod.rs index 60b7a973c..e6534df89 100644 --- a/library/std/src/sys/hermit/mod.rs +++ b/library/std/src/sys/hermit/mod.rs @@ -25,6 +25,7 @@ pub mod cmath; pub mod env; pub mod fd; pub mod fs; +pub mod futex; #[path = "../unsupported/io.rs"] pub mod io; pub mod memchr; @@ -45,14 +46,14 @@ pub mod thread_local_dtor; pub mod thread_local_key; pub mod time; -mod condvar; -mod mutex; -mod rwlock; - +#[path = "../unix/locks"] pub mod locks { - pub use super::condvar::*; - pub use super::mutex::*; - pub use super::rwlock::*; + mod futex_condvar; + mod futex_mutex; + mod futex_rwlock; + pub(crate) use futex_condvar::MovableCondvar; + pub(crate) use futex_mutex::{MovableMutex, Mutex}; + pub(crate) use futex_rwlock::{MovableRwLock, RwLock}; } use crate::io::ErrorKind; @@ -98,16 +99,14 @@ pub extern "C" fn __rust_abort() { // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. -pub unsafe fn init(argc: isize, argv: *const *const u8) { +pub unsafe fn init(argc: isize, argv: *const *const u8, _sigpipe: u8) { let _ = net::init(); args::init(argc, argv); } // SAFETY: must be called only once during runtime cleanup. // NOTE: this is not guaranteed to run, for example when the program aborts. -pub unsafe fn cleanup() { - args::cleanup(); -} +pub unsafe fn cleanup() {} #[cfg(not(test))] #[no_mangle] diff --git a/library/std/src/sys/hermit/mutex.rs b/library/std/src/sys/hermit/mutex.rs deleted file mode 100644 index eb15a04ff..000000000 --- a/library/std/src/sys/hermit/mutex.rs +++ /dev/null @@ -1,216 +0,0 @@ -use crate::cell::UnsafeCell; -use crate::collections::VecDeque; -use crate::hint; -use crate::ops::{Deref, DerefMut, Drop}; -use crate::ptr; -use crate::sync::atomic::{AtomicUsize, Ordering}; -use crate::sys::hermit::abi; - -/// This type provides a lock based on busy waiting to realize mutual exclusion -/// -/// # Description -/// -/// This structure behaves a lot like a common mutex. There are some differences: -/// -/// - By using busy waiting, it can be used outside the runtime. -/// - It is a so called ticket lock and is completely fair. -#[cfg_attr(target_arch = "x86_64", repr(align(128)))] -#[cfg_attr(not(target_arch = "x86_64"), repr(align(64)))] -struct Spinlock<T: ?Sized> { - queue: AtomicUsize, - dequeue: AtomicUsize, - data: UnsafeCell<T>, -} - -unsafe impl<T: ?Sized + Send> Sync for Spinlock<T> {} -unsafe impl<T: ?Sized + Send> Send for Spinlock<T> {} - -/// A guard to which the protected data can be accessed -/// -/// When the guard falls out of scope it will release the lock. -struct SpinlockGuard<'a, T: ?Sized + 'a> { - dequeue: &'a AtomicUsize, - data: &'a mut T, -} - -impl<T> Spinlock<T> { - pub const fn new(user_data: T) -> Spinlock<T> { - Spinlock { - queue: AtomicUsize::new(0), - dequeue: AtomicUsize::new(1), - data: UnsafeCell::new(user_data), - } - } - - #[inline] - fn obtain_lock(&self) { - let ticket = self.queue.fetch_add(1, Ordering::SeqCst) + 1; - let mut counter: u16 = 0; - while self.dequeue.load(Ordering::SeqCst) != ticket { - counter += 1; - if counter < 100 { - hint::spin_loop(); - } else { - counter = 0; - unsafe { - abi::yield_now(); - } - } - } - } - - #[inline] - pub unsafe fn lock(&self) -> SpinlockGuard<'_, T> { - self.obtain_lock(); - SpinlockGuard { dequeue: &self.dequeue, data: &mut *self.data.get() } - } -} - -impl<T: ?Sized + Default> Default for Spinlock<T> { - fn default() -> Spinlock<T> { - Spinlock::new(Default::default()) - } -} - -impl<'a, T: ?Sized> Deref for SpinlockGuard<'a, T> { - type Target = T; - fn deref(&self) -> &T { - &*self.data - } -} - -impl<'a, T: ?Sized> DerefMut for SpinlockGuard<'a, T> { - fn deref_mut(&mut self) -> &mut T { - &mut *self.data - } -} - -impl<'a, T: ?Sized> Drop for SpinlockGuard<'a, T> { - /// The dropping of the SpinlockGuard will release the lock it was created from. - fn drop(&mut self) { - self.dequeue.fetch_add(1, Ordering::SeqCst); - } -} - -/// Realize a priority queue for tasks -struct PriorityQueue { - queues: [Option<VecDeque<abi::Tid>>; abi::NO_PRIORITIES], - prio_bitmap: u64, -} - -impl PriorityQueue { - pub const fn new() -> PriorityQueue { - PriorityQueue { - queues: [ - None, None, None, None, None, None, None, None, None, None, None, None, None, None, - None, None, None, None, None, None, None, None, None, None, None, None, None, None, - None, None, None, - ], - prio_bitmap: 0, - } - } - - /// Add a task id by its priority to the queue - pub fn push(&mut self, prio: abi::Priority, id: abi::Tid) { - let i: usize = prio.into().into(); - self.prio_bitmap |= (1 << i) as u64; - if let Some(queue) = &mut self.queues[i] { - queue.push_back(id); - } else { - let mut queue = VecDeque::new(); - queue.push_back(id); - self.queues[i] = Some(queue); - } - } - - fn pop_from_queue(&mut self, queue_index: usize) -> Option<abi::Tid> { - if let Some(queue) = &mut self.queues[queue_index] { - let id = queue.pop_front(); - - if queue.is_empty() { - self.prio_bitmap &= !(1 << queue_index as u64); - } - - id - } else { - None - } - } - - /// Pop the task handle with the highest priority from the queue - pub fn pop(&mut self) -> Option<abi::Tid> { - for i in 0..abi::NO_PRIORITIES { - if self.prio_bitmap & (1 << i) != 0 { - return self.pop_from_queue(i); - } - } - - None - } -} - -struct MutexInner { - locked: bool, - blocked_task: PriorityQueue, -} - -impl MutexInner { - pub const fn new() -> MutexInner { - MutexInner { locked: false, blocked_task: PriorityQueue::new() } - } -} - -pub struct Mutex { - inner: Spinlock<MutexInner>, -} - -pub type MovableMutex = Mutex; - -unsafe impl Send for Mutex {} -unsafe impl Sync for Mutex {} - -impl Mutex { - pub const fn new() -> Mutex { - Mutex { inner: Spinlock::new(MutexInner::new()) } - } - - #[inline] - pub unsafe fn init(&mut self) {} - - #[inline] - pub unsafe fn lock(&self) { - loop { - let mut guard = self.inner.lock(); - if guard.locked == false { - guard.locked = true; - return; - } else { - let prio = abi::get_priority(); - let id = abi::getpid(); - - guard.blocked_task.push(prio, id); - abi::block_current_task(); - drop(guard); - abi::yield_now(); - } - } - } - - #[inline] - pub unsafe fn unlock(&self) { - let mut guard = self.inner.lock(); - guard.locked = false; - if let Some(tid) = guard.blocked_task.pop() { - abi::wakeup_task(tid); - } - } - - #[inline] - pub unsafe fn try_lock(&self) -> bool { - let mut guard = self.inner.lock(); - if guard.locked == false { - guard.locked = true; - } - guard.locked - } -} diff --git a/library/std/src/sys/hermit/net.rs b/library/std/src/sys/hermit/net.rs index 745476171..8a13879d8 100644 --- a/library/std/src/sys/hermit/net.rs +++ b/library/std/src/sys/hermit/net.rs @@ -487,6 +487,4 @@ pub mod netc { #[derive(Copy, Clone)] pub struct sockaddr {} - - pub type socklen_t = usize; } diff --git a/library/std/src/sys/hermit/rwlock.rs b/library/std/src/sys/hermit/rwlock.rs deleted file mode 100644 index 9701bab1f..000000000 --- a/library/std/src/sys/hermit/rwlock.rs +++ /dev/null @@ -1,144 +0,0 @@ -use crate::cell::UnsafeCell; -use crate::sys::locks::{MovableCondvar, Mutex}; -use crate::sys_common::lazy_box::{LazyBox, LazyInit}; - -pub struct RwLock { - lock: Mutex, - cond: MovableCondvar, - state: UnsafeCell<State>, -} - -pub type MovableRwLock = RwLock; - -enum State { - Unlocked, - Reading(usize), - Writing, -} - -unsafe impl Send for RwLock {} -unsafe impl Sync for RwLock {} - -// This rwlock implementation is a relatively simple implementation which has a -// condition variable for readers/writers as well as a mutex protecting the -// internal state of the lock. A current downside of the implementation is that -// unlocking the lock will notify *all* waiters rather than just readers or just -// writers. This can cause lots of "thundering stampede" problems. While -// hopefully correct this implementation is very likely to want to be changed in -// the future. - -impl RwLock { - pub const fn new() -> RwLock { - RwLock { - lock: Mutex::new(), - cond: MovableCondvar::new(), - state: UnsafeCell::new(State::Unlocked), - } - } - - #[inline] - pub unsafe fn read(&self) { - self.lock.lock(); - while !(*self.state.get()).inc_readers() { - self.cond.wait(&self.lock); - } - self.lock.unlock(); - } - - #[inline] - pub unsafe fn try_read(&self) -> bool { - self.lock.lock(); - let ok = (*self.state.get()).inc_readers(); - self.lock.unlock(); - return ok; - } - - #[inline] - pub unsafe fn write(&self) { - self.lock.lock(); - while !(*self.state.get()).inc_writers() { - self.cond.wait(&self.lock); - } - self.lock.unlock(); - } - - #[inline] - pub unsafe fn try_write(&self) -> bool { - self.lock.lock(); - let ok = (*self.state.get()).inc_writers(); - self.lock.unlock(); - return ok; - } - - #[inline] - pub unsafe fn read_unlock(&self) { - self.lock.lock(); - let notify = (*self.state.get()).dec_readers(); - self.lock.unlock(); - if notify { - // FIXME: should only wake up one of these some of the time - self.cond.notify_all(); - } - } - - #[inline] - pub unsafe fn write_unlock(&self) { - self.lock.lock(); - (*self.state.get()).dec_writers(); - self.lock.unlock(); - // FIXME: should only wake up one of these some of the time - self.cond.notify_all(); - } -} - -impl State { - fn inc_readers(&mut self) -> bool { - match *self { - State::Unlocked => { - *self = State::Reading(1); - true - } - State::Reading(ref mut cnt) => { - *cnt += 1; - true - } - State::Writing => false, - } - } - - fn inc_writers(&mut self) -> bool { - match *self { - State::Unlocked => { - *self = State::Writing; - true - } - State::Reading(_) | State::Writing => false, - } - } - - fn dec_readers(&mut self) -> bool { - let zero = match *self { - State::Reading(ref mut cnt) => { - *cnt -= 1; - *cnt == 0 - } - State::Unlocked | State::Writing => invalid(), - }; - if zero { - *self = State::Unlocked; - } - zero - } - - fn dec_writers(&mut self) { - match *self { - State::Writing => {} - State::Unlocked | State::Reading(_) => invalid(), - } - *self = State::Unlocked; - } -} - -fn invalid() -> ! { - panic!("inconsistent rwlock"); -} diff --git a/library/std/src/sys/itron/mutex.rs b/library/std/src/sys/itron/mutex.rs index 715e94c3b..085662e6d 100644 --- a/library/std/src/sys/itron/mutex.rs +++ b/library/std/src/sys/itron/mutex.rs @@ -31,12 +31,6 @@ impl Mutex { Mutex { mtx: SpinIdOnceCell::new() } } - pub unsafe fn init(&mut self) { - // Initialize `self.mtx` eagerly - let id = new_mtx().unwrap_or_else(|e| fail(e, &"acre_mtx")); - unsafe { self.mtx.set_unchecked((id, ())) }; - } - /// Get the inner mutex's ID, which is lazily created. fn raw(&self) -> abi::ID { match self.mtx.get_or_try_init(|| new_mtx().map(|id| (id, ()))) { diff --git a/library/std/src/sys/mod.rs b/library/std/src/sys/mod.rs index 167c918c9..c080c176a 100644 --- a/library/std/src/sys/mod.rs +++ b/library/std/src/sys/mod.rs @@ -22,7 +22,7 @@ #![allow(missing_debug_implementations)] -mod common; +pub mod common; cfg_if::cfg_if! { if #[cfg(unix)] { diff --git a/library/std/src/sys/sgx/abi/thread.rs b/library/std/src/sys/sgx/abi/thread.rs index ef55b821a..2b23e368c 100644 --- a/library/std/src/sys/sgx/abi/thread.rs +++ b/library/std/src/sys/sgx/abi/thread.rs @@ -7,7 +7,11 @@ use fortanix_sgx_abi::Tcs; #[unstable(feature = "sgx_platform", issue = "56975")] pub fn current() -> Tcs { extern "C" { - fn get_tcs_addr() -> Tcs; + fn get_tcs_addr() -> *mut u8; + } + let addr = unsafe { get_tcs_addr() }; + match Tcs::new(addr) { + Some(tcs) => tcs, + None => rtabort!("TCS must not be placed at address zero (this is a linker error)"), } - unsafe { get_tcs_addr() } } diff --git a/library/std/src/sys/sgx/abi/tls/mod.rs b/library/std/src/sys/sgx/abi/tls/mod.rs index 13d96e9a6..09c4ab3d3 100644 --- a/library/std/src/sys/sgx/abi/tls/mod.rs +++ b/library/std/src/sys/sgx/abi/tls/mod.rs @@ -111,6 +111,7 @@ impl Tls { rtabort!("TLS limit exceeded") }; TLS_DESTRUCTOR[index].store(dtor.map_or(0, |f| f as usize), Ordering::Relaxed); + unsafe { Self::current() }.data[index].set(ptr::null_mut()); Key::from_index(index) } diff --git a/library/std/src/sys/sgx/abi/usercalls/alloc.rs b/library/std/src/sys/sgx/abi/usercalls/alloc.rs index ea24fedd0..0d934318c 100644 --- a/library/std/src/sys/sgx/abi/usercalls/alloc.rs +++ b/library/std/src/sys/sgx/abi/usercalls/alloc.rs @@ -56,6 +56,8 @@ unsafe impl UserSafeSized for Usercall {} #[unstable(feature = "sgx_platform", issue = "56975")] unsafe impl UserSafeSized for Return {} #[unstable(feature = "sgx_platform", issue = "56975")] +unsafe impl UserSafeSized for Cancel {} +#[unstable(feature = "sgx_platform", issue = "56975")] unsafe impl<T: UserSafeSized> UserSafeSized for [T; 2] {} /// A type that can be represented in memory as one or more `UserSafeSized`s. @@ -115,7 +117,7 @@ pub unsafe trait UserSafe { /// * the pointer is null. /// * the pointed-to range is not in user memory. unsafe fn check_ptr(ptr: *const Self) { - let is_aligned = |p| -> bool { 0 == (p as usize) & (Self::align_of() - 1) }; + let is_aligned = |p: *const u8| -> bool { p.is_aligned_to(Self::align_of()) }; assert!(is_aligned(ptr as *const u8)); assert!(is_user_range(ptr as _, mem::size_of_val(unsafe { &*ptr }))); @@ -305,6 +307,34 @@ where } } +// Split a memory region ptr..ptr + len into three parts: +// +--------+ +// | small0 | Chunk smaller than 8 bytes +// +--------+ +// | big | Chunk 8-byte aligned, and size a multiple of 8 bytes +// +--------+ +// | small1 | Chunk smaller than 8 bytes +// +--------+ +fn region_as_aligned_chunks(ptr: *const u8, len: usize) -> (usize, usize, usize) { + let small0_size = if ptr.is_aligned_to(8) { 0 } else { 8 - ptr.addr() % 8 }; + let small1_size = (len - small0_size) % 8; + let big_size = len - small0_size - small1_size; + + (small0_size, big_size, small1_size) +} + +unsafe fn copy_quadwords(src: *const u8, dst: *mut u8, len: usize) { + unsafe { + asm!( + "rep movsq (%rsi), (%rdi)", + inout("rcx") len / 8 => _, + inout("rdi") dst => _, + inout("rsi") src => _, + options(att_syntax, nostack, preserves_flags) + ); + } +} + /// Copies `len` bytes of data from enclave pointer `src` to userspace `dst` /// /// This function mitigates stale data vulnerabilities by ensuring all writes to untrusted memory are either: @@ -334,8 +364,8 @@ pub(crate) unsafe fn copy_to_userspace(src: *const u8, dst: *mut u8, len: usize) mfence lfence ", - val = in(reg_byte) *src.offset(off as isize), - dst = in(reg) dst.offset(off as isize), + val = in(reg_byte) *src.add(off), + dst = in(reg) dst.add(off), seg_sel = in(reg) &mut seg_sel, options(nostack, att_syntax) ); @@ -343,34 +373,23 @@ pub(crate) unsafe fn copy_to_userspace(src: *const u8, dst: *mut u8, len: usize) } } - unsafe fn copy_aligned_quadwords_to_userspace(src: *const u8, dst: *mut u8, len: usize) { - unsafe { - asm!( - "rep movsq (%rsi), (%rdi)", - inout("rcx") len / 8 => _, - inout("rdi") dst => _, - inout("rsi") src => _, - options(att_syntax, nostack, preserves_flags) - ); - } - } assert!(!src.is_null()); assert!(!dst.is_null()); assert!(is_enclave_range(src, len)); assert!(is_user_range(dst, len)); assert!(len < isize::MAX as usize); - assert!(!(src as usize).overflowing_add(len).1); - assert!(!(dst as usize).overflowing_add(len).1); + assert!(!src.addr().overflowing_add(len).1); + assert!(!dst.addr().overflowing_add(len).1); if len < 8 { // Can't align on 8 byte boundary: copy safely byte per byte unsafe { copy_bytewise_to_userspace(src, dst, len); } - } else if len % 8 == 0 && dst as usize % 8 == 0 { + } else if len % 8 == 0 && dst.is_aligned_to(8) { // Copying 8-byte aligned quadwords: copy quad word per quad word unsafe { - copy_aligned_quadwords_to_userspace(src, dst, len); + copy_quadwords(src, dst, len); } } else { // Split copies into three parts: @@ -381,25 +400,121 @@ pub(crate) unsafe fn copy_to_userspace(src: *const u8, dst: *mut u8, len: usize) // +--------+ // | small1 | Chunk smaller than 8 bytes // +--------+ + let (small0_size, big_size, small1_size) = region_as_aligned_chunks(dst, len); unsafe { // Copy small0 - let small0_size = (8 - dst as usize % 8) as u8; - let small0_src = src; - let small0_dst = dst; - copy_bytewise_to_userspace(small0_src as _, small0_dst, small0_size as _); + copy_bytewise_to_userspace(src, dst, small0_size); // Copy big - let small1_size = ((len - small0_size as usize) % 8) as u8; - let big_size = len - small0_size as usize - small1_size as usize; - let big_src = src.offset(small0_size as _); - let big_dst = dst.offset(small0_size as _); - copy_aligned_quadwords_to_userspace(big_src as _, big_dst, big_size); + let big_src = src.add(small0_size); + let big_dst = dst.add(small0_size); + copy_quadwords(big_src, big_dst, big_size); // Copy small1 - let small1_src = src.offset(big_size as isize + small0_size as isize); - let small1_dst = dst.offset(big_size as isize + small0_size as isize); - copy_bytewise_to_userspace(small1_src, small1_dst, small1_size as _); + let small1_src = src.add(big_size + small0_size); + let small1_dst = dst.add(big_size + small0_size); + copy_bytewise_to_userspace(small1_src, small1_dst, small1_size); + } + } +} + +/// Copies `len` bytes of data from userspace pointer `src` to enclave pointer `dst` +/// +/// This function mitigates AEPIC leak vulnerabilities by ensuring all reads from untrusted memory are 8-byte aligned +/// +/// # Panics +/// This function panics if: +/// +/// * The `src` pointer is null +/// * The `dst` pointer is null +/// * The `src` memory range is not in user memory +/// * The `dst` memory range is not in enclave memory +/// +/// # References +/// - https://www.intel.com/content/www/us/en/security-center/advisory/intel-sa-00657.html +/// - https://www.intel.com/content/www/us/en/developer/articles/technical/software-security-guidance/advisory-guidance/stale-data-read-from-xapic.html +pub(crate) unsafe fn copy_from_userspace(src: *const u8, dst: *mut u8, len: usize) { + // Copies memory region `src..src + len` to the enclave at `dst`. The source memory region + // is: + // - strictly less than 8 bytes in size and may be + // - located at a misaligned memory location + fn copy_misaligned_chunk_to_enclave(src: *const u8, dst: *mut u8, len: usize) { + let mut tmp_buff = [0u8; 16]; + + unsafe { + // Compute an aligned memory region to read from + // +--------+ <-- aligned_src + aligned_len (8B-aligned) + // | pad1 | + // +--------+ <-- src + len (misaligned) + // | | + // | | + // | | + // +--------+ <-- src (misaligned) + // | pad0 | + // +--------+ <-- aligned_src (8B-aligned) + let pad0_size = src as usize % 8; + let aligned_src = src.sub(pad0_size); + + let pad1_size = 8 - (src.add(len) as usize % 8); + let aligned_len = pad0_size + len + pad1_size; + + debug_assert!(len < 8); + debug_assert_eq!(aligned_src as usize % 8, 0); + debug_assert_eq!(aligned_len % 8, 0); + debug_assert!(aligned_len <= 16); + + // Copy the aligned buffer to a temporary buffer + // Note: copying from a slightly different memory location is a bit odd. In this case it + // can't lead to page faults or inadvertent copying from the enclave as we only ensured + // that the `src` pointer is aligned at an 8 byte boundary. As pages are 4096 bytes + // aligned, `aligned_src` must be on the same page as `src`. A similar argument can be made + // for `src + len` + copy_quadwords(aligned_src as _, tmp_buff.as_mut_ptr(), aligned_len); + + // Copy the correct parts of the temporary buffer to the destination + ptr::copy(tmp_buff.as_ptr().add(pad0_size), dst, len); + } + } + + assert!(!src.is_null()); + assert!(!dst.is_null()); + assert!(is_user_range(src, len)); + assert!(is_enclave_range(dst, len)); + assert!(!(src as usize).overflowing_add(len + 8).1); + assert!(!(dst as usize).overflowing_add(len + 8).1); + + if len < 8 { + copy_misaligned_chunk_to_enclave(src, dst, len); + } else if len % 8 == 0 && src as usize % 8 == 0 { + // Copying 8-byte aligned quadwords: copy quad word per quad word + unsafe { + copy_quadwords(src, dst, len); + } + } else { + // Split copies into three parts: + // +--------+ + // | small0 | Chunk smaller than 8 bytes + // +--------+ + // | big | Chunk 8-byte aligned, and size a multiple of 8 bytes + // +--------+ + // | small1 | Chunk smaller than 8 bytes + // +--------+ + let (small0_size, big_size, small1_size) = region_as_aligned_chunks(dst, len); + + unsafe { + // Copy small0 + copy_misaligned_chunk_to_enclave(src, dst, small0_size); + + // Copy big + let big_src = src.add(small0_size); + let big_dst = dst.add(small0_size); + copy_quadwords(big_src, big_dst, big_size); + + // Copy small1 + let small1_src = src.add(big_size + small0_size); + let small1_dst = dst.add(big_size + small0_size); + copy_misaligned_chunk_to_enclave(small1_src, small1_dst, small1_size); } } } @@ -468,7 +583,7 @@ where pub fn copy_to_enclave(&self, dest: &mut T) { unsafe { assert_eq!(mem::size_of_val(dest), mem::size_of_val(&*self.0.get())); - ptr::copy( + copy_from_userspace( self.0.get() as *const T as *const u8, dest as *mut T as *mut u8, mem::size_of_val(dest), @@ -494,7 +609,11 @@ where { /// Copies the value from user memory into enclave memory. pub fn to_enclave(&self) -> T { - unsafe { ptr::read(self.0.get()) } + unsafe { + let mut data: T = mem::MaybeUninit::uninit().assume_init(); + copy_from_userspace(self.0.get() as _, &mut data as *mut T as _, mem::size_of::<T>()); + data + } } } diff --git a/library/std/src/sys/sgx/abi/usercalls/mod.rs b/library/std/src/sys/sgx/abi/usercalls/mod.rs index 79d1db5e1..e19e84326 100644 --- a/library/std/src/sys/sgx/abi/usercalls/mod.rs +++ b/library/std/src/sys/sgx/abi/usercalls/mod.rs @@ -292,12 +292,17 @@ fn check_os_error(err: Result) -> i32 { } } -trait FromSgxResult { +/// Translate the raw result of an SGX usercall. +#[unstable(feature = "sgx_platform", issue = "56975")] +pub trait FromSgxResult { + /// Return type type Return; + /// Translate the raw result of an SGX usercall. fn from_sgx_result(self) -> IoResult<Self::Return>; } +#[unstable(feature = "sgx_platform", issue = "56975")] impl<T> FromSgxResult for (Result, T) { type Return = T; @@ -310,6 +315,7 @@ impl<T> FromSgxResult for (Result, T) { } } +#[unstable(feature = "sgx_platform", issue = "56975")] impl FromSgxResult for Result { type Return = (); diff --git a/library/std/src/sys/sgx/abi/usercalls/raw.rs b/library/std/src/sys/sgx/abi/usercalls/raw.rs index 4267b96cc..10c1456d4 100644 --- a/library/std/src/sys/sgx/abi/usercalls/raw.rs +++ b/library/std/src/sys/sgx/abi/usercalls/raw.rs @@ -37,14 +37,23 @@ pub unsafe fn do_usercall( (a, b) } -type Register = u64; +/// A value passed or returned in a CPU register. +#[unstable(feature = "sgx_platform", issue = "56975")] +pub type Register = u64; -trait RegisterArgument { +/// Translate a type from/to Register to be used as an argument. +#[unstable(feature = "sgx_platform", issue = "56975")] +pub trait RegisterArgument { + /// Translate a Register to Self. fn from_register(_: Register) -> Self; + /// Translate self to a Register. fn into_register(self) -> Register; } -trait ReturnValue { +/// Translate a pair of Registers to the raw usercall return value. +#[unstable(feature = "sgx_platform", issue = "56975")] +pub trait ReturnValue { + /// Translate a pair of Registers to the raw usercall return value. fn from_registers(call: &'static str, regs: (Register, Register)) -> Self; } @@ -68,6 +77,7 @@ macro_rules! define_usercalls { macro_rules! define_ra { (< $i:ident > $t:ty) => { + #[unstable(feature = "sgx_platform", issue = "56975")] impl<$i> RegisterArgument for $t { fn from_register(a: Register) -> Self { a as _ @@ -78,6 +88,7 @@ macro_rules! define_ra { } }; ($i:ty as $t:ty) => { + #[unstable(feature = "sgx_platform", issue = "56975")] impl RegisterArgument for $t { fn from_register(a: Register) -> Self { a as $i as _ @@ -88,6 +99,7 @@ macro_rules! define_ra { } }; ($t:ty) => { + #[unstable(feature = "sgx_platform", issue = "56975")] impl RegisterArgument for $t { fn from_register(a: Register) -> Self { a as _ @@ -112,6 +124,7 @@ define_ra!(usize as isize); define_ra!(<T> *const T); define_ra!(<T> *mut T); +#[unstable(feature = "sgx_platform", issue = "56975")] impl RegisterArgument for bool { fn from_register(a: Register) -> bool { if a != 0 { true } else { false } @@ -121,6 +134,7 @@ impl RegisterArgument for bool { } } +#[unstable(feature = "sgx_platform", issue = "56975")] impl<T: RegisterArgument> RegisterArgument for Option<NonNull<T>> { fn from_register(a: Register) -> Option<NonNull<T>> { NonNull::new(a as _) @@ -130,12 +144,14 @@ impl<T: RegisterArgument> RegisterArgument for Option<NonNull<T>> { } } +#[unstable(feature = "sgx_platform", issue = "56975")] impl ReturnValue for ! { fn from_registers(call: &'static str, _regs: (Register, Register)) -> Self { rtabort!("Usercall {call}: did not expect to be re-entered"); } } +#[unstable(feature = "sgx_platform", issue = "56975")] impl ReturnValue for () { fn from_registers(call: &'static str, usercall_retval: (Register, Register)) -> Self { rtassert!(usercall_retval.0 == 0); @@ -144,6 +160,7 @@ impl ReturnValue for () { } } +#[unstable(feature = "sgx_platform", issue = "56975")] impl<T: RegisterArgument> ReturnValue for T { fn from_registers(call: &'static str, usercall_retval: (Register, Register)) -> Self { rtassert!(usercall_retval.1 == 0); @@ -151,6 +168,7 @@ impl<T: RegisterArgument> ReturnValue for T { } } +#[unstable(feature = "sgx_platform", issue = "56975")] impl<T: RegisterArgument, U: RegisterArgument> ReturnValue for (T, U) { fn from_registers(_call: &'static str, regs: (Register, Register)) -> Self { (T::from_register(regs.0), U::from_register(regs.1)) diff --git a/library/std/src/sys/sgx/abi/usercalls/tests.rs b/library/std/src/sys/sgx/abi/usercalls/tests.rs index cbf7d7d54..58b8eb215 100644 --- a/library/std/src/sys/sgx/abi/usercalls/tests.rs +++ b/library/std/src/sys/sgx/abi/usercalls/tests.rs @@ -1,8 +1,8 @@ -use super::alloc::copy_to_userspace; use super::alloc::User; +use super::alloc::{copy_from_userspace, copy_to_userspace}; #[test] -fn test_copy_function() { +fn test_copy_to_userspace_function() { let mut src = [0u8; 100]; let mut dst = User::<[u8]>::uninitialized(100); @@ -17,12 +17,38 @@ fn test_copy_function() { dst.copy_from_enclave(&[0u8; 100]); // Copy src[0..size] to dst + offset - unsafe { copy_to_userspace(src.as_ptr(), dst.as_mut_ptr().offset(offset), size) }; + unsafe { copy_to_userspace(src.as_ptr(), dst.as_mut_ptr().add(offset), size) }; // Verify copy for byte in 0..size { unsafe { - assert_eq!(*dst.as_ptr().offset(offset + byte as isize), src[byte as usize]); + assert_eq!(*dst.as_ptr().add(offset + byte), src[byte as usize]); + } + } + } + } +} + +#[test] +fn test_copy_from_userspace_function() { + let mut dst = [0u8; 100]; + let mut src = User::<[u8]>::uninitialized(100); + + src.copy_from_enclave(&[0u8; 100]); + + for size in 0..48 { + // For all possible alignment + for offset in 0..8 { + // overwrite complete dst + dst = [0u8; 100]; + + // Copy src[0..size] to dst + offset + unsafe { copy_from_userspace(src.as_ptr().offset(offset), dst.as_mut_ptr(), size) }; + + // Verify copy + for byte in 0..size { + unsafe { + assert_eq!(dst[byte as usize], *src.as_ptr().offset(offset + byte as isize)); } } } diff --git a/library/std/src/sys/sgx/mod.rs b/library/std/src/sys/sgx/mod.rs index 696400670..b1d32929e 100644 --- a/library/std/src/sys/sgx/mod.rs +++ b/library/std/src/sys/sgx/mod.rs @@ -47,7 +47,7 @@ pub mod locks { // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. -pub unsafe fn init(argc: isize, argv: *const *const u8) { +pub unsafe fn init(argc: isize, argv: *const *const u8, _sigpipe: u8) { unsafe { args::init(argc, argv); } diff --git a/library/std/src/sys/sgx/mutex.rs b/library/std/src/sys/sgx/mutex.rs index 513cd77fd..aa747d56b 100644 --- a/library/std/src/sys/sgx/mutex.rs +++ b/library/std/src/sys/sgx/mutex.rs @@ -21,9 +21,6 @@ impl Mutex { } #[inline] - pub unsafe fn init(&mut self) {} - - #[inline] pub unsafe fn lock(&self) { let mut guard = self.inner.lock(); if *guard.lock_var() { diff --git a/library/std/src/sys/sgx/thread_local_key.rs b/library/std/src/sys/sgx/thread_local_key.rs index b21784475..c7a57d3a3 100644 --- a/library/std/src/sys/sgx/thread_local_key.rs +++ b/library/std/src/sys/sgx/thread_local_key.rs @@ -21,8 +21,3 @@ pub unsafe fn get(key: Key) -> *mut u8 { pub unsafe fn destroy(key: Key) { Tls::destroy(AbiKey::from_usize(key)) } - -#[inline] -pub fn requires_synchronized_create() -> bool { - false -} diff --git a/library/std/src/sys/solid/fs.rs b/library/std/src/sys/solid/fs.rs index a2cbee4dc..6c66b93a3 100644 --- a/library/std/src/sys/solid/fs.rs +++ b/library/std/src/sys/solid/fs.rs @@ -2,7 +2,7 @@ use super::{abi, error}; use crate::{ ffi::{CStr, CString, OsStr, OsString}, fmt, - io::{self, IoSlice, IoSliceMut, ReadBuf, SeekFrom}, + io::{self, BorrowedCursor, IoSlice, IoSliceMut, SeekFrom}, mem::MaybeUninit, os::raw::{c_int, c_short}, os::solid::ffi::OsStrExt, @@ -77,6 +77,9 @@ pub struct OpenOptions { custom_flags: i32, } +#[derive(Copy, Clone, Debug, Default)] +pub struct FileTimes {} + #[derive(Clone, PartialEq, Eq, Debug)] pub struct FilePermissions(c_short); @@ -126,6 +129,11 @@ impl FilePermissions { } } +impl FileTimes { + pub fn set_accessed(&mut self, _t: SystemTime) {} + pub fn set_modified(&mut self, _t: SystemTime) {} +} + impl FileType { pub fn is_dir(&self) -> bool { self.is(abi::S_IFDIR) @@ -167,15 +175,19 @@ impl Iterator for ReadDir { type Item = io::Result<DirEntry>; fn next(&mut self) -> Option<io::Result<DirEntry>> { - unsafe { - let mut out_dirent = MaybeUninit::uninit(); - error::SolidError::err_if_negative(abi::SOLID_FS_ReadDir( + let entry = unsafe { + let mut out_entry = MaybeUninit::uninit(); + match error::SolidError::err_if_negative(abi::SOLID_FS_ReadDir( self.inner.dirp, - out_dirent.as_mut_ptr(), - )) - .ok()?; - Some(Ok(DirEntry { entry: out_dirent.assume_init(), inner: Arc::clone(&self.inner) })) - } + out_entry.as_mut_ptr(), + )) { + Ok(_) => out_entry.assume_init(), + Err(e) if e.as_raw() == abi::SOLID_ERR_NOTFOUND => return None, + Err(e) => return Some(Err(e.as_io_error())), + } + }; + + (entry.d_name[0] != 0).then(|| Ok(DirEntry { entry, inner: Arc::clone(&self.inner) })) } } @@ -358,13 +370,13 @@ impl File { } } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { + pub fn read_buf(&self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> { unsafe { - let len = buf.remaining(); + let len = cursor.capacity(); let mut out_num_bytes = MaybeUninit::uninit(); error::SolidError::err_if_negative(abi::SOLID_FS_Read( self.fd.raw(), - buf.unfilled_mut().as_mut_ptr() as *mut u8, + cursor.as_mut().as_mut_ptr() as *mut u8, len, out_num_bytes.as_mut_ptr(), )) @@ -376,9 +388,7 @@ impl File { // Safety: `num_bytes_read` bytes were written to the unfilled // portion of the buffer - buf.assume_init(num_bytes_read); - - buf.add_filled(num_bytes_read); + cursor.advance(num_bytes_read); Ok(()) } @@ -452,6 +462,10 @@ impl File { pub fn set_permissions(&self, _perm: FilePermissions) -> io::Result<()> { unsupported() } + + pub fn set_times(&self, _times: FileTimes) -> io::Result<()> { + unsupported() + } } impl Drop for File { diff --git a/library/std/src/sys/solid/mod.rs b/library/std/src/sys/solid/mod.rs index 778a589d1..5867979a2 100644 --- a/library/std/src/sys/solid/mod.rs +++ b/library/std/src/sys/solid/mod.rs @@ -56,7 +56,7 @@ pub mod locks { // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. -pub unsafe fn init(_argc: isize, _argv: *const *const u8) {} +pub unsafe fn init(_argc: isize, _argv: *const *const u8, _sigpipe: u8) {} // SAFETY: must be called only once during runtime cleanup. pub unsafe fn cleanup() {} diff --git a/library/std/src/sys/solid/os.rs b/library/std/src/sys/solid/os.rs index b5649d6e0..4906c6268 100644 --- a/library/std/src/sys/solid/os.rs +++ b/library/std/src/sys/solid/os.rs @@ -1,4 +1,5 @@ use super::unsupported; +use crate::convert::TryFrom; use crate::error::Error as StdError; use crate::ffi::{CStr, CString, OsStr, OsString}; use crate::fmt; @@ -8,7 +9,8 @@ use crate::os::{ solid::ffi::{OsStrExt, OsStringExt}, }; use crate::path::{self, PathBuf}; -use crate::sys_common::rwlock::StaticRwLock; +use crate::sync::RwLock; +use crate::sys::common::small_c_string::run_with_cstr; use crate::vec; use super::{error, itron, memchr}; @@ -78,7 +80,7 @@ pub fn current_exe() -> io::Result<PathBuf> { unsupported() } -static ENV_LOCK: StaticRwLock = StaticRwLock::new(); +static ENV_LOCK: RwLock<()> = RwLock::new(()); pub struct Env { iter: vec::IntoIter<(OsString, OsString)>, @@ -139,35 +141,33 @@ pub fn env() -> Env { pub fn getenv(k: &OsStr) -> Option<OsString> { // environment variables with a nul byte can't be set, so their value is // always None as well - let k = CString::new(k.as_bytes()).ok()?; - unsafe { + let s = run_with_cstr(k.as_bytes(), |k| { let _guard = ENV_LOCK.read(); - let s = libc::getenv(k.as_ptr()) as *const libc::c_char; - if s.is_null() { - None - } else { - Some(OsStringExt::from_vec(CStr::from_ptr(s).to_bytes().to_vec())) - } + Ok(unsafe { libc::getenv(k.as_ptr()) } as *const libc::c_char) + }) + .ok()?; + + if s.is_null() { + None + } else { + Some(OsStringExt::from_vec(unsafe { CStr::from_ptr(s) }.to_bytes().to_vec())) } } pub fn setenv(k: &OsStr, v: &OsStr) -> io::Result<()> { - let k = CString::new(k.as_bytes())?; - let v = CString::new(v.as_bytes())?; - - unsafe { - let _guard = ENV_LOCK.write(); - cvt_env(libc::setenv(k.as_ptr(), v.as_ptr(), 1)).map(drop) - } + run_with_cstr(k.as_bytes(), |k| { + run_with_cstr(v.as_bytes(), |v| { + let _guard = ENV_LOCK.write(); + cvt_env(unsafe { libc::setenv(k.as_ptr(), v.as_ptr(), 1) }).map(drop) + }) + }) } pub fn unsetenv(n: &OsStr) -> io::Result<()> { - let nbuf = CString::new(n.as_bytes())?; - - unsafe { + run_with_cstr(n.as_bytes(), |nbuf| { let _guard = ENV_LOCK.write(); - cvt_env(libc::unsetenv(nbuf.as_ptr())).map(drop) - } + cvt_env(unsafe { libc::unsetenv(nbuf.as_ptr()) }).map(drop) + }) } /// In kmclib, `setenv` and `unsetenv` don't always set `errno`, so this diff --git a/library/std/src/sys/solid/thread_local_key.rs b/library/std/src/sys/solid/thread_local_key.rs index b17521f70..b37bf9996 100644 --- a/library/std/src/sys/solid/thread_local_key.rs +++ b/library/std/src/sys/solid/thread_local_key.rs @@ -19,8 +19,3 @@ pub unsafe fn get(_key: Key) -> *mut u8 { pub unsafe fn destroy(_key: Key) { panic!("should not be used on the solid target"); } - -#[inline] -pub fn requires_synchronized_create() -> bool { - panic!("should not be used on the solid target"); -} diff --git a/library/std/src/sys/unix/fd.rs b/library/std/src/sys/unix/fd.rs index 30812dabb..dbaa3c33e 100644 --- a/library/std/src/sys/unix/fd.rs +++ b/library/std/src/sys/unix/fd.rs @@ -4,7 +4,7 @@ mod tests; use crate::cmp; -use crate::io::{self, IoSlice, IoSliceMut, Read, ReadBuf}; +use crate::io::{self, BorrowedCursor, IoSlice, IoSliceMut, Read}; use crate::os::unix::io::{AsFd, AsRawFd, BorrowedFd, FromRawFd, IntoRawFd, OwnedFd, RawFd}; use crate::sys::cvt; use crate::sys_common::{AsInner, FromInner, IntoInner}; @@ -131,20 +131,19 @@ impl FileDesc { } } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { + pub fn read_buf(&self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> { let ret = cvt(unsafe { libc::read( self.as_raw_fd(), - buf.unfilled_mut().as_mut_ptr() as *mut libc::c_void, - cmp::min(buf.remaining(), READ_LIMIT), + cursor.as_mut().as_mut_ptr() as *mut libc::c_void, + cmp::min(cursor.capacity(), READ_LIMIT), ) })?; // Safety: `ret` bytes were written to the initialized portion of the buffer unsafe { - buf.assume_init(ret as usize); + cursor.advance(ret as usize); } - buf.add_filled(ret as usize); Ok(()) } diff --git a/library/std/src/sys/unix/fs.rs b/library/std/src/sys/unix/fs.rs index b5cc8038c..37a49f2d7 100644 --- a/library/std/src/sys/unix/fs.rs +++ b/library/std/src/sys/unix/fs.rs @@ -1,13 +1,26 @@ +// miri has some special hacks here that make things unused. +#![cfg_attr(miri, allow(unused))] + use crate::os::unix::prelude::*; -use crate::ffi::{CStr, CString, OsStr, OsString}; +use crate::ffi::{CStr, OsStr, OsString}; use crate::fmt; -use crate::io::{self, Error, IoSlice, IoSliceMut, ReadBuf, SeekFrom}; +use crate::io::{self, BorrowedCursor, Error, IoSlice, IoSliceMut, SeekFrom}; use crate::mem; +#[cfg(any( + target_os = "android", + target_os = "linux", + target_os = "solaris", + target_os = "fuchsia", + target_os = "redox", + target_os = "illumos" +))] +use crate::mem::MaybeUninit; use crate::os::unix::io::{AsFd, AsRawFd, BorrowedFd, FromRawFd, IntoRawFd}; use crate::path::{Path, PathBuf}; use crate::ptr; use crate::sync::Arc; +use crate::sys::common::small_c_string::run_path_with_cstr; use crate::sys::fd::FileDesc; use crate::sys::time::SystemTime; use crate::sys::{cvt, cvt_r}; @@ -260,7 +273,7 @@ pub struct DirEntry { // We need to store an owned copy of the entry name on platforms that use // readdir() (not readdir_r()), because a) struct dirent may use a flexible // array to store the name, b) it lives only until the next readdir() call. - name: CString, + name: crate::ffi::CString, } // Define a minimal subset of fields we need from `dirent64`, especially since @@ -313,8 +326,11 @@ pub struct FilePermissions { mode: mode_t, } -#[derive(Copy, Clone)] -pub struct FileTimes([libc::timespec; 2]); +#[derive(Copy, Clone, Debug, Default)] +pub struct FileTimes { + accessed: Option<SystemTime>, + modified: Option<SystemTime>, +} #[derive(Copy, Clone, PartialEq, Eq, Hash, Debug)] pub struct FileType { @@ -512,45 +528,11 @@ impl FilePermissions { impl FileTimes { pub fn set_accessed(&mut self, t: SystemTime) { - self.0[0] = t.t.to_timespec().expect("Invalid system time"); + self.accessed = Some(t); } pub fn set_modified(&mut self, t: SystemTime) { - self.0[1] = t.t.to_timespec().expect("Invalid system time"); - } -} - -struct TimespecDebugAdapter<'a>(&'a libc::timespec); - -impl fmt::Debug for TimespecDebugAdapter<'_> { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - f.debug_struct("timespec") - .field("tv_sec", &self.0.tv_sec) - .field("tv_nsec", &self.0.tv_nsec) - .finish() - } -} - -impl fmt::Debug for FileTimes { - fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - f.debug_struct("FileTimes") - .field("accessed", &TimespecDebugAdapter(&self.0[0])) - .field("modified", &TimespecDebugAdapter(&self.0[1])) - .finish() - } -} - -impl Default for FileTimes { - fn default() -> Self { - // Redox doesn't appear to support `UTIME_OMIT`, so we stub it out here, and always return - // an error in `set_times`. - // ESP-IDF does not support `futimens` at all and the behavior for that OS is therefore - // the same as for Redox. - #[cfg(any(target_os = "redox", target_os = "espidf"))] - let omit = libc::timespec { tv_sec: 0, tv_nsec: 0 }; - #[cfg(not(any(target_os = "redox", target_os = "espidf")))] - let omit = libc::timespec { tv_sec: 0, tv_nsec: libc::UTIME_OMIT as _ }; - Self([omit; 2]) + self.modified = Some(t); } } @@ -614,33 +596,69 @@ impl Iterator for ReadDir { }; } - // Only d_reclen bytes of *entry_ptr are valid, so we can't just copy the - // whole thing (#93384). Instead, copy everything except the name. - let mut copy: dirent64 = mem::zeroed(); - // Can't dereference entry_ptr, so use the local entry to get - // offsetof(struct dirent, d_name) - let copy_bytes = &mut copy as *mut _ as *mut u8; - let copy_name = &mut copy.d_name as *mut _ as *mut u8; - let name_offset = copy_name.offset_from(copy_bytes) as usize; - let entry_bytes = entry_ptr as *const u8; - let entry_name = entry_bytes.add(name_offset); - ptr::copy_nonoverlapping(entry_bytes, copy_bytes, name_offset); + // The dirent64 struct is a weird imaginary thing that isn't ever supposed + // to be worked with by value. Its trailing d_name field is declared + // variously as [c_char; 256] or [c_char; 1] on different systems but + // either way that size is meaningless; only the offset of d_name is + // meaningful. The dirent64 pointers that libc returns from readdir64 are + // allowed to point to allocations smaller _or_ LARGER than implied by the + // definition of the struct. + // + // As such, we need to be even more careful with dirent64 than if its + // contents were "simply" partially initialized data. + // + // Like for uninitialized contents, converting entry_ptr to `&dirent64` + // would not be legal. However, unique to dirent64 is that we don't even + // get to use `addr_of!((*entry_ptr).d_name)` because that operation + // requires the full extent of *entry_ptr to be in bounds of the same + // allocation, which is not necessarily the case here. + // + // Absent any other way to obtain a pointer to `(*entry_ptr).d_name` + // legally in Rust analogously to how it would be done in C, we instead + // need to make our own non-libc allocation that conforms to the weird + // imaginary definition of dirent64, and use that for a field offset + // computation. + macro_rules! offset_ptr { + ($entry_ptr:expr, $field:ident) => {{ + const OFFSET: isize = { + let delusion = MaybeUninit::<dirent64>::uninit(); + let entry_ptr = delusion.as_ptr(); + unsafe { + ptr::addr_of!((*entry_ptr).$field) + .cast::<u8>() + .offset_from(entry_ptr.cast::<u8>()) + } + }; + if true { + // Cast to the same type determined by the else branch. + $entry_ptr.byte_offset(OFFSET).cast::<_>() + } else { + #[allow(deref_nullptr)] + { + ptr::addr_of!((*ptr::null::<dirent64>()).$field) + } + } + }}; + } + + // d_name is guaranteed to be null-terminated. + let name = CStr::from_ptr(offset_ptr!(entry_ptr, d_name).cast()); + let name_bytes = name.to_bytes(); + if name_bytes == b"." || name_bytes == b".." { + continue; + } let entry = dirent64_min { - d_ino: copy.d_ino as u64, + d_ino: *offset_ptr!(entry_ptr, d_ino) as u64, #[cfg(not(any(target_os = "solaris", target_os = "illumos")))] - d_type: copy.d_type as u8, + d_type: *offset_ptr!(entry_ptr, d_type) as u8, }; - let ret = DirEntry { + return Some(Ok(DirEntry { entry, - // d_name is guaranteed to be null-terminated. - name: CStr::from_ptr(entry_name as *const _).to_owned(), + name: name.to_owned(), dir: Arc::clone(&self.inner), - }; - if ret.name_bytes() != b"." && ret.name_bytes() != b".." { - return Some(Ok(ret)); - } + })); } } } @@ -687,7 +705,11 @@ impl Iterator for ReadDir { impl Drop for Dir { fn drop(&mut self) { let r = unsafe { libc::closedir(self.0) }; - debug_assert_eq!(r, 0); + assert!( + r == 0 || crate::io::Error::last_os_error().kind() == crate::io::ErrorKind::Interrupted, + "unexpected error during closedir: {:?}", + crate::io::Error::last_os_error() + ); } } @@ -700,7 +722,10 @@ impl DirEntry { self.file_name_os_str().to_os_string() } - #[cfg(any(target_os = "linux", target_os = "emscripten", target_os = "android"))] + #[cfg(all( + any(target_os = "linux", target_os = "emscripten", target_os = "android"), + not(miri) + ))] pub fn metadata(&self) -> io::Result<FileAttr> { let fd = cvt(unsafe { dirfd(self.dir.dirp.0) })?; let name = self.name_cstr().as_ptr(); @@ -721,7 +746,10 @@ impl DirEntry { Ok(FileAttr::from_stat64(stat)) } - #[cfg(not(any(target_os = "linux", target_os = "emscripten", target_os = "android")))] + #[cfg(any( + not(any(target_os = "linux", target_os = "emscripten", target_os = "android")), + miri + ))] pub fn metadata(&self) -> io::Result<FileAttr> { lstat(&self.path()) } @@ -925,8 +953,7 @@ impl OpenOptions { impl File { pub fn open(path: &Path, opts: &OpenOptions) -> io::Result<File> { - let path = cstr(path)?; - File::open_c(&path, opts) + run_path_with_cstr(path, |path| File::open_c(path, opts)) } pub fn open_c(path: &CStr, opts: &OpenOptions) -> io::Result<File> { @@ -1031,8 +1058,8 @@ impl File { self.0.read_at(buf, offset) } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - self.0.read_buf(buf) + pub fn read_buf(&self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + self.0.read_buf(cursor) } pub fn write(&self, buf: &[u8]) -> io::Result<usize> { @@ -1078,10 +1105,21 @@ impl File { } pub fn set_times(&self, times: FileTimes) -> io::Result<()> { + #[cfg(not(any(target_os = "redox", target_os = "espidf", target_os = "horizon")))] + let to_timespec = |time: Option<SystemTime>| { + match time { + Some(time) if let Some(ts) = time.t.to_timespec() => Ok(ts), + Some(time) if time > crate::sys::time::UNIX_EPOCH => Err(io::const_io_error!(io::ErrorKind::InvalidInput, "timestamp is too large to set as a file time")), + Some(_) => Err(io::const_io_error!(io::ErrorKind::InvalidInput, "timestamp is too small to set as a file time")), + None => Ok(libc::timespec { tv_sec: 0, tv_nsec: libc::UTIME_OMIT as _ }), + } + }; + #[cfg(not(any(target_os = "redox", target_os = "espidf", target_os = "horizon")))] + let times = [to_timespec(times.accessed)?, to_timespec(times.modified)?]; cfg_if::cfg_if! { - if #[cfg(any(target_os = "redox", target_os = "espidf"))] { + if #[cfg(any(target_os = "redox", target_os = "espidf", target_os = "horizon"))] { // Redox doesn't appear to support `UTIME_OMIT`. - // ESP-IDF does not support `futimens` at all and the behavior for that OS is therefore + // ESP-IDF and HorizonOS do not support `futimens` at all and the behavior for those OS is therefore // the same as for Redox. drop(times); Err(io::const_io_error!( @@ -1093,7 +1131,7 @@ impl File { cvt(unsafe { weak!(fn futimens(c_int, *const libc::timespec) -> c_int); match futimens.get() { - Some(futimens) => futimens(self.as_raw_fd(), times.0.as_ptr()), + Some(futimens) => futimens(self.as_raw_fd(), times.as_ptr()), #[cfg(target_os = "macos")] None => { fn ts_to_tv(ts: &libc::timespec) -> libc::timeval { @@ -1102,7 +1140,7 @@ impl File { tv_usec: (ts.tv_nsec / 1000) as _ } } - let timevals = [ts_to_tv(×.0[0]), ts_to_tv(×.0[1])]; + let timevals = [ts_to_tv(×[0]), ts_to_tv(×[1])]; libc::futimes(self.as_raw_fd(), timevals.as_ptr()) } // futimes requires even newer Android. @@ -1115,7 +1153,7 @@ impl File { })?; Ok(()) } else { - cvt(unsafe { libc::futimens(self.as_raw_fd(), times.0.as_ptr()) })?; + cvt(unsafe { libc::futimens(self.as_raw_fd(), times.as_ptr()) })?; Ok(()) } } @@ -1128,9 +1166,7 @@ impl DirBuilder { } pub fn mkdir(&self, p: &Path) -> io::Result<()> { - let p = cstr(p)?; - cvt(unsafe { libc::mkdir(p.as_ptr(), self.mode) })?; - Ok(()) + run_path_with_cstr(p, |p| cvt(unsafe { libc::mkdir(p.as_ptr(), self.mode) }).map(|_| ())) } pub fn set_mode(&mut self, mode: u32) { @@ -1138,10 +1174,6 @@ impl DirBuilder { } } -fn cstr(path: &Path) -> io::Result<CString> { - Ok(CString::new(path.as_os_str().as_bytes())?) -} - impl AsInner<FileDesc> for File { fn as_inner(&self) -> &FileDesc { &self.0 @@ -1192,7 +1224,12 @@ impl FromRawFd for File { impl fmt::Debug for File { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { - #[cfg(any(target_os = "linux", target_os = "netbsd"))] + #[cfg(any( + target_os = "linux", + target_os = "netbsd", + target_os = "illumos", + target_os = "solaris" + ))] fn get_path(fd: c_int) -> Option<PathBuf> { let mut p = PathBuf::from("/proc/self/fd"); p.push(&fd.to_string()); @@ -1247,14 +1284,23 @@ impl fmt::Debug for File { target_os = "macos", target_os = "vxworks", all(target_os = "freebsd", target_arch = "x86_64"), - target_os = "netbsd" + target_os = "netbsd", + target_os = "illumos", + target_os = "solaris" )))] fn get_path(_fd: c_int) -> Option<PathBuf> { // FIXME(#24570): implement this for other Unix platforms None } - #[cfg(any(target_os = "linux", target_os = "macos", target_os = "vxworks"))] + #[cfg(any( + target_os = "linux", + target_os = "macos", + target_os = "freebsd", + target_os = "netbsd", + target_os = "openbsd", + target_os = "vxworks" + ))] fn get_mode(fd: c_int) -> Option<(bool, bool)> { let mode = unsafe { libc::fcntl(fd, libc::F_GETFL) }; if mode == -1 { @@ -1268,7 +1314,14 @@ impl fmt::Debug for File { } } - #[cfg(not(any(target_os = "linux", target_os = "macos", target_os = "vxworks")))] + #[cfg(not(any( + target_os = "linux", + target_os = "macos", + target_os = "freebsd", + target_os = "netbsd", + target_os = "openbsd", + target_os = "vxworks" + )))] fn get_mode(_fd: c_int) -> Option<(bool, bool)> { // FIXME(#24570): implement this for other Unix platforms None @@ -1287,173 +1340,170 @@ impl fmt::Debug for File { } } -pub fn readdir(p: &Path) -> io::Result<ReadDir> { - let root = p.to_path_buf(); - let p = cstr(p)?; - unsafe { - let ptr = libc::opendir(p.as_ptr()); - if ptr.is_null() { - Err(Error::last_os_error()) - } else { - let inner = InnerReadDir { dirp: Dir(ptr), root }; - Ok(ReadDir { - inner: Arc::new(inner), - #[cfg(not(any( - target_os = "android", - target_os = "linux", - target_os = "solaris", - target_os = "illumos", - target_os = "fuchsia", - target_os = "redox", - )))] - end_of_stream: false, - }) - } +pub fn readdir(path: &Path) -> io::Result<ReadDir> { + let ptr = run_path_with_cstr(path, |p| unsafe { Ok(libc::opendir(p.as_ptr())) })?; + if ptr.is_null() { + Err(Error::last_os_error()) + } else { + let root = path.to_path_buf(); + let inner = InnerReadDir { dirp: Dir(ptr), root }; + Ok(ReadDir { + inner: Arc::new(inner), + #[cfg(not(any( + target_os = "android", + target_os = "linux", + target_os = "solaris", + target_os = "illumos", + target_os = "fuchsia", + target_os = "redox", + )))] + end_of_stream: false, + }) } } pub fn unlink(p: &Path) -> io::Result<()> { - let p = cstr(p)?; - cvt(unsafe { libc::unlink(p.as_ptr()) })?; - Ok(()) + run_path_with_cstr(p, |p| cvt(unsafe { libc::unlink(p.as_ptr()) }).map(|_| ())) } pub fn rename(old: &Path, new: &Path) -> io::Result<()> { - let old = cstr(old)?; - let new = cstr(new)?; - cvt(unsafe { libc::rename(old.as_ptr(), new.as_ptr()) })?; - Ok(()) + run_path_with_cstr(old, |old| { + run_path_with_cstr(new, |new| { + cvt(unsafe { libc::rename(old.as_ptr(), new.as_ptr()) }).map(|_| ()) + }) + }) } pub fn set_perm(p: &Path, perm: FilePermissions) -> io::Result<()> { - let p = cstr(p)?; - cvt_r(|| unsafe { libc::chmod(p.as_ptr(), perm.mode) })?; - Ok(()) + run_path_with_cstr(p, |p| cvt_r(|| unsafe { libc::chmod(p.as_ptr(), perm.mode) }).map(|_| ())) } pub fn rmdir(p: &Path) -> io::Result<()> { - let p = cstr(p)?; - cvt(unsafe { libc::rmdir(p.as_ptr()) })?; - Ok(()) + run_path_with_cstr(p, |p| cvt(unsafe { libc::rmdir(p.as_ptr()) }).map(|_| ())) } pub fn readlink(p: &Path) -> io::Result<PathBuf> { - let c_path = cstr(p)?; - let p = c_path.as_ptr(); + run_path_with_cstr(p, |c_path| { + let p = c_path.as_ptr(); - let mut buf = Vec::with_capacity(256); + let mut buf = Vec::with_capacity(256); - loop { - let buf_read = - cvt(unsafe { libc::readlink(p, buf.as_mut_ptr() as *mut _, buf.capacity()) })? as usize; + loop { + let buf_read = + cvt(unsafe { libc::readlink(p, buf.as_mut_ptr() as *mut _, buf.capacity()) })? + as usize; - unsafe { - buf.set_len(buf_read); - } + unsafe { + buf.set_len(buf_read); + } - if buf_read != buf.capacity() { - buf.shrink_to_fit(); + if buf_read != buf.capacity() { + buf.shrink_to_fit(); - return Ok(PathBuf::from(OsString::from_vec(buf))); - } + return Ok(PathBuf::from(OsString::from_vec(buf))); + } - // Trigger the internal buffer resizing logic of `Vec` by requiring - // more space than the current capacity. The length is guaranteed to be - // the same as the capacity due to the if statement above. - buf.reserve(1); - } + // Trigger the internal buffer resizing logic of `Vec` by requiring + // more space than the current capacity. The length is guaranteed to be + // the same as the capacity due to the if statement above. + buf.reserve(1); + } + }) } pub fn symlink(original: &Path, link: &Path) -> io::Result<()> { - let original = cstr(original)?; - let link = cstr(link)?; - cvt(unsafe { libc::symlink(original.as_ptr(), link.as_ptr()) })?; - Ok(()) + run_path_with_cstr(original, |original| { + run_path_with_cstr(link, |link| { + cvt(unsafe { libc::symlink(original.as_ptr(), link.as_ptr()) }).map(|_| ()) + }) + }) } pub fn link(original: &Path, link: &Path) -> io::Result<()> { - let original = cstr(original)?; - let link = cstr(link)?; - cfg_if::cfg_if! { - if #[cfg(any(target_os = "vxworks", target_os = "redox", target_os = "android", target_os = "espidf", target_os = "horizon"))] { - // VxWorks, Redox and ESP-IDF lack `linkat`, so use `link` instead. POSIX leaves - // it implementation-defined whether `link` follows symlinks, so rely on the - // `symlink_hard_link` test in library/std/src/fs/tests.rs to check the behavior. - // Android has `linkat` on newer versions, but we happen to know `link` - // always has the correct behavior, so it's here as well. - cvt(unsafe { libc::link(original.as_ptr(), link.as_ptr()) })?; - } else if #[cfg(target_os = "macos")] { - // On MacOS, older versions (<=10.9) lack support for linkat while newer - // versions have it. We want to use linkat if it is available, so we use weak! - // to check. `linkat` is preferable to `link` because it gives us a flag to - // specify how symlinks should be handled. We pass 0 as the flags argument, - // meaning it shouldn't follow symlinks. - weak!(fn linkat(c_int, *const c_char, c_int, *const c_char, c_int) -> c_int); - - if let Some(f) = linkat.get() { - cvt(unsafe { f(libc::AT_FDCWD, original.as_ptr(), libc::AT_FDCWD, link.as_ptr(), 0) })?; - } else { - cvt(unsafe { libc::link(original.as_ptr(), link.as_ptr()) })?; - }; - } else { - // Where we can, use `linkat` instead of `link`; see the comment above - // this one for details on why. - cvt(unsafe { libc::linkat(libc::AT_FDCWD, original.as_ptr(), libc::AT_FDCWD, link.as_ptr(), 0) })?; - } - } - Ok(()) + run_path_with_cstr(original, |original| { + run_path_with_cstr(link, |link| { + cfg_if::cfg_if! { + if #[cfg(any(target_os = "vxworks", target_os = "redox", target_os = "android", target_os = "espidf", target_os = "horizon"))] { + // VxWorks, Redox and ESP-IDF lack `linkat`, so use `link` instead. POSIX leaves + // it implementation-defined whether `link` follows symlinks, so rely on the + // `symlink_hard_link` test in library/std/src/fs/tests.rs to check the behavior. + // Android has `linkat` on newer versions, but we happen to know `link` + // always has the correct behavior, so it's here as well. + cvt(unsafe { libc::link(original.as_ptr(), link.as_ptr()) })?; + } else if #[cfg(target_os = "macos")] { + // On MacOS, older versions (<=10.9) lack support for linkat while newer + // versions have it. We want to use linkat if it is available, so we use weak! + // to check. `linkat` is preferable to `link` because it gives us a flag to + // specify how symlinks should be handled. We pass 0 as the flags argument, + // meaning it shouldn't follow symlinks. + weak!(fn linkat(c_int, *const c_char, c_int, *const c_char, c_int) -> c_int); + + if let Some(f) = linkat.get() { + cvt(unsafe { f(libc::AT_FDCWD, original.as_ptr(), libc::AT_FDCWD, link.as_ptr(), 0) })?; + } else { + cvt(unsafe { libc::link(original.as_ptr(), link.as_ptr()) })?; + }; + } else { + // Where we can, use `linkat` instead of `link`; see the comment above + // this one for details on why. + cvt(unsafe { libc::linkat(libc::AT_FDCWD, original.as_ptr(), libc::AT_FDCWD, link.as_ptr(), 0) })?; + } + } + Ok(()) + }) + }) } pub fn stat(p: &Path) -> io::Result<FileAttr> { - let p = cstr(p)?; - - cfg_has_statx! { - if let Some(ret) = unsafe { try_statx( - libc::AT_FDCWD, - p.as_ptr(), - libc::AT_STATX_SYNC_AS_STAT, - libc::STATX_ALL, - ) } { - return ret; + run_path_with_cstr(p, |p| { + cfg_has_statx! { + if let Some(ret) = unsafe { try_statx( + libc::AT_FDCWD, + p.as_ptr(), + libc::AT_STATX_SYNC_AS_STAT, + libc::STATX_ALL, + ) } { + return ret; + } } - } - let mut stat: stat64 = unsafe { mem::zeroed() }; - cvt(unsafe { stat64(p.as_ptr(), &mut stat) })?; - Ok(FileAttr::from_stat64(stat)) + let mut stat: stat64 = unsafe { mem::zeroed() }; + cvt(unsafe { stat64(p.as_ptr(), &mut stat) })?; + Ok(FileAttr::from_stat64(stat)) + }) } pub fn lstat(p: &Path) -> io::Result<FileAttr> { - let p = cstr(p)?; - - cfg_has_statx! { - if let Some(ret) = unsafe { try_statx( - libc::AT_FDCWD, - p.as_ptr(), - libc::AT_SYMLINK_NOFOLLOW | libc::AT_STATX_SYNC_AS_STAT, - libc::STATX_ALL, - ) } { - return ret; + run_path_with_cstr(p, |p| { + cfg_has_statx! { + if let Some(ret) = unsafe { try_statx( + libc::AT_FDCWD, + p.as_ptr(), + libc::AT_SYMLINK_NOFOLLOW | libc::AT_STATX_SYNC_AS_STAT, + libc::STATX_ALL, + ) } { + return ret; + } } - } - let mut stat: stat64 = unsafe { mem::zeroed() }; - cvt(unsafe { lstat64(p.as_ptr(), &mut stat) })?; - Ok(FileAttr::from_stat64(stat)) + let mut stat: stat64 = unsafe { mem::zeroed() }; + cvt(unsafe { lstat64(p.as_ptr(), &mut stat) })?; + Ok(FileAttr::from_stat64(stat)) + }) } pub fn canonicalize(p: &Path) -> io::Result<PathBuf> { - let path = CString::new(p.as_os_str().as_bytes())?; - let buf; - unsafe { - let r = libc::realpath(path.as_ptr(), ptr::null_mut()); - if r.is_null() { - return Err(io::Error::last_os_error()); - } - buf = CStr::from_ptr(r).to_bytes().to_vec(); - libc::free(r as *mut _); + let r = run_path_with_cstr(p, |path| unsafe { + Ok(libc::realpath(path.as_ptr(), ptr::null_mut())) + })?; + if r.is_null() { + return Err(io::Error::last_os_error()); } - Ok(PathBuf::from(OsString::from_vec(buf))) + Ok(PathBuf::from(OsString::from_vec(unsafe { + let buf = CStr::from_ptr(r).to_bytes().to_vec(); + libc::free(r as *mut _); + buf + }))) } fn open_from(from: &Path) -> io::Result<(crate::fs::File, crate::fs::Metadata)> { @@ -1603,9 +1653,9 @@ pub fn copy(from: &Path, to: &Path) -> io::Result<u64> { // Opportunistically attempt to create a copy-on-write clone of `from` // using `fclonefileat`. if HAS_FCLONEFILEAT.load(Ordering::Relaxed) { - let to = cstr(to)?; - let clonefile_result = - cvt(unsafe { fclonefileat(reader.as_raw_fd(), libc::AT_FDCWD, to.as_ptr(), 0) }); + let clonefile_result = run_path_with_cstr(to, |to| { + cvt(unsafe { fclonefileat(reader.as_raw_fd(), libc::AT_FDCWD, to.as_ptr(), 0) }) + }); match clonefile_result { Ok(_) => return Ok(reader_metadata.len()), Err(err) => match err.raw_os_error() { @@ -1649,9 +1699,10 @@ pub fn copy(from: &Path, to: &Path) -> io::Result<u64> { } pub fn chown(path: &Path, uid: u32, gid: u32) -> io::Result<()> { - let path = cstr(path)?; - cvt(unsafe { libc::chown(path.as_ptr(), uid as libc::uid_t, gid as libc::gid_t) })?; - Ok(()) + run_path_with_cstr(path, |path| { + cvt(unsafe { libc::chown(path.as_ptr(), uid as libc::uid_t, gid as libc::gid_t) }) + .map(|_| ()) + }) } pub fn fchown(fd: c_int, uid: u32, gid: u32) -> io::Result<()> { @@ -1660,16 +1711,15 @@ pub fn fchown(fd: c_int, uid: u32, gid: u32) -> io::Result<()> { } pub fn lchown(path: &Path, uid: u32, gid: u32) -> io::Result<()> { - let path = cstr(path)?; - cvt(unsafe { libc::lchown(path.as_ptr(), uid as libc::uid_t, gid as libc::gid_t) })?; - Ok(()) + run_path_with_cstr(path, |path| { + cvt(unsafe { libc::lchown(path.as_ptr(), uid as libc::uid_t, gid as libc::gid_t) }) + .map(|_| ()) + }) } #[cfg(not(any(target_os = "fuchsia", target_os = "vxworks")))] pub fn chroot(dir: &Path) -> io::Result<()> { - let dir = cstr(dir)?; - cvt(unsafe { libc::chroot(dir.as_ptr()) })?; - Ok(()) + run_path_with_cstr(dir, |dir| cvt(unsafe { libc::chroot(dir.as_ptr()) }).map(|_| ())) } pub use remove_dir_impl::remove_dir_all; @@ -1683,13 +1733,14 @@ mod remove_dir_impl { // Modern implementation using openat(), unlinkat() and fdopendir() #[cfg(not(any(target_os = "redox", target_os = "espidf", target_os = "horizon", miri)))] mod remove_dir_impl { - use super::{cstr, lstat, Dir, DirEntry, InnerReadDir, ReadDir}; + use super::{lstat, Dir, DirEntry, InnerReadDir, ReadDir}; use crate::ffi::CStr; use crate::io; use crate::os::unix::io::{AsRawFd, FromRawFd, IntoRawFd}; use crate::os::unix::prelude::{OwnedFd, RawFd}; use crate::path::{Path, PathBuf}; use crate::sync::Arc; + use crate::sys::common::small_c_string::run_path_with_cstr; use crate::sys::{cvt, cvt_r}; #[cfg(not(all(target_os = "macos", not(target_arch = "aarch64")),))] @@ -1856,7 +1907,7 @@ mod remove_dir_impl { if attr.file_type().is_symlink() { crate::fs::remove_file(p) } else { - remove_dir_all_recursive(None, &cstr(p)?) + run_path_with_cstr(p, |p| remove_dir_all_recursive(None, &p)) } } diff --git a/library/std/src/sys/unix/io.rs b/library/std/src/sys/unix/io.rs index deb5ee76b..29c340dd3 100644 --- a/library/std/src/sys/unix/io.rs +++ b/library/std/src/sys/unix/io.rs @@ -1,4 +1,5 @@ use crate::marker::PhantomData; +use crate::os::fd::{AsFd, AsRawFd}; use crate::slice; use libc::{c_void, iovec}; @@ -74,3 +75,8 @@ impl<'a> IoSliceMut<'a> { unsafe { slice::from_raw_parts_mut(self.vec.iov_base as *mut u8, self.vec.iov_len) } } } + +pub fn is_terminal(fd: &impl AsFd) -> bool { + let fd = fd.as_fd(); + unsafe { libc::isatty(fd.as_raw_fd()) != 0 } +} diff --git a/library/std/src/sys/unix/kernel_copy.rs b/library/std/src/sys/unix/kernel_copy.rs index 8f7abb55e..94546ca09 100644 --- a/library/std/src/sys/unix/kernel_copy.rs +++ b/library/std/src/sys/unix/kernel_copy.rs @@ -20,7 +20,7 @@ //! Since those syscalls have requirements that cannot be fully checked in advance and //! gathering additional information about file descriptors would require additional syscalls //! anyway it simply attempts to use them one after another (guided by inaccurate hints) to -//! figure out which one works and and falls back to the generic read-write copy loop if none of them +//! figure out which one works and falls back to the generic read-write copy loop if none of them //! does. //! Once a working syscall is found for a pair of file descriptors it will be called in a loop //! until the copy operation is completed. diff --git a/library/std/src/sys/unix/locks/fuchsia_mutex.rs b/library/std/src/sys/unix/locks/fuchsia_mutex.rs index ce427599c..117611ce4 100644 --- a/library/std/src/sys/unix/locks/fuchsia_mutex.rs +++ b/library/std/src/sys/unix/locks/fuchsia_mutex.rs @@ -86,9 +86,6 @@ impl Mutex { } #[inline] - pub unsafe fn init(&mut self) {} - - #[inline] pub unsafe fn try_lock(&self) -> bool { let thread_self = zx_thread_self(); self.futex.compare_exchange(UNLOCKED, to_state(thread_self), Acquire, Relaxed).is_ok() @@ -138,7 +135,7 @@ impl Mutex { } } - // The state has changed or a wakeup occured, try to lock the mutex. + // The state has changed or a wakeup occurred, try to lock the mutex. match self.futex.compare_exchange(UNLOCKED, owned_state, Acquire, Relaxed) { Ok(_) => return, Err(updated) => state = updated, diff --git a/library/std/src/sys/unix/locks/futex_mutex.rs b/library/std/src/sys/unix/locks/futex_mutex.rs index 99ba86e5f..33b13dad4 100644 --- a/library/std/src/sys/unix/locks/futex_mutex.rs +++ b/library/std/src/sys/unix/locks/futex_mutex.rs @@ -20,9 +20,6 @@ impl Mutex { } #[inline] - pub unsafe fn init(&mut self) {} - - #[inline] pub unsafe fn try_lock(&self) -> bool { self.futex.compare_exchange(0, 1, Acquire, Relaxed).is_ok() } @@ -53,7 +50,7 @@ impl Mutex { // We avoid an unnecessary write if it as already set to 2, // to be friendlier for the caches. if state != 2 && self.futex.swap(2, Acquire) == 0 { - // We changed it from 0 to 2, so we just succesfully locked it. + // We changed it from 0 to 2, so we just successfully locked it. return; } diff --git a/library/std/src/sys/unix/locks/futex_rwlock.rs b/library/std/src/sys/unix/locks/futex_rwlock.rs index b3bbbf743..0cc92244e 100644 --- a/library/std/src/sys/unix/locks/futex_rwlock.rs +++ b/library/std/src/sys/unix/locks/futex_rwlock.rs @@ -54,7 +54,7 @@ fn is_read_lockable(state: u32) -> bool { // We don't allow read-locking if there's readers waiting, even if the lock is unlocked // and there's no writers waiting. The only situation when this happens is after unlocking, // at which point the unlocking thread might be waking up writers, which have priority over readers. - // The unlocking thread will clear the readers waiting bit and wake up readers, if necssary. + // The unlocking thread will clear the readers waiting bit and wake up readers, if necessary. state & MASK < MAX_READERS && !has_readers_waiting(state) && !has_writers_waiting(state) } diff --git a/library/std/src/sys/unix/locks/mod.rs b/library/std/src/sys/unix/locks/mod.rs index f5f92f693..9bb314b70 100644 --- a/library/std/src/sys/unix/locks/mod.rs +++ b/library/std/src/sys/unix/locks/mod.rs @@ -11,21 +11,21 @@ cfg_if::cfg_if! { mod futex_rwlock; mod futex_condvar; pub(crate) use futex_mutex::{Mutex, MovableMutex}; - pub(crate) use futex_rwlock::{RwLock, MovableRwLock}; + pub(crate) use futex_rwlock::MovableRwLock; pub(crate) use futex_condvar::MovableCondvar; } else if #[cfg(target_os = "fuchsia")] { mod fuchsia_mutex; mod futex_rwlock; mod futex_condvar; pub(crate) use fuchsia_mutex::{Mutex, MovableMutex}; - pub(crate) use futex_rwlock::{RwLock, MovableRwLock}; + pub(crate) use futex_rwlock::MovableRwLock; pub(crate) use futex_condvar::MovableCondvar; } else { mod pthread_mutex; mod pthread_rwlock; mod pthread_condvar; pub(crate) use pthread_mutex::{Mutex, MovableMutex}; - pub(crate) use pthread_rwlock::{RwLock, MovableRwLock}; + pub(crate) use pthread_rwlock::MovableRwLock; pub(crate) use pthread_condvar::MovableCondvar; } } diff --git a/library/std/src/sys/unix/locks/pthread_condvar.rs b/library/std/src/sys/unix/locks/pthread_condvar.rs index abf27e7db..4741c0c67 100644 --- a/library/std/src/sys/unix/locks/pthread_condvar.rs +++ b/library/std/src/sys/unix/locks/pthread_condvar.rs @@ -172,7 +172,7 @@ impl Condvar { let mut sys_now = libc::timeval { tv_sec: 0, tv_usec: 0 }; let stable_now = Instant::now(); let r = libc::gettimeofday(&mut sys_now, ptr::null_mut()); - debug_assert_eq!(r, 0); + assert_eq!(r, 0, "unexpected error: {:?}", crate::io::Error::last_os_error()); let nsec = dur.subsec_nanos() as libc::c_long + (sys_now.tv_usec * 1000) as libc::c_long; let extra = (nsec / 1_000_000_000) as libc::time_t; diff --git a/library/std/src/sys/unix/locks/pthread_mutex.rs b/library/std/src/sys/unix/locks/pthread_mutex.rs index 98afee69b..5964935dd 100644 --- a/library/std/src/sys/unix/locks/pthread_mutex.rs +++ b/library/std/src/sys/unix/locks/pthread_mutex.rs @@ -52,7 +52,7 @@ impl Mutex { Mutex { inner: UnsafeCell::new(libc::PTHREAD_MUTEX_INITIALIZER) } } #[inline] - pub unsafe fn init(&mut self) { + unsafe fn init(&mut self) { // Issue #33770 // // A pthread mutex initialized with PTHREAD_MUTEX_INITIALIZER will have diff --git a/library/std/src/sys/unix/mod.rs b/library/std/src/sys/unix/mod.rs index 3d0d91460..9055a011c 100644 --- a/library/std/src/sys/unix/mod.rs +++ b/library/std/src/sys/unix/mod.rs @@ -44,12 +44,13 @@ pub mod thread_parker; pub mod time; #[cfg(target_os = "espidf")] -pub fn init(argc: isize, argv: *const *const u8) {} +pub fn init(argc: isize, argv: *const *const u8, _sigpipe: u8) {} #[cfg(not(target_os = "espidf"))] // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. -pub unsafe fn init(argc: isize, argv: *const *const u8) { +// See `fn init()` in `library/std/src/rt.rs` for docs on `sigpipe`. +pub unsafe fn init(argc: isize, argv: *const *const u8, sigpipe: u8) { // The standard streams might be closed on application startup. To prevent // std::io::{stdin, stdout,stderr} objects from using other unrelated file // resources opened later, we reopen standards streams when they are closed. @@ -61,8 +62,9 @@ pub unsafe fn init(argc: isize, argv: *const *const u8) { // want! // // Hence, we set SIGPIPE to ignore when the program starts up in order - // to prevent this problem. - reset_sigpipe(); + // to prevent this problem. Add `#[unix_sigpipe = "..."]` above `fn main()` to + // alter this behavior. + reset_sigpipe(sigpipe); stack_overflow::init(); args::init(argc, argv); @@ -151,12 +153,64 @@ pub unsafe fn init(argc: isize, argv: *const *const u8) { } } - unsafe fn reset_sigpipe() { + unsafe fn reset_sigpipe(#[allow(unused_variables)] sigpipe: u8) { #[cfg(not(any(target_os = "emscripten", target_os = "fuchsia", target_os = "horizon")))] - rtassert!(signal(libc::SIGPIPE, libc::SIG_IGN) != libc::SIG_ERR); + { + // We don't want to add this as a public type to libstd, nor do we + // want to `include!` a file from the compiler (which would break + // Miri and xargo for example), so we choose to duplicate these + // constants from `compiler/rustc_session/src/config/sigpipe.rs`. + // See the other file for docs. NOTE: Make sure to keep them in + // sync! + mod sigpipe { + pub const DEFAULT: u8 = 0; + pub const INHERIT: u8 = 1; + pub const SIG_IGN: u8 = 2; + pub const SIG_DFL: u8 = 3; + } + + let (sigpipe_attr_specified, handler) = match sigpipe { + sigpipe::DEFAULT => (false, Some(libc::SIG_IGN)), + sigpipe::INHERIT => (true, None), + sigpipe::SIG_IGN => (true, Some(libc::SIG_IGN)), + sigpipe::SIG_DFL => (true, Some(libc::SIG_DFL)), + _ => unreachable!(), + }; + // The bootstrap compiler doesn't know about sigpipe::DEFAULT, and always passes in + // SIG_IGN. This causes some tests to fail because they expect SIGPIPE to be reset to + // default on process spawning (which doesn't happen if #[unix_sigpipe] is specified). + // Since we can't differentiate between the cases here, treat SIG_IGN as DEFAULT + // unconditionally. + if sigpipe_attr_specified && !(cfg!(bootstrap) && sigpipe == sigpipe::SIG_IGN) { + UNIX_SIGPIPE_ATTR_SPECIFIED.store(true, crate::sync::atomic::Ordering::Relaxed); + } + if let Some(handler) = handler { + rtassert!(signal(libc::SIGPIPE, handler) != libc::SIG_ERR); + } + } } } +// This is set (up to once) in reset_sigpipe. +#[cfg(not(any( + target_os = "espidf", + target_os = "emscripten", + target_os = "fuchsia", + target_os = "horizon" +)))] +static UNIX_SIGPIPE_ATTR_SPECIFIED: crate::sync::atomic::AtomicBool = + crate::sync::atomic::AtomicBool::new(false); + +#[cfg(not(any( + target_os = "espidf", + target_os = "emscripten", + target_os = "fuchsia", + target_os = "horizon" +)))] +pub(crate) fn unix_sigpipe_attr_specified() -> bool { + UNIX_SIGPIPE_ATTR_SPECIFIED.load(crate::sync::atomic::Ordering::Relaxed) +} + // SAFETY: must be called only once during runtime cleanup. // NOTE: this is not guaranteed to run, for example when the program aborts. pub unsafe fn cleanup() { @@ -295,8 +349,10 @@ pub fn abort_internal() -> ! { cfg_if::cfg_if! { if #[cfg(target_os = "android")] { - #[link(name = "dl")] - #[link(name = "log")] + #[link(name = "dl", kind = "static", modifiers = "-bundle", + cfg(target_feature = "crt-static"))] + #[link(name = "dl", cfg(not(target_feature = "crt-static")))] + #[link(name = "log", cfg(not(target_feature = "crt-static")))] extern "C" {} } else if #[cfg(target_os = "freebsd")] { #[link(name = "execinfo")] @@ -326,16 +382,12 @@ cfg_if::cfg_if! { extern "C" {} } else if #[cfg(target_os = "macos")] { #[link(name = "System")] - // res_init and friends require -lresolv on macOS/iOS. - // See #41582 and https://blog.achernya.com/2013/03/os-x-has-silly-libsystem.html - #[link(name = "resolv")] extern "C" {} } else if #[cfg(any(target_os = "ios", target_os = "watchos"))] { #[link(name = "System")] #[link(name = "objc")] #[link(name = "Security", kind = "framework")] #[link(name = "Foundation", kind = "framework")] - #[link(name = "resolv")] extern "C" {} } else if #[cfg(target_os = "fuchsia")] { #[link(name = "zircon")] diff --git a/library/std/src/sys/unix/net.rs b/library/std/src/sys/unix/net.rs index 462a45b01..b84bf8f92 100644 --- a/library/std/src/sys/unix/net.rs +++ b/library/std/src/sys/unix/net.rs @@ -393,6 +393,17 @@ impl Socket { } #[cfg(any(target_os = "android", target_os = "linux",))] + pub fn set_quickack(&self, quickack: bool) -> io::Result<()> { + setsockopt(self, libc::IPPROTO_TCP, libc::TCP_QUICKACK, quickack as c_int) + } + + #[cfg(any(target_os = "android", target_os = "linux",))] + pub fn quickack(&self) -> io::Result<bool> { + let raw: c_int = getsockopt(self, libc::IPPROTO_TCP, libc::TCP_QUICKACK)?; + Ok(raw != 0) + } + + #[cfg(any(target_os = "android", target_os = "linux",))] pub fn set_passcred(&self, passcred: bool) -> io::Result<()> { setsockopt(self, libc::SOL_SOCKET, libc::SO_PASSCRED, passcred as libc::c_int) } @@ -427,6 +438,17 @@ impl Socket { self.0.set_nonblocking(nonblocking) } + #[cfg(any(target_os = "linux", target_os = "freebsd", target_os = "openbsd"))] + pub fn set_mark(&self, mark: u32) -> io::Result<()> { + #[cfg(target_os = "linux")] + let option = libc::SO_MARK; + #[cfg(target_os = "freebsd")] + let option = libc::SO_USER_COOKIE; + #[cfg(target_os = "openbsd")] + let option = libc::SO_RTABLE; + setsockopt(self, libc::SOL_SOCKET, option, mark as libc::c_int) + } + pub fn take_error(&self) -> io::Result<Option<io::Error>> { let raw: c_int = getsockopt(self, libc::SOL_SOCKET, libc::SO_ERROR)?; if raw == 0 { Ok(None) } else { Ok(Some(io::Error::from_raw_os_error(raw as i32))) } diff --git a/library/std/src/sys/unix/os.rs b/library/std/src/sys/unix/os.rs index 46545a083..2f2663db6 100644 --- a/library/std/src/sys/unix/os.rs +++ b/library/std/src/sys/unix/os.rs @@ -7,6 +7,7 @@ mod tests; use crate::os::unix::prelude::*; +use crate::convert::TryFrom; use crate::error::Error as StdError; use crate::ffi::{CStr, CString, OsStr, OsString}; use crate::fmt; @@ -17,10 +18,11 @@ use crate::path::{self, PathBuf}; use crate::ptr; use crate::slice; use crate::str; +use crate::sync::{PoisonError, RwLock}; +use crate::sys::common::small_c_string::{run_path_with_cstr, run_with_cstr}; use crate::sys::cvt; use crate::sys::fd; use crate::sys::memchr; -use crate::sys_common::rwlock::{StaticRwLock, StaticRwLockReadGuard}; use crate::vec; #[cfg(all(target_env = "gnu", not(target_os = "vxworks")))] @@ -125,7 +127,9 @@ pub fn error_string(errno: i32) -> String { } let p = p as *const _; - str::from_utf8(CStr::from_ptr(p).to_bytes()).unwrap().to_owned() + // We can't always expect a UTF-8 environment. When we don't get that luxury, + // it's better to give a low-quality error message than none at all. + String::from_utf8_lossy(CStr::from_ptr(p).to_bytes()).into() } } @@ -168,12 +172,8 @@ pub fn chdir(p: &path::Path) -> io::Result<()> { #[cfg(not(target_os = "espidf"))] pub fn chdir(p: &path::Path) -> io::Result<()> { - let p: &OsStr = p.as_ref(); - let p = CString::new(p.as_bytes())?; - if unsafe { libc::chdir(p.as_ptr()) } != 0 { - return Err(io::Error::last_os_error()); - } - Ok(()) + let result = run_path_with_cstr(p, |p| unsafe { Ok(libc::chdir(p.as_ptr())) })?; + if result == 0 { Ok(()) } else { Err(io::Error::last_os_error()) } } pub struct SplitPaths<'a> { @@ -501,10 +501,10 @@ pub unsafe fn environ() -> *mut *const *const c_char { ptr::addr_of_mut!(environ) } -static ENV_LOCK: StaticRwLock = StaticRwLock::new(); +static ENV_LOCK: RwLock<()> = RwLock::new(()); -pub fn env_read_lock() -> StaticRwLockReadGuard { - ENV_LOCK.read() +pub fn env_read_lock() -> impl Drop { + ENV_LOCK.read().unwrap_or_else(PoisonError::into_inner) } /// Returns a vector of (variable, value) byte-vector pairs for all the @@ -546,35 +546,32 @@ pub fn env() -> Env { pub fn getenv(k: &OsStr) -> Option<OsString> { // environment variables with a nul byte can't be set, so their value is // always None as well - let k = CString::new(k.as_bytes()).ok()?; - unsafe { + let s = run_with_cstr(k.as_bytes(), |k| { let _guard = env_read_lock(); - let s = libc::getenv(k.as_ptr()) as *const libc::c_char; - if s.is_null() { - None - } else { - Some(OsStringExt::from_vec(CStr::from_ptr(s).to_bytes().to_vec())) - } + Ok(unsafe { libc::getenv(k.as_ptr()) } as *const libc::c_char) + }) + .ok()?; + if s.is_null() { + None + } else { + Some(OsStringExt::from_vec(unsafe { CStr::from_ptr(s) }.to_bytes().to_vec())) } } pub fn setenv(k: &OsStr, v: &OsStr) -> io::Result<()> { - let k = CString::new(k.as_bytes())?; - let v = CString::new(v.as_bytes())?; - - unsafe { - let _guard = ENV_LOCK.write(); - cvt(libc::setenv(k.as_ptr(), v.as_ptr(), 1)).map(drop) - } + run_with_cstr(k.as_bytes(), |k| { + run_with_cstr(v.as_bytes(), |v| { + let _guard = ENV_LOCK.write(); + cvt(unsafe { libc::setenv(k.as_ptr(), v.as_ptr(), 1) }).map(drop) + }) + }) } pub fn unsetenv(n: &OsStr) -> io::Result<()> { - let nbuf = CString::new(n.as_bytes())?; - - unsafe { + run_with_cstr(n.as_bytes(), |nbuf| { let _guard = ENV_LOCK.write(); - cvt(libc::unsetenv(nbuf.as_ptr())).map(drop) - } + cvt(unsafe { libc::unsetenv(nbuf.as_ptr()) }).map(drop) + }) } #[cfg(not(target_os = "espidf"))] diff --git a/library/std/src/sys/unix/os_str.rs b/library/std/src/sys/unix/os_str.rs index ccbc18224..017e2af29 100644 --- a/library/std/src/sys/unix/os_str.rs +++ b/library/std/src/sys/unix/os_str.rs @@ -11,7 +11,7 @@ use crate::str; use crate::sync::Arc; use crate::sys_common::{AsInner, IntoInner}; -use core::str::lossy::{Utf8Lossy, Utf8LossyChunk}; +use core::str::Utf8Chunks; #[cfg(test)] #[path = "../unix/os_str/tests.rs"] @@ -29,26 +29,32 @@ pub struct Slice { } impl fmt::Debug for Slice { - fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { - // Writes out a valid unicode string with the correct escape sequences - - formatter.write_str("\"")?; - for Utf8LossyChunk { valid, broken } in Utf8Lossy::from_bytes(&self.inner).chunks() { - for c in valid.chars().flat_map(|c| c.escape_debug()) { - formatter.write_char(c)? - } - - for b in broken { - write!(formatter, "\\x{:02X}", b)?; - } - } - formatter.write_str("\"") + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + fmt::Debug::fmt(&Utf8Chunks::new(&self.inner).debug(), f) } } impl fmt::Display for Slice { - fn fmt(&self, formatter: &mut fmt::Formatter<'_>) -> fmt::Result { - fmt::Display::fmt(&Utf8Lossy::from_bytes(&self.inner), formatter) + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + // If we're the empty string then our iterator won't actually yield + // anything, so perform the formatting manually + if self.inner.is_empty() { + return "".fmt(f); + } + + for chunk in Utf8Chunks::new(&self.inner) { + let valid = chunk.valid(); + // If we successfully decoded the whole chunk as a valid string then + // we can return a direct formatting of the string which will also + // respect various formatting flags if possible. + if chunk.invalid().is_empty() { + return valid.fmt(f); + } + + f.write_str(valid)?; + f.write_char(char::REPLACEMENT_CHARACTER)?; + } + Ok(()) } } diff --git a/library/std/src/sys/unix/os_str/tests.rs b/library/std/src/sys/unix/os_str/tests.rs index 213277f01..22ba0c923 100644 --- a/library/std/src/sys/unix/os_str/tests.rs +++ b/library/std/src/sys/unix/os_str/tests.rs @@ -8,3 +8,11 @@ fn slice_debug_output() { assert_eq!(output, expected); } + +#[test] +fn display() { + assert_eq!( + "Hello\u{FFFD}\u{FFFD} There\u{FFFD} Goodbye", + Slice::from_u8_slice(b"Hello\xC0\x80 There\xE6\x83 Goodbye").to_string(), + ); +} diff --git a/library/std/src/sys/unix/process/process_common.rs b/library/std/src/sys/unix/process/process_common.rs index bca1b65a7..848adca78 100644 --- a/library/std/src/sys/unix/process/process_common.rs +++ b/library/std/src/sys/unix/process/process_common.rs @@ -39,17 +39,39 @@ cfg_if::cfg_if! { // https://github.com/aosp-mirror/platform_bionic/blob/ad8dcd6023294b646e5a8288c0ed431b0845da49/libc/include/android/legacy_signal_inlines.h cfg_if::cfg_if! { if #[cfg(target_os = "android")] { + #[allow(dead_code)] pub unsafe fn sigemptyset(set: *mut libc::sigset_t) -> libc::c_int { set.write_bytes(0u8, 1); return 0; } + #[allow(dead_code)] pub unsafe fn sigaddset(set: *mut libc::sigset_t, signum: libc::c_int) -> libc::c_int { - use crate::{slice, mem}; + use crate::{ + mem::{align_of, size_of}, + slice, + }; + use libc::{c_ulong, sigset_t}; + + // The implementations from bionic (android libc) type pun `sigset_t` as an + // array of `c_ulong`. This works, but lets add a smoke check to make sure + // that doesn't change. + const _: () = assert!( + align_of::<c_ulong>() == align_of::<sigset_t>() + && (size_of::<sigset_t>() % size_of::<c_ulong>()) == 0 + ); - let raw = slice::from_raw_parts_mut(set as *mut u8, mem::size_of::<libc::sigset_t>()); let bit = (signum - 1) as usize; - raw[bit / 8] |= 1 << (bit % 8); + if set.is_null() || bit >= (8 * size_of::<sigset_t>()) { + crate::sys::unix::os::set_errno(libc::EINVAL); + return -1; + } + let raw = slice::from_raw_parts_mut( + set as *mut c_ulong, + size_of::<sigset_t>() / size_of::<c_ulong>(), + ); + const LONG_BIT: usize = size_of::<c_ulong>() * 8; + raw[bit / LONG_BIT] |= 1 << (bit % LONG_BIT); return 0; } } else { @@ -72,6 +94,7 @@ pub struct Command { argv: Argv, env: CommandEnv, + program_kind: ProgramKind, cwd: Option<CString>, uid: Option<uid_t>, gid: Option<gid_t>, @@ -128,15 +151,40 @@ pub enum Stdio { Fd(FileDesc), } +#[derive(Copy, Clone, Debug, Eq, PartialEq)] +pub enum ProgramKind { + /// A program that would be looked up on the PATH (e.g. `ls`) + PathLookup, + /// A relative path (e.g. `my-dir/foo`, `../foo`, `./foo`) + Relative, + /// An absolute path. + Absolute, +} + +impl ProgramKind { + fn new(program: &OsStr) -> Self { + if program.bytes().starts_with(b"/") { + Self::Absolute + } else if program.bytes().contains(&b'/') { + // If the program has more than one component in it, it is a relative path. + Self::Relative + } else { + Self::PathLookup + } + } +} + impl Command { #[cfg(not(target_os = "linux"))] pub fn new(program: &OsStr) -> Command { let mut saw_nul = false; + let program_kind = ProgramKind::new(program.as_ref()); let program = os2c(program, &mut saw_nul); Command { argv: Argv(vec![program.as_ptr(), ptr::null()]), args: vec![program.clone()], program, + program_kind, env: Default::default(), cwd: None, uid: None, @@ -154,11 +202,13 @@ impl Command { #[cfg(target_os = "linux")] pub fn new(program: &OsStr) -> Command { let mut saw_nul = false; + let program_kind = ProgramKind::new(program.as_ref()); let program = os2c(program, &mut saw_nul); Command { argv: Argv(vec![program.as_ptr(), ptr::null()]), args: vec![program.clone()], program, + program_kind, env: Default::default(), cwd: None, uid: None, @@ -234,6 +284,11 @@ impl Command { OsStr::from_bytes(self.program.as_bytes()) } + #[allow(dead_code)] + pub fn get_program_kind(&self) -> ProgramKind { + self.program_kind + } + pub fn get_args(&self) -> CommandArgs<'_> { let mut iter = self.args.iter(); iter.next(); diff --git a/library/std/src/sys/unix/process/process_common/tests.rs b/library/std/src/sys/unix/process/process_common/tests.rs index 1956b3692..03631e4e3 100644 --- a/library/std/src/sys/unix/process/process_common/tests.rs +++ b/library/std/src/sys/unix/process/process_common/tests.rs @@ -31,41 +31,54 @@ macro_rules! t { ignore )] fn test_process_mask() { - unsafe { - // Test to make sure that a signal mask does not get inherited. - let mut cmd = Command::new(OsStr::new("cat")); - - let mut set = mem::MaybeUninit::<libc::sigset_t>::uninit(); - let mut old_set = mem::MaybeUninit::<libc::sigset_t>::uninit(); - t!(cvt(sigemptyset(set.as_mut_ptr()))); - t!(cvt(sigaddset(set.as_mut_ptr(), libc::SIGINT))); - t!(cvt_nz(libc::pthread_sigmask(libc::SIG_SETMASK, set.as_ptr(), old_set.as_mut_ptr()))); - - cmd.stdin(Stdio::MakePipe); - cmd.stdout(Stdio::MakePipe); - - let (mut cat, mut pipes) = t!(cmd.spawn(Stdio::Null, true)); - let stdin_write = pipes.stdin.take().unwrap(); - let stdout_read = pipes.stdout.take().unwrap(); - - t!(cvt_nz(libc::pthread_sigmask(libc::SIG_SETMASK, old_set.as_ptr(), ptr::null_mut()))); - - t!(cvt(libc::kill(cat.id() as libc::pid_t, libc::SIGINT))); - // We need to wait until SIGINT is definitely delivered. The - // easiest way is to write something to cat, and try to read it - // back: if SIGINT is unmasked, it'll get delivered when cat is - // next scheduled. - let _ = stdin_write.write(b"Hello"); - drop(stdin_write); - - // Either EOF or failure (EPIPE) is okay. - let mut buf = [0; 5]; - if let Ok(ret) = stdout_read.read(&mut buf) { - assert_eq!(ret, 0); + // Test to make sure that a signal mask *does* get inherited. + fn test_inner(mut cmd: Command) { + unsafe { + let mut set = mem::MaybeUninit::<libc::sigset_t>::uninit(); + let mut old_set = mem::MaybeUninit::<libc::sigset_t>::uninit(); + t!(cvt(sigemptyset(set.as_mut_ptr()))); + t!(cvt(sigaddset(set.as_mut_ptr(), libc::SIGINT))); + t!(cvt_nz(libc::pthread_sigmask( + libc::SIG_SETMASK, + set.as_ptr(), + old_set.as_mut_ptr() + ))); + + cmd.stdin(Stdio::MakePipe); + cmd.stdout(Stdio::MakePipe); + + let (mut cat, mut pipes) = t!(cmd.spawn(Stdio::Null, true)); + let stdin_write = pipes.stdin.take().unwrap(); + let stdout_read = pipes.stdout.take().unwrap(); + + t!(cvt_nz(libc::pthread_sigmask(libc::SIG_SETMASK, old_set.as_ptr(), ptr::null_mut()))); + + t!(cvt(libc::kill(cat.id() as libc::pid_t, libc::SIGINT))); + // We need to wait until SIGINT is definitely delivered. The + // easiest way is to write something to cat, and try to read it + // back: if SIGINT is unmasked, it'll get delivered when cat is + // next scheduled. + let _ = stdin_write.write(b"Hello"); + drop(stdin_write); + + // Exactly 5 bytes should be read. + let mut buf = [0; 5]; + let ret = t!(stdout_read.read(&mut buf)); + assert_eq!(ret, 5); + assert_eq!(&buf, b"Hello"); + + t!(cat.wait()); } - - t!(cat.wait()); } + + // A plain `Command::new` uses the posix_spawn path on many platforms. + let cmd = Command::new(OsStr::new("cat")); + test_inner(cmd); + + // Specifying `pre_exec` forces the fork/exec path. + let mut cmd = Command::new(OsStr::new("cat")); + unsafe { cmd.pre_exec(Box::new(|| Ok(()))) }; + test_inner(cmd); } #[test] @@ -122,3 +135,27 @@ fn test_process_group_no_posix_spawn() { t!(cat.wait()); } } + +#[test] +fn test_program_kind() { + let vectors = &[ + ("foo", ProgramKind::PathLookup), + ("foo.out", ProgramKind::PathLookup), + ("./foo", ProgramKind::Relative), + ("../foo", ProgramKind::Relative), + ("dir/foo", ProgramKind::Relative), + // Note that paths on Unix can't contain / in them, so this is actually the directory "fo\\" + // followed by the file "o". + ("fo\\/o", ProgramKind::Relative), + ("/foo", ProgramKind::Absolute), + ("/dir/../foo", ProgramKind::Absolute), + ]; + + for (program, expected_kind) in vectors { + assert_eq!( + ProgramKind::new(program.as_ref()), + *expected_kind, + "actual != expected program kind for input {program}", + ); + } +} diff --git a/library/std/src/sys/unix/process/process_fuchsia.rs b/library/std/src/sys/unix/process/process_fuchsia.rs index 73f5d3a61..66ea3db20 100644 --- a/library/std/src/sys/unix/process/process_fuchsia.rs +++ b/library/std/src/sys/unix/process/process_fuchsia.rs @@ -287,7 +287,7 @@ impl ExitStatus { // SuS and POSIX) say a wait status is, but Fuchsia apparently uses a u64, so it won't // necessarily fit. // - // It seems to me that that the right answer would be to provide std::os::fuchsia with its + // It seems to me that the right answer would be to provide std::os::fuchsia with its // own ExitStatusExt, rather that trying to provide a not very convincing imitation of // Unix. Ie, std::os::unix::process:ExitStatusExt ought not to exist on Fuchsia. But // fixing this up that is beyond the scope of my efforts now. diff --git a/library/std/src/sys/unix/process/process_unix.rs b/library/std/src/sys/unix/process/process_unix.rs index 75bb92437..56a805cef 100644 --- a/library/std/src/sys/unix/process/process_unix.rs +++ b/library/std/src/sys/unix/process/process_unix.rs @@ -2,7 +2,6 @@ use crate::fmt; use crate::io::{self, Error, ErrorKind}; use crate::mem; use crate::num::NonZeroI32; -use crate::ptr; use crate::sys; use crate::sys::cvt; use crate::sys::process::process_common::*; @@ -310,7 +309,7 @@ impl Command { //FIXME: Redox kernel does not support setgroups yet #[cfg(not(target_os = "redox"))] if libc::getuid() == 0 && self.get_groups().is_none() { - cvt(libc::setgroups(0, ptr::null()))?; + cvt(libc::setgroups(0, crate::ptr::null()))?; } cvt(libc::setuid(u as uid_t))?; } @@ -326,30 +325,26 @@ impl Command { // emscripten has no signal support. #[cfg(not(target_os = "emscripten"))] { - use crate::mem::MaybeUninit; - use crate::sys::cvt_nz; - // Reset signal handling so the child process starts in a - // standardized state. libstd ignores SIGPIPE, and signal-handling - // libraries often set a mask. Child processes inherit ignored - // signals and the signal mask from their parent, but most - // UNIX programs do not reset these things on their own, so we - // need to clean things up now to avoid confusing the program - // we're about to run. - let mut set = MaybeUninit::<libc::sigset_t>::uninit(); - cvt(sigemptyset(set.as_mut_ptr()))?; - cvt_nz(libc::pthread_sigmask(libc::SIG_SETMASK, set.as_ptr(), ptr::null_mut()))?; - - #[cfg(target_os = "android")] // see issue #88585 - { - let mut action: libc::sigaction = mem::zeroed(); - action.sa_sigaction = libc::SIG_DFL; - cvt(libc::sigaction(libc::SIGPIPE, &action, ptr::null_mut()))?; - } - #[cfg(not(target_os = "android"))] - { - let ret = sys::signal(libc::SIGPIPE, libc::SIG_DFL); - if ret == libc::SIG_ERR { - return Err(io::Error::last_os_error()); + // Inherit the signal mask from the parent rather than resetting it (i.e. do not call + // pthread_sigmask). + + // If #[unix_sigpipe] is specified, don't reset SIGPIPE to SIG_DFL. + // If #[unix_sigpipe] is not specified, reset SIGPIPE to SIG_DFL for backward compatibility. + // + // #[unix_sigpipe] is an opportunity to change the default here. + if !crate::sys::unix_sigpipe_attr_specified() { + #[cfg(target_os = "android")] // see issue #88585 + { + let mut action: libc::sigaction = mem::zeroed(); + action.sa_sigaction = libc::SIG_DFL; + cvt(libc::sigaction(libc::SIGPIPE, &action, crate::ptr::null_mut()))?; + } + #[cfg(not(target_os = "android"))] + { + let ret = sys::signal(libc::SIGPIPE, libc::SIG_DFL); + if ret == libc::SIG_ERR { + return Err(io::Error::last_os_error()); + } } } } @@ -411,7 +406,7 @@ impl Command { envp: Option<&CStringArray>, ) -> io::Result<Option<Process>> { use crate::mem::MaybeUninit; - use crate::sys::{self, cvt_nz}; + use crate::sys::{self, cvt_nz, unix_sigpipe_attr_specified}; if self.get_gid().is_some() || self.get_uid().is_some() @@ -453,7 +448,9 @@ impl Command { // successfully launch the program, but erroneously return // ENOENT when used with posix_spawn_file_actions_addchdir_np // which was introduced in macOS 10.15. - return Ok(None); + if self.get_program_kind() == ProgramKind::Relative { + return Ok(None); + } } match posix_spawn_file_actions_addchdir_np.get() { Some(f) => Some((f, cwd)), @@ -529,13 +526,24 @@ impl Command { cvt_nz(libc::posix_spawnattr_setpgroup(attrs.0.as_mut_ptr(), pgroup))?; } - let mut set = MaybeUninit::<libc::sigset_t>::uninit(); - cvt(sigemptyset(set.as_mut_ptr()))?; - cvt_nz(libc::posix_spawnattr_setsigmask(attrs.0.as_mut_ptr(), set.as_ptr()))?; - cvt(sigaddset(set.as_mut_ptr(), libc::SIGPIPE))?; - cvt_nz(libc::posix_spawnattr_setsigdefault(attrs.0.as_mut_ptr(), set.as_ptr()))?; + // Inherit the signal mask from this process rather than resetting it (i.e. do not call + // posix_spawnattr_setsigmask). + + // If #[unix_sigpipe] is specified, don't reset SIGPIPE to SIG_DFL. + // If #[unix_sigpipe] is not specified, reset SIGPIPE to SIG_DFL for backward compatibility. + // + // #[unix_sigpipe] is an opportunity to change the default here. + if !unix_sigpipe_attr_specified() { + let mut default_set = MaybeUninit::<libc::sigset_t>::uninit(); + cvt(sigemptyset(default_set.as_mut_ptr()))?; + cvt(sigaddset(default_set.as_mut_ptr(), libc::SIGPIPE))?; + cvt_nz(libc::posix_spawnattr_setsigdefault( + attrs.0.as_mut_ptr(), + default_set.as_ptr(), + ))?; + flags |= libc::POSIX_SPAWN_SETSIGDEF; + } - flags |= libc::POSIX_SPAWN_SETSIGDEF | libc::POSIX_SPAWN_SETSIGMASK; cvt_nz(libc::posix_spawnattr_setflags(attrs.0.as_mut_ptr(), flags as _))?; // Make sure we synchronize access to the global `environ` resource @@ -820,14 +828,14 @@ impl crate::os::linux::process::ChildExt for crate::process::Child { self.handle .pidfd .as_ref() - .ok_or_else(|| Error::new(ErrorKind::Other, "No pidfd was created.")) + .ok_or_else(|| Error::new(ErrorKind::Uncategorized, "No pidfd was created.")) } fn take_pidfd(&mut self) -> io::Result<PidFd> { self.handle .pidfd .take() - .ok_or_else(|| Error::new(ErrorKind::Other, "No pidfd was created.")) + .ok_or_else(|| Error::new(ErrorKind::Uncategorized, "No pidfd was created.")) } } diff --git a/library/std/src/sys/unix/rand.rs b/library/std/src/sys/unix/rand.rs index bf4920488..a6fe07873 100644 --- a/library/std/src/sys/unix/rand.rs +++ b/library/std/src/sys/unix/rand.rs @@ -1,13 +1,13 @@ -use crate::mem; -use crate::slice; - pub fn hashmap_random_keys() -> (u64, u64) { - let mut v = (0, 0); - unsafe { - let view = slice::from_raw_parts_mut(&mut v as *mut _ as *mut u8, mem::size_of_val(&v)); - imp::fill_bytes(view); - } - v + const KEY_LEN: usize = core::mem::size_of::<u64>(); + + let mut v = [0u8; KEY_LEN * 2]; + imp::fill_bytes(&mut v); + + let key1 = v[0..KEY_LEN].try_into().unwrap(); + let key2 = v[KEY_LEN..].try_into().unwrap(); + + (u64::from_ne_bytes(key1), u64::from_ne_bytes(key2)) } #[cfg(all( diff --git a/library/std/src/sys/unix/stdio.rs b/library/std/src/sys/unix/stdio.rs index 329f9433d..b3626c564 100644 --- a/library/std/src/sys/unix/stdio.rs +++ b/library/std/src/sys/unix/stdio.rs @@ -1,6 +1,6 @@ use crate::io::{self, IoSlice, IoSliceMut}; use crate::mem::ManuallyDrop; -use crate::os::unix::io::{AsFd, BorrowedFd, FromRawFd}; +use crate::os::unix::io::FromRawFd; use crate::sys::fd::FileDesc; pub struct Stdin(()); @@ -91,51 +91,3 @@ pub const STDIN_BUF_SIZE: usize = crate::sys_common::io::DEFAULT_BUF_SIZE; pub fn panic_output() -> Option<impl io::Write> { Some(Stderr::new()) } - -#[stable(feature = "io_safety", since = "1.63.0")] -impl AsFd for io::Stdin { - #[inline] - fn as_fd(&self) -> BorrowedFd<'_> { - unsafe { BorrowedFd::borrow_raw(libc::STDIN_FILENO) } - } -} - -#[stable(feature = "io_safety", since = "1.63.0")] -impl<'a> AsFd for io::StdinLock<'a> { - #[inline] - fn as_fd(&self) -> BorrowedFd<'_> { - unsafe { BorrowedFd::borrow_raw(libc::STDIN_FILENO) } - } -} - -#[stable(feature = "io_safety", since = "1.63.0")] -impl AsFd for io::Stdout { - #[inline] - fn as_fd(&self) -> BorrowedFd<'_> { - unsafe { BorrowedFd::borrow_raw(libc::STDOUT_FILENO) } - } -} - -#[stable(feature = "io_safety", since = "1.63.0")] -impl<'a> AsFd for io::StdoutLock<'a> { - #[inline] - fn as_fd(&self) -> BorrowedFd<'_> { - unsafe { BorrowedFd::borrow_raw(libc::STDOUT_FILENO) } - } -} - -#[stable(feature = "io_safety", since = "1.63.0")] -impl AsFd for io::Stderr { - #[inline] - fn as_fd(&self) -> BorrowedFd<'_> { - unsafe { BorrowedFd::borrow_raw(libc::STDERR_FILENO) } - } -} - -#[stable(feature = "io_safety", since = "1.63.0")] -impl<'a> AsFd for io::StderrLock<'a> { - #[inline] - fn as_fd(&self) -> BorrowedFd<'_> { - unsafe { BorrowedFd::borrow_raw(libc::STDERR_FILENO) } - } -} diff --git a/library/std/src/sys/unix/thread.rs b/library/std/src/sys/unix/thread.rs index 36a3fa602..c1d30dd9d 100644 --- a/library/std/src/sys/unix/thread.rs +++ b/library/std/src/sys/unix/thread.rs @@ -116,11 +116,9 @@ impl Thread { debug_assert_eq!(ret, 0); } - #[cfg(any(target_os = "linux", target_os = "android"))] + #[cfg(target_os = "android")] pub fn set_name(name: &CStr) { const PR_SET_NAME: libc::c_int = 15; - // pthread wrapper only appeared in glibc 2.12, so we use syscall - // directly. unsafe { libc::prctl( PR_SET_NAME, @@ -132,6 +130,19 @@ impl Thread { } } + #[cfg(target_os = "linux")] + pub fn set_name(name: &CStr) { + const TASK_COMM_LEN: usize = 16; + + unsafe { + // Available since glibc 2.12, musl 1.1.16, and uClibc 1.0.20. + let name = truncate_cstr(name, TASK_COMM_LEN); + let res = libc::pthread_setname_np(libc::pthread_self(), name.as_ptr()); + // We have no good way of propagating errors here, but in debug-builds let's check that this actually worked. + debug_assert_eq!(res, 0); + } + } + #[cfg(any(target_os = "freebsd", target_os = "dragonfly", target_os = "openbsd"))] pub fn set_name(name: &CStr) { unsafe { @@ -142,20 +153,23 @@ impl Thread { #[cfg(any(target_os = "macos", target_os = "ios", target_os = "watchos"))] pub fn set_name(name: &CStr) { unsafe { - libc::pthread_setname_np(name.as_ptr()); + let name = truncate_cstr(name, libc::MAXTHREADNAMESIZE); + let res = libc::pthread_setname_np(name.as_ptr()); + // We have no good way of propagating errors here, but in debug-builds let's check that this actually worked. + debug_assert_eq!(res, 0); } } #[cfg(target_os = "netbsd")] pub fn set_name(name: &CStr) { - use crate::ffi::CString; - let cname = CString::new(&b"%s"[..]).unwrap(); unsafe { - libc::pthread_setname_np( + let cname = CStr::from_bytes_with_nul_unchecked(b"%s\0".as_slice()); + let res = libc::pthread_setname_np( libc::pthread_self(), cname.as_ptr(), name.as_ptr() as *mut libc::c_void, ); + debug_assert_eq!(res, 0); } } @@ -168,9 +182,8 @@ impl Thread { } if let Some(f) = pthread_setname_np.get() { - unsafe { - f(libc::pthread_self(), name.as_ptr()); - } + let res = unsafe { f(libc::pthread_self(), name.as_ptr()) }; + debug_assert_eq!(res, 0); } } @@ -271,6 +284,20 @@ impl Drop for Thread { } } +#[cfg(any(target_os = "linux", target_os = "macos", target_os = "ios", target_os = "watchos"))] +fn truncate_cstr(cstr: &CStr, max_with_nul: usize) -> crate::borrow::Cow<'_, CStr> { + use crate::{borrow::Cow, ffi::CString}; + + if cstr.to_bytes_with_nul().len() > max_with_nul { + let bytes = cstr.to_bytes()[..max_with_nul - 1].to_vec(); + // SAFETY: the non-nul bytes came straight from a CStr. + // (CString will add the terminating nul.) + Cow::Owned(unsafe { CString::from_vec_unchecked(bytes) }) + } else { + Cow::Borrowed(cstr) + } +} + pub fn available_parallelism() -> io::Result<NonZeroUsize> { cfg_if::cfg_if! { if #[cfg(any( @@ -423,7 +450,7 @@ mod cgroups { Some(b"") => Cgroup::V2, Some(controllers) if from_utf8(controllers) - .is_ok_and(|c| c.split(",").any(|c| c == "cpu")) => + .is_ok_and(|c| c.split(',').any(|c| c == "cpu")) => { Cgroup::V1 } @@ -761,6 +788,16 @@ pub mod guard { const GUARD_PAGES: usize = 1; let guard = guardaddr..guardaddr + GUARD_PAGES * page_size; Some(guard) + } else if cfg!(target_os = "openbsd") { + // OpenBSD stack already includes a guard page, and stack is + // immutable. + // + // We'll just note where we expect rlimit to start + // faulting, so our handler can report "stack overflow", and + // trust that the kernel's own stack guard will work. + let stackptr = get_stack_start_aligned()?; + let stackaddr = stackptr.addr(); + Some(stackaddr - page_size..stackaddr) } else { // Reallocate the last page of the stack. // This ensures SIGBUS will be raised on diff --git a/library/std/src/sys/unix/thread_local_dtor.rs b/library/std/src/sys/unix/thread_local_dtor.rs index 6e8be2a91..d7fd2130f 100644 --- a/library/std/src/sys/unix/thread_local_dtor.rs +++ b/library/std/src/sys/unix/thread_local_dtor.rs @@ -17,6 +17,7 @@ target_os = "redox", target_os = "emscripten" ))] +#[cfg_attr(target_family = "wasm", allow(unused))] // might remain unused depending on target details (e.g. wasm32-unknown-emscripten) pub unsafe fn register_dtor(t: *mut u8, dtor: unsafe extern "C" fn(*mut u8)) { use crate::mem; use crate::sys_common::thread_local_dtor::register_dtor_fallback; diff --git a/library/std/src/sys/unix/thread_local_key.rs b/library/std/src/sys/unix/thread_local_key.rs index 2c5b94b1e..2b2d079ee 100644 --- a/library/std/src/sys/unix/thread_local_key.rs +++ b/library/std/src/sys/unix/thread_local_key.rs @@ -27,8 +27,3 @@ pub unsafe fn destroy(key: Key) { let r = libc::pthread_key_delete(key); debug_assert_eq!(r, 0); } - -#[inline] -pub fn requires_synchronized_create() -> bool { - false -} diff --git a/library/std/src/sys/unix/thread_parker.rs b/library/std/src/sys/unix/thread_parker.rs deleted file mode 100644 index ca1a7138f..000000000 --- a/library/std/src/sys/unix/thread_parker.rs +++ /dev/null @@ -1,281 +0,0 @@ -//! Thread parking without `futex` using the `pthread` synchronization primitives. - -#![cfg(not(any( - target_os = "linux", - target_os = "android", - all(target_os = "emscripten", target_feature = "atomics"), - target_os = "freebsd", - target_os = "openbsd", - target_os = "dragonfly", - target_os = "fuchsia", -)))] - -use crate::cell::UnsafeCell; -use crate::marker::PhantomPinned; -use crate::pin::Pin; -use crate::ptr::addr_of_mut; -use crate::sync::atomic::AtomicUsize; -use crate::sync::atomic::Ordering::SeqCst; -use crate::time::Duration; - -const EMPTY: usize = 0; -const PARKED: usize = 1; -const NOTIFIED: usize = 2; - -unsafe fn lock(lock: *mut libc::pthread_mutex_t) { - let r = libc::pthread_mutex_lock(lock); - debug_assert_eq!(r, 0); -} - -unsafe fn unlock(lock: *mut libc::pthread_mutex_t) { - let r = libc::pthread_mutex_unlock(lock); - debug_assert_eq!(r, 0); -} - -unsafe fn notify_one(cond: *mut libc::pthread_cond_t) { - let r = libc::pthread_cond_signal(cond); - debug_assert_eq!(r, 0); -} - -unsafe fn wait(cond: *mut libc::pthread_cond_t, lock: *mut libc::pthread_mutex_t) { - let r = libc::pthread_cond_wait(cond, lock); - debug_assert_eq!(r, 0); -} - -const TIMESPEC_MAX: libc::timespec = - libc::timespec { tv_sec: <libc::time_t>::MAX, tv_nsec: 1_000_000_000 - 1 }; - -unsafe fn wait_timeout( - cond: *mut libc::pthread_cond_t, - lock: *mut libc::pthread_mutex_t, - dur: Duration, -) { - // Use the system clock on systems that do not support pthread_condattr_setclock. - // This unfortunately results in problems when the system time changes. - #[cfg(any( - target_os = "macos", - target_os = "ios", - target_os = "watchos", - target_os = "espidf" - ))] - let (now, dur) = { - use super::time::SystemTime; - use crate::cmp::min; - - // OSX implementation of `pthread_cond_timedwait` is buggy - // with super long durations. When duration is greater than - // 0x100_0000_0000_0000 seconds, `pthread_cond_timedwait` - // in macOS Sierra return error 316. - // - // This program demonstrates the issue: - // https://gist.github.com/stepancheg/198db4623a20aad2ad7cddb8fda4a63c - // - // To work around this issue, and possible bugs of other OSes, timeout - // is clamped to 1000 years, which is allowable per the API of `park_timeout` - // because of spurious wakeups. - let dur = min(dur, Duration::from_secs(1000 * 365 * 86400)); - let now = SystemTime::now().t; - (now, dur) - }; - // Use the monotonic clock on other systems. - #[cfg(not(any( - target_os = "macos", - target_os = "ios", - target_os = "watchos", - target_os = "espidf" - )))] - let (now, dur) = { - use super::time::Timespec; - - (Timespec::now(libc::CLOCK_MONOTONIC), dur) - }; - - let timeout = - now.checked_add_duration(&dur).and_then(|t| t.to_timespec()).unwrap_or(TIMESPEC_MAX); - let r = libc::pthread_cond_timedwait(cond, lock, &timeout); - debug_assert!(r == libc::ETIMEDOUT || r == 0); -} - -pub struct Parker { - state: AtomicUsize, - lock: UnsafeCell<libc::pthread_mutex_t>, - cvar: UnsafeCell<libc::pthread_cond_t>, - // The `pthread` primitives require a stable address, so make this struct `!Unpin`. - _pinned: PhantomPinned, -} - -impl Parker { - /// Construct the UNIX parker in-place. - /// - /// # Safety - /// The constructed parker must never be moved. - pub unsafe fn new(parker: *mut Parker) { - // Use the default mutex implementation to allow for simpler initialization. - // This could lead to undefined behaviour when deadlocking. This is avoided - // by not deadlocking. Note in particular the unlocking operation before any - // panic, as code after the panic could try to park again. - addr_of_mut!((*parker).state).write(AtomicUsize::new(EMPTY)); - addr_of_mut!((*parker).lock).write(UnsafeCell::new(libc::PTHREAD_MUTEX_INITIALIZER)); - - cfg_if::cfg_if! { - if #[cfg(any( - target_os = "macos", - target_os = "ios", - target_os = "watchos", - target_os = "l4re", - target_os = "android", - target_os = "redox" - ))] { - addr_of_mut!((*parker).cvar).write(UnsafeCell::new(libc::PTHREAD_COND_INITIALIZER)); - } else if #[cfg(any(target_os = "espidf", target_os = "horizon"))] { - let r = libc::pthread_cond_init(addr_of_mut!((*parker).cvar).cast(), crate::ptr::null()); - assert_eq!(r, 0); - } else { - use crate::mem::MaybeUninit; - let mut attr = MaybeUninit::<libc::pthread_condattr_t>::uninit(); - let r = libc::pthread_condattr_init(attr.as_mut_ptr()); - assert_eq!(r, 0); - let r = libc::pthread_condattr_setclock(attr.as_mut_ptr(), libc::CLOCK_MONOTONIC); - assert_eq!(r, 0); - let r = libc::pthread_cond_init(addr_of_mut!((*parker).cvar).cast(), attr.as_ptr()); - assert_eq!(r, 0); - let r = libc::pthread_condattr_destroy(attr.as_mut_ptr()); - assert_eq!(r, 0); - } - } - } - - // This implementation doesn't require `unsafe`, but other implementations - // may assume this is only called by the thread that owns the Parker. - pub unsafe fn park(self: Pin<&Self>) { - // If we were previously notified then we consume this notification and - // return quickly. - if self.state.compare_exchange(NOTIFIED, EMPTY, SeqCst, SeqCst).is_ok() { - return; - } - - // Otherwise we need to coordinate going to sleep - lock(self.lock.get()); - match self.state.compare_exchange(EMPTY, PARKED, SeqCst, SeqCst) { - Ok(_) => {} - Err(NOTIFIED) => { - // We must read here, even though we know it will be `NOTIFIED`. - // This is because `unpark` may have been called again since we read - // `NOTIFIED` in the `compare_exchange` above. We must perform an - // acquire operation that synchronizes with that `unpark` to observe - // any writes it made before the call to unpark. To do that we must - // read from the write it made to `state`. - let old = self.state.swap(EMPTY, SeqCst); - - unlock(self.lock.get()); - - assert_eq!(old, NOTIFIED, "park state changed unexpectedly"); - return; - } // should consume this notification, so prohibit spurious wakeups in next park. - Err(_) => { - unlock(self.lock.get()); - - panic!("inconsistent park state") - } - } - - loop { - wait(self.cvar.get(), self.lock.get()); - - match self.state.compare_exchange(NOTIFIED, EMPTY, SeqCst, SeqCst) { - Ok(_) => break, // got a notification - Err(_) => {} // spurious wakeup, go back to sleep - } - } - - unlock(self.lock.get()); - } - - // This implementation doesn't require `unsafe`, but other implementations - // may assume this is only called by the thread that owns the Parker. Use - // `Pin` to guarantee a stable address for the mutex and condition variable. - pub unsafe fn park_timeout(self: Pin<&Self>, dur: Duration) { - // Like `park` above we have a fast path for an already-notified thread, and - // afterwards we start coordinating for a sleep. - // return quickly. - if self.state.compare_exchange(NOTIFIED, EMPTY, SeqCst, SeqCst).is_ok() { - return; - } - - lock(self.lock.get()); - match self.state.compare_exchange(EMPTY, PARKED, SeqCst, SeqCst) { - Ok(_) => {} - Err(NOTIFIED) => { - // We must read again here, see `park`. - let old = self.state.swap(EMPTY, SeqCst); - unlock(self.lock.get()); - - assert_eq!(old, NOTIFIED, "park state changed unexpectedly"); - return; - } // should consume this notification, so prohibit spurious wakeups in next park. - Err(_) => { - unlock(self.lock.get()); - panic!("inconsistent park_timeout state") - } - } - - // Wait with a timeout, and if we spuriously wake up or otherwise wake up - // from a notification we just want to unconditionally set the state back to - // empty, either consuming a notification or un-flagging ourselves as - // parked. - wait_timeout(self.cvar.get(), self.lock.get(), dur); - - match self.state.swap(EMPTY, SeqCst) { - NOTIFIED => unlock(self.lock.get()), // got a notification, hurray! - PARKED => unlock(self.lock.get()), // no notification, alas - n => { - unlock(self.lock.get()); - panic!("inconsistent park_timeout state: {n}") - } - } - } - - pub fn unpark(self: Pin<&Self>) { - // To ensure the unparked thread will observe any writes we made - // before this call, we must perform a release operation that `park` - // can synchronize with. To do that we must write `NOTIFIED` even if - // `state` is already `NOTIFIED`. That is why this must be a swap - // rather than a compare-and-swap that returns if it reads `NOTIFIED` - // on failure. - match self.state.swap(NOTIFIED, SeqCst) { - EMPTY => return, // no one was waiting - NOTIFIED => return, // already unparked - PARKED => {} // gotta go wake someone up - _ => panic!("inconsistent state in unpark"), - } - - // There is a period between when the parked thread sets `state` to - // `PARKED` (or last checked `state` in the case of a spurious wake - // up) and when it actually waits on `cvar`. If we were to notify - // during this period it would be ignored and then when the parked - // thread went to sleep it would never wake up. Fortunately, it has - // `lock` locked at this stage so we can acquire `lock` to wait until - // it is ready to receive the notification. - // - // Releasing `lock` before the call to `notify_one` means that when the - // parked thread wakes it doesn't get woken only to have to wait for us - // to release `lock`. - unsafe { - lock(self.lock.get()); - unlock(self.lock.get()); - notify_one(self.cvar.get()); - } - } -} - -impl Drop for Parker { - fn drop(&mut self) { - unsafe { - libc::pthread_cond_destroy(self.cvar.get_mut()); - libc::pthread_mutex_destroy(self.lock.get_mut()); - } - } -} - -unsafe impl Sync for Parker {} -unsafe impl Send for Parker {} diff --git a/library/std/src/sys/unix/thread_parker/darwin.rs b/library/std/src/sys/unix/thread_parker/darwin.rs new file mode 100644 index 000000000..2f5356fe2 --- /dev/null +++ b/library/std/src/sys/unix/thread_parker/darwin.rs @@ -0,0 +1,131 @@ +//! Thread parking for Darwin-based systems. +//! +//! Darwin actually has futex syscalls (`__ulock_wait`/`__ulock_wake`), but they +//! cannot be used in `std` because they are non-public (their use will lead to +//! rejection from the App Store) and because they are only available starting +//! with macOS version 10.12, even though the minimum target version is 10.7. +//! +//! Therefore, we need to look for other synchronization primitives. Luckily, Darwin +//! supports semaphores, which allow us to implement the behaviour we need with +//! only one primitive (as opposed to a mutex-condvar pair). We use the semaphore +//! provided by libdispatch, as the underlying Mach semaphore is only dubiously +//! public. + +use crate::pin::Pin; +use crate::sync::atomic::{ + AtomicI8, + Ordering::{Acquire, Release}, +}; +use crate::time::Duration; + +type dispatch_semaphore_t = *mut crate::ffi::c_void; +type dispatch_time_t = u64; + +const DISPATCH_TIME_NOW: dispatch_time_t = 0; +const DISPATCH_TIME_FOREVER: dispatch_time_t = !0; + +// Contained in libSystem.dylib, which is linked by default. +extern "C" { + fn dispatch_time(when: dispatch_time_t, delta: i64) -> dispatch_time_t; + fn dispatch_semaphore_create(val: isize) -> dispatch_semaphore_t; + fn dispatch_semaphore_wait(dsema: dispatch_semaphore_t, timeout: dispatch_time_t) -> isize; + fn dispatch_semaphore_signal(dsema: dispatch_semaphore_t) -> isize; + fn dispatch_release(object: *mut crate::ffi::c_void); +} + +const EMPTY: i8 = 0; +const NOTIFIED: i8 = 1; +const PARKED: i8 = -1; + +pub struct Parker { + semaphore: dispatch_semaphore_t, + state: AtomicI8, +} + +unsafe impl Sync for Parker {} +unsafe impl Send for Parker {} + +impl Parker { + pub unsafe fn new(parker: *mut Parker) { + let semaphore = dispatch_semaphore_create(0); + assert!( + !semaphore.is_null(), + "failed to create dispatch semaphore for thread synchronization" + ); + parker.write(Parker { semaphore, state: AtomicI8::new(EMPTY) }) + } + + // Does not need `Pin`, but other implementation do. + pub unsafe fn park(self: Pin<&Self>) { + // The semaphore counter must be zero at this point, because unparking + // threads will not actually increase it until we signalled that we + // are waiting. + + // Change NOTIFIED to EMPTY and EMPTY to PARKED. + if self.state.fetch_sub(1, Acquire) == NOTIFIED { + return; + } + + // Another thread may increase the semaphore counter from this point on. + // If it is faster than us, we will decrement it again immediately below. + // If we are faster, we wait. + + // Ensure that the semaphore counter has actually been decremented, even + // if the call timed out for some reason. + while dispatch_semaphore_wait(self.semaphore, DISPATCH_TIME_FOREVER) != 0 {} + + // At this point, the semaphore counter is zero again. + + // We were definitely woken up, so we don't need to check the state. + // Still, we need to reset the state using a swap to observe the state + // change with acquire ordering. + self.state.swap(EMPTY, Acquire); + } + + // Does not need `Pin`, but other implementation do. + pub unsafe fn park_timeout(self: Pin<&Self>, dur: Duration) { + if self.state.fetch_sub(1, Acquire) == NOTIFIED { + return; + } + + let nanos = dur.as_nanos().try_into().unwrap_or(i64::MAX); + let timeout = dispatch_time(DISPATCH_TIME_NOW, nanos); + + let timeout = dispatch_semaphore_wait(self.semaphore, timeout) != 0; + + let state = self.state.swap(EMPTY, Acquire); + if state == NOTIFIED && timeout { + // If the state was NOTIFIED but semaphore_wait returned without + // decrementing the count because of a timeout, it means another + // thread is about to call semaphore_signal. We must wait for that + // to happen to ensure the semaphore count is reset. + while dispatch_semaphore_wait(self.semaphore, DISPATCH_TIME_FOREVER) != 0 {} + } else { + // Either a timeout occurred and we reset the state before any thread + // tried to wake us up, or we were woken up and reset the state, + // making sure to observe the state change with acquire ordering. + // Either way, the semaphore counter is now zero again. + } + } + + // Does not need `Pin`, but other implementation do. + pub fn unpark(self: Pin<&Self>) { + let state = self.state.swap(NOTIFIED, Release); + if state == PARKED { + unsafe { + dispatch_semaphore_signal(self.semaphore); + } + } + } +} + +impl Drop for Parker { + fn drop(&mut self) { + // SAFETY: + // We always ensure that the semaphore count is reset, so this will + // never cause an exception. + unsafe { + dispatch_release(self.semaphore); + } + } +} diff --git a/library/std/src/sys/unix/thread_parker/mod.rs b/library/std/src/sys/unix/thread_parker/mod.rs new file mode 100644 index 000000000..35f1e68a8 --- /dev/null +++ b/library/std/src/sys/unix/thread_parker/mod.rs @@ -0,0 +1,32 @@ +//! Thread parking on systems without futex support. + +#![cfg(not(any( + target_os = "linux", + target_os = "android", + all(target_os = "emscripten", target_feature = "atomics"), + target_os = "freebsd", + target_os = "openbsd", + target_os = "dragonfly", + target_os = "fuchsia", +)))] + +cfg_if::cfg_if! { + if #[cfg(all( + any( + target_os = "macos", + target_os = "ios", + target_os = "watchos", + target_os = "tvos", + ), + not(miri), + ))] { + mod darwin; + pub use darwin::Parker; + } else if #[cfg(target_os = "netbsd")] { + mod netbsd; + pub use netbsd::Parker; + } else { + mod pthread; + pub use pthread::Parker; + } +} diff --git a/library/std/src/sys/unix/thread_parker/netbsd.rs b/library/std/src/sys/unix/thread_parker/netbsd.rs new file mode 100644 index 000000000..7657605b5 --- /dev/null +++ b/library/std/src/sys/unix/thread_parker/netbsd.rs @@ -0,0 +1,113 @@ +use crate::ffi::{c_int, c_void}; +use crate::pin::Pin; +use crate::ptr::{null, null_mut}; +use crate::sync::atomic::{ + AtomicU64, + Ordering::{Acquire, Relaxed, Release}, +}; +use crate::time::Duration; +use libc::{_lwp_self, clockid_t, lwpid_t, time_t, timespec, CLOCK_MONOTONIC}; + +extern "C" { + fn ___lwp_park60( + clock_id: clockid_t, + flags: c_int, + ts: *mut timespec, + unpark: lwpid_t, + hint: *const c_void, + unparkhint: *const c_void, + ) -> c_int; + fn _lwp_unpark(lwp: lwpid_t, hint: *const c_void) -> c_int; +} + +/// The thread is not parked and the token is not available. +/// +/// Zero cannot be a valid LWP id, since it is used as empty value for the unpark +/// argument in _lwp_park. +const EMPTY: u64 = 0; +/// The token is available. Do not park anymore. +const NOTIFIED: u64 = u64::MAX; + +pub struct Parker { + /// The parker state. Contains either one of the two state values above or the LWP + /// id of the parked thread. + state: AtomicU64, +} + +impl Parker { + pub unsafe fn new(parker: *mut Parker) { + parker.write(Parker { state: AtomicU64::new(EMPTY) }) + } + + // Does not actually need `unsafe` or `Pin`, but the pthread implementation does. + pub unsafe fn park(self: Pin<&Self>) { + // If the token has already been made available, we can skip + // a bit of work, so check for it here. + if self.state.load(Acquire) != NOTIFIED { + let parked = _lwp_self() as u64; + let hint = self.state.as_mut_ptr().cast(); + if self.state.compare_exchange(EMPTY, parked, Relaxed, Acquire).is_ok() { + // Loop to guard against spurious wakeups. + loop { + ___lwp_park60(0, 0, null_mut(), 0, hint, null()); + if self.state.load(Acquire) == NOTIFIED { + break; + } + } + } + } + + // At this point, the change to NOTIFIED has always been observed with acquire + // ordering, so we can just use a relaxed store here (instead of a swap). + self.state.store(EMPTY, Relaxed); + } + + // Does not actually need `unsafe` or `Pin`, but the pthread implementation does. + pub unsafe fn park_timeout(self: Pin<&Self>, dur: Duration) { + if self.state.load(Acquire) != NOTIFIED { + let parked = _lwp_self() as u64; + let hint = self.state.as_mut_ptr().cast(); + let mut timeout = timespec { + // Saturate so that the operation will definitely time out + // (even if it is after the heat death of the universe). + tv_sec: dur.as_secs().try_into().ok().unwrap_or(time_t::MAX), + tv_nsec: dur.subsec_nanos().into(), + }; + + if self.state.compare_exchange(EMPTY, parked, Relaxed, Acquire).is_ok() { + // Timeout needs to be mutable since it is modified on NetBSD 9.0 and + // above. + ___lwp_park60(CLOCK_MONOTONIC, 0, &mut timeout, 0, hint, null()); + // Use a swap to get acquire ordering even if the token was set after + // the timeout occurred. + self.state.swap(EMPTY, Acquire); + return; + } + } + + self.state.store(EMPTY, Relaxed); + } + + // Does not actually need `Pin`, but the pthread implementation does. + pub fn unpark(self: Pin<&Self>) { + let state = self.state.swap(NOTIFIED, Release); + if !matches!(state, EMPTY | NOTIFIED) { + let lwp = state as lwpid_t; + let hint = self.state.as_mut_ptr().cast(); + + // If the parking thread terminated and did not actually park, this will + // probably return an error, which is OK. In the worst case, another + // thread has received the same LWP id. It will then receive a spurious + // wakeup, but those are allowable per the API contract. The same reasoning + // applies if a timeout occurred before this call, but the state was not + // yet reset. + + // SAFETY: + // The syscall has no invariants to hold. Only unsafe because it is an + // extern function. + unsafe { + _lwp_unpark(lwp, hint); + } + } + } +} diff --git a/library/std/src/sys/unix/thread_parker/pthread.rs b/library/std/src/sys/unix/thread_parker/pthread.rs new file mode 100644 index 000000000..3dfc0026e --- /dev/null +++ b/library/std/src/sys/unix/thread_parker/pthread.rs @@ -0,0 +1,271 @@ +//! Thread parking without `futex` using the `pthread` synchronization primitives. + +use crate::cell::UnsafeCell; +use crate::marker::PhantomPinned; +use crate::pin::Pin; +use crate::ptr::addr_of_mut; +use crate::sync::atomic::AtomicUsize; +use crate::sync::atomic::Ordering::SeqCst; +use crate::time::Duration; + +const EMPTY: usize = 0; +const PARKED: usize = 1; +const NOTIFIED: usize = 2; + +unsafe fn lock(lock: *mut libc::pthread_mutex_t) { + let r = libc::pthread_mutex_lock(lock); + debug_assert_eq!(r, 0); +} + +unsafe fn unlock(lock: *mut libc::pthread_mutex_t) { + let r = libc::pthread_mutex_unlock(lock); + debug_assert_eq!(r, 0); +} + +unsafe fn notify_one(cond: *mut libc::pthread_cond_t) { + let r = libc::pthread_cond_signal(cond); + debug_assert_eq!(r, 0); +} + +unsafe fn wait(cond: *mut libc::pthread_cond_t, lock: *mut libc::pthread_mutex_t) { + let r = libc::pthread_cond_wait(cond, lock); + debug_assert_eq!(r, 0); +} + +const TIMESPEC_MAX: libc::timespec = + libc::timespec { tv_sec: <libc::time_t>::MAX, tv_nsec: 1_000_000_000 - 1 }; + +unsafe fn wait_timeout( + cond: *mut libc::pthread_cond_t, + lock: *mut libc::pthread_mutex_t, + dur: Duration, +) { + // Use the system clock on systems that do not support pthread_condattr_setclock. + // This unfortunately results in problems when the system time changes. + #[cfg(any( + target_os = "macos", + target_os = "ios", + target_os = "watchos", + target_os = "espidf" + ))] + let (now, dur) = { + use crate::cmp::min; + use crate::sys::time::SystemTime; + + // OSX implementation of `pthread_cond_timedwait` is buggy + // with super long durations. When duration is greater than + // 0x100_0000_0000_0000 seconds, `pthread_cond_timedwait` + // in macOS Sierra return error 316. + // + // This program demonstrates the issue: + // https://gist.github.com/stepancheg/198db4623a20aad2ad7cddb8fda4a63c + // + // To work around this issue, and possible bugs of other OSes, timeout + // is clamped to 1000 years, which is allowable per the API of `park_timeout` + // because of spurious wakeups. + let dur = min(dur, Duration::from_secs(1000 * 365 * 86400)); + let now = SystemTime::now().t; + (now, dur) + }; + // Use the monotonic clock on other systems. + #[cfg(not(any( + target_os = "macos", + target_os = "ios", + target_os = "watchos", + target_os = "espidf" + )))] + let (now, dur) = { + use crate::sys::time::Timespec; + + (Timespec::now(libc::CLOCK_MONOTONIC), dur) + }; + + let timeout = + now.checked_add_duration(&dur).and_then(|t| t.to_timespec()).unwrap_or(TIMESPEC_MAX); + let r = libc::pthread_cond_timedwait(cond, lock, &timeout); + debug_assert!(r == libc::ETIMEDOUT || r == 0); +} + +pub struct Parker { + state: AtomicUsize, + lock: UnsafeCell<libc::pthread_mutex_t>, + cvar: UnsafeCell<libc::pthread_cond_t>, + // The `pthread` primitives require a stable address, so make this struct `!Unpin`. + _pinned: PhantomPinned, +} + +impl Parker { + /// Construct the UNIX parker in-place. + /// + /// # Safety + /// The constructed parker must never be moved. + pub unsafe fn new(parker: *mut Parker) { + // Use the default mutex implementation to allow for simpler initialization. + // This could lead to undefined behaviour when deadlocking. This is avoided + // by not deadlocking. Note in particular the unlocking operation before any + // panic, as code after the panic could try to park again. + addr_of_mut!((*parker).state).write(AtomicUsize::new(EMPTY)); + addr_of_mut!((*parker).lock).write(UnsafeCell::new(libc::PTHREAD_MUTEX_INITIALIZER)); + + cfg_if::cfg_if! { + if #[cfg(any( + target_os = "macos", + target_os = "ios", + target_os = "watchos", + target_os = "l4re", + target_os = "android", + target_os = "redox" + ))] { + addr_of_mut!((*parker).cvar).write(UnsafeCell::new(libc::PTHREAD_COND_INITIALIZER)); + } else if #[cfg(any(target_os = "espidf", target_os = "horizon"))] { + let r = libc::pthread_cond_init(addr_of_mut!((*parker).cvar).cast(), crate::ptr::null()); + assert_eq!(r, 0); + } else { + use crate::mem::MaybeUninit; + let mut attr = MaybeUninit::<libc::pthread_condattr_t>::uninit(); + let r = libc::pthread_condattr_init(attr.as_mut_ptr()); + assert_eq!(r, 0); + let r = libc::pthread_condattr_setclock(attr.as_mut_ptr(), libc::CLOCK_MONOTONIC); + assert_eq!(r, 0); + let r = libc::pthread_cond_init(addr_of_mut!((*parker).cvar).cast(), attr.as_ptr()); + assert_eq!(r, 0); + let r = libc::pthread_condattr_destroy(attr.as_mut_ptr()); + assert_eq!(r, 0); + } + } + } + + // This implementation doesn't require `unsafe`, but other implementations + // may assume this is only called by the thread that owns the Parker. + pub unsafe fn park(self: Pin<&Self>) { + // If we were previously notified then we consume this notification and + // return quickly. + if self.state.compare_exchange(NOTIFIED, EMPTY, SeqCst, SeqCst).is_ok() { + return; + } + + // Otherwise we need to coordinate going to sleep + lock(self.lock.get()); + match self.state.compare_exchange(EMPTY, PARKED, SeqCst, SeqCst) { + Ok(_) => {} + Err(NOTIFIED) => { + // We must read here, even though we know it will be `NOTIFIED`. + // This is because `unpark` may have been called again since we read + // `NOTIFIED` in the `compare_exchange` above. We must perform an + // acquire operation that synchronizes with that `unpark` to observe + // any writes it made before the call to unpark. To do that we must + // read from the write it made to `state`. + let old = self.state.swap(EMPTY, SeqCst); + + unlock(self.lock.get()); + + assert_eq!(old, NOTIFIED, "park state changed unexpectedly"); + return; + } // should consume this notification, so prohibit spurious wakeups in next park. + Err(_) => { + unlock(self.lock.get()); + + panic!("inconsistent park state") + } + } + + loop { + wait(self.cvar.get(), self.lock.get()); + + match self.state.compare_exchange(NOTIFIED, EMPTY, SeqCst, SeqCst) { + Ok(_) => break, // got a notification + Err(_) => {} // spurious wakeup, go back to sleep + } + } + + unlock(self.lock.get()); + } + + // This implementation doesn't require `unsafe`, but other implementations + // may assume this is only called by the thread that owns the Parker. Use + // `Pin` to guarantee a stable address for the mutex and condition variable. + pub unsafe fn park_timeout(self: Pin<&Self>, dur: Duration) { + // Like `park` above we have a fast path for an already-notified thread, and + // afterwards we start coordinating for a sleep. + // return quickly. + if self.state.compare_exchange(NOTIFIED, EMPTY, SeqCst, SeqCst).is_ok() { + return; + } + + lock(self.lock.get()); + match self.state.compare_exchange(EMPTY, PARKED, SeqCst, SeqCst) { + Ok(_) => {} + Err(NOTIFIED) => { + // We must read again here, see `park`. + let old = self.state.swap(EMPTY, SeqCst); + unlock(self.lock.get()); + + assert_eq!(old, NOTIFIED, "park state changed unexpectedly"); + return; + } // should consume this notification, so prohibit spurious wakeups in next park. + Err(_) => { + unlock(self.lock.get()); + panic!("inconsistent park_timeout state") + } + } + + // Wait with a timeout, and if we spuriously wake up or otherwise wake up + // from a notification we just want to unconditionally set the state back to + // empty, either consuming a notification or un-flagging ourselves as + // parked. + wait_timeout(self.cvar.get(), self.lock.get(), dur); + + match self.state.swap(EMPTY, SeqCst) { + NOTIFIED => unlock(self.lock.get()), // got a notification, hurray! + PARKED => unlock(self.lock.get()), // no notification, alas + n => { + unlock(self.lock.get()); + panic!("inconsistent park_timeout state: {n}") + } + } + } + + pub fn unpark(self: Pin<&Self>) { + // To ensure the unparked thread will observe any writes we made + // before this call, we must perform a release operation that `park` + // can synchronize with. To do that we must write `NOTIFIED` even if + // `state` is already `NOTIFIED`. That is why this must be a swap + // rather than a compare-and-swap that returns if it reads `NOTIFIED` + // on failure. + match self.state.swap(NOTIFIED, SeqCst) { + EMPTY => return, // no one was waiting + NOTIFIED => return, // already unparked + PARKED => {} // gotta go wake someone up + _ => panic!("inconsistent state in unpark"), + } + + // There is a period between when the parked thread sets `state` to + // `PARKED` (or last checked `state` in the case of a spurious wake + // up) and when it actually waits on `cvar`. If we were to notify + // during this period it would be ignored and then when the parked + // thread went to sleep it would never wake up. Fortunately, it has + // `lock` locked at this stage so we can acquire `lock` to wait until + // it is ready to receive the notification. + // + // Releasing `lock` before the call to `notify_one` means that when the + // parked thread wakes it doesn't get woken only to have to wait for us + // to release `lock`. + unsafe { + lock(self.lock.get()); + unlock(self.lock.get()); + notify_one(self.cvar.get()); + } + } +} + +impl Drop for Parker { + fn drop(&mut self) { + unsafe { + libc::pthread_cond_destroy(self.cvar.get_mut()); + libc::pthread_mutex_destroy(self.lock.get_mut()); + } + } +} + +unsafe impl Sync for Parker {} +unsafe impl Send for Parker {} diff --git a/library/std/src/sys/unix/time.rs b/library/std/src/sys/unix/time.rs index dff973f59..cca9c6767 100644 --- a/library/std/src/sys/unix/time.rs +++ b/library/std/src/sys/unix/time.rs @@ -7,6 +7,12 @@ const NSEC_PER_SEC: u64 = 1_000_000_000; pub const UNIX_EPOCH: SystemTime = SystemTime { t: Timespec::zero() }; #[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash)] +#[repr(transparent)] +#[rustc_layout_scalar_valid_range_start(0)] +#[rustc_layout_scalar_valid_range_end(999_999_999)] +struct Nanoseconds(u32); + +#[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash)] pub struct SystemTime { pub(in crate::sys::unix) t: Timespec, } @@ -14,7 +20,7 @@ pub struct SystemTime { #[derive(Copy, Clone, PartialEq, Eq, PartialOrd, Ord, Hash)] pub(in crate::sys::unix) struct Timespec { tv_sec: i64, - tv_nsec: i64, + tv_nsec: Nanoseconds, } impl SystemTime { @@ -46,18 +52,20 @@ impl fmt::Debug for SystemTime { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { f.debug_struct("SystemTime") .field("tv_sec", &self.t.tv_sec) - .field("tv_nsec", &self.t.tv_nsec) + .field("tv_nsec", &self.t.tv_nsec.0) .finish() } } impl Timespec { pub const fn zero() -> Timespec { - Timespec { tv_sec: 0, tv_nsec: 0 } + Timespec::new(0, 0) } - fn new(tv_sec: i64, tv_nsec: i64) -> Timespec { - Timespec { tv_sec, tv_nsec } + const fn new(tv_sec: i64, tv_nsec: i64) -> Timespec { + assert!(tv_nsec >= 0 && tv_nsec < NSEC_PER_SEC as i64); + // SAFETY: The assert above checks tv_nsec is within the valid range + Timespec { tv_sec, tv_nsec: unsafe { Nanoseconds(tv_nsec as u32) } } } pub fn sub_timespec(&self, other: &Timespec) -> Result<Duration, Duration> { @@ -75,12 +83,12 @@ impl Timespec { // // Ideally this code could be rearranged such that it more // directly expresses the lower-cost behavior we want from it. - let (secs, nsec) = if self.tv_nsec >= other.tv_nsec { - ((self.tv_sec - other.tv_sec) as u64, (self.tv_nsec - other.tv_nsec) as u32) + let (secs, nsec) = if self.tv_nsec.0 >= other.tv_nsec.0 { + ((self.tv_sec - other.tv_sec) as u64, self.tv_nsec.0 - other.tv_nsec.0) } else { ( (self.tv_sec - other.tv_sec - 1) as u64, - self.tv_nsec as u32 + (NSEC_PER_SEC as u32) - other.tv_nsec as u32, + self.tv_nsec.0 + (NSEC_PER_SEC as u32) - other.tv_nsec.0, ) }; @@ -102,7 +110,7 @@ impl Timespec { // Nano calculations can't overflow because nanos are <1B which fit // in a u32. - let mut nsec = other.subsec_nanos() + self.tv_nsec as u32; + let mut nsec = other.subsec_nanos() + self.tv_nsec.0; if nsec >= NSEC_PER_SEC as u32 { nsec -= NSEC_PER_SEC as u32; secs = secs.checked_add(1)?; @@ -118,7 +126,7 @@ impl Timespec { .and_then(|secs| self.tv_sec.checked_sub(secs))?; // Similar to above, nanos can't overflow. - let mut nsec = self.tv_nsec as i32 - other.subsec_nanos() as i32; + let mut nsec = self.tv_nsec.0 as i32 - other.subsec_nanos() as i32; if nsec < 0 { nsec += NSEC_PER_SEC as i32; secs = secs.checked_sub(1)?; @@ -130,7 +138,7 @@ impl Timespec { pub fn to_timespec(&self) -> Option<libc::timespec> { Some(libc::timespec { tv_sec: self.tv_sec.try_into().ok()?, - tv_nsec: self.tv_nsec.try_into().ok()?, + tv_nsec: self.tv_nsec.0.try_into().ok()?, }) } } @@ -293,7 +301,7 @@ mod inner { fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { f.debug_struct("Instant") .field("tv_sec", &self.t.tv_sec) - .field("tv_nsec", &self.t.tv_nsec) + .field("tv_nsec", &self.t.tv_nsec.0) .finish() } } @@ -334,7 +342,7 @@ mod inner { let mut t = MaybeUninit::uninit(); cvt(unsafe { clock_gettime64(clock, t.as_mut_ptr()) }).unwrap(); let t = unsafe { t.assume_init() }; - return Timespec { tv_sec: t.tv_sec, tv_nsec: t.tv_nsec as i64 }; + return Timespec::new(t.tv_sec, t.tv_nsec as i64); } } diff --git a/library/std/src/sys/unsupported/alloc.rs b/library/std/src/sys/unsupported/alloc.rs index 8d5d0a2f5..d715ae454 100644 --- a/library/std/src/sys/unsupported/alloc.rs +++ b/library/std/src/sys/unsupported/alloc.rs @@ -1,15 +1,16 @@ use crate::alloc::{GlobalAlloc, Layout, System}; +use crate::ptr::null_mut; #[stable(feature = "alloc_system_type", since = "1.28.0")] unsafe impl GlobalAlloc for System { #[inline] unsafe fn alloc(&self, _layout: Layout) -> *mut u8 { - 0 as *mut u8 + null_mut() } #[inline] unsafe fn alloc_zeroed(&self, _layout: Layout) -> *mut u8 { - 0 as *mut u8 + null_mut() } #[inline] @@ -17,6 +18,6 @@ unsafe impl GlobalAlloc for System { #[inline] unsafe fn realloc(&self, _ptr: *mut u8, _layout: Layout, _new_size: usize) -> *mut u8 { - 0 as *mut u8 + null_mut() } } diff --git a/library/std/src/sys/unsupported/common.rs b/library/std/src/sys/unsupported/common.rs index 4c9ade4a8..5cd9e57de 100644 --- a/library/std/src/sys/unsupported/common.rs +++ b/library/std/src/sys/unsupported/common.rs @@ -6,7 +6,7 @@ pub mod memchr { // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. -pub unsafe fn init(_argc: isize, _argv: *const *const u8) {} +pub unsafe fn init(_argc: isize, _argv: *const *const u8, _sigpipe: u8) {} // SAFETY: must be called only once during runtime cleanup. // NOTE: this is not guaranteed to run, for example when the program aborts. diff --git a/library/std/src/sys/unsupported/fs.rs b/library/std/src/sys/unsupported/fs.rs index 0e1a6257e..6ac1b5d2b 100644 --- a/library/std/src/sys/unsupported/fs.rs +++ b/library/std/src/sys/unsupported/fs.rs @@ -1,7 +1,7 @@ use crate::ffi::OsString; use crate::fmt; use crate::hash::{Hash, Hasher}; -use crate::io::{self, IoSlice, IoSliceMut, ReadBuf, SeekFrom}; +use crate::io::{self, BorrowedCursor, IoSlice, IoSliceMut, SeekFrom}; use crate::path::{Path, PathBuf}; use crate::sys::time::SystemTime; use crate::sys::unsupported; @@ -214,7 +214,7 @@ impl File { self.0 } - pub fn read_buf(&self, _buf: &mut ReadBuf<'_>) -> io::Result<()> { + pub fn read_buf(&self, _cursor: BorrowedCursor<'_>) -> io::Result<()> { self.0 } diff --git a/library/std/src/sys/unsupported/io.rs b/library/std/src/sys/unsupported/io.rs index d5f475b43..82610ffab 100644 --- a/library/std/src/sys/unsupported/io.rs +++ b/library/std/src/sys/unsupported/io.rs @@ -45,3 +45,7 @@ impl<'a> IoSliceMut<'a> { self.0 } } + +pub fn is_terminal<T>(_: &T) -> bool { + false +} diff --git a/library/std/src/sys/unsupported/locks/condvar.rs b/library/std/src/sys/unsupported/locks/condvar.rs index e703fd0d2..527a26a12 100644 --- a/library/std/src/sys/unsupported/locks/condvar.rs +++ b/library/std/src/sys/unsupported/locks/condvar.rs @@ -7,6 +7,7 @@ pub type MovableCondvar = Condvar; impl Condvar { #[inline] + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> Condvar { Condvar {} } diff --git a/library/std/src/sys/unsupported/locks/mod.rs b/library/std/src/sys/unsupported/locks/mod.rs index d412ff152..602a2d623 100644 --- a/library/std/src/sys/unsupported/locks/mod.rs +++ b/library/std/src/sys/unsupported/locks/mod.rs @@ -3,4 +3,4 @@ mod mutex; mod rwlock; pub use condvar::{Condvar, MovableCondvar}; pub use mutex::{MovableMutex, Mutex}; -pub use rwlock::{MovableRwLock, RwLock}; +pub use rwlock::MovableRwLock; diff --git a/library/std/src/sys/unsupported/locks/mutex.rs b/library/std/src/sys/unsupported/locks/mutex.rs index d7cb12e0c..87ea475c6 100644 --- a/library/std/src/sys/unsupported/locks/mutex.rs +++ b/library/std/src/sys/unsupported/locks/mutex.rs @@ -12,14 +12,12 @@ unsafe impl Sync for Mutex {} // no threads on this platform impl Mutex { #[inline] + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> Mutex { Mutex { locked: Cell::new(false) } } #[inline] - pub unsafe fn init(&mut self) {} - - #[inline] pub unsafe fn lock(&self) { assert_eq!(self.locked.replace(true), false, "cannot recursively acquire mutex"); } diff --git a/library/std/src/sys/unsupported/locks/rwlock.rs b/library/std/src/sys/unsupported/locks/rwlock.rs index aca5fb715..5292691b9 100644 --- a/library/std/src/sys/unsupported/locks/rwlock.rs +++ b/library/std/src/sys/unsupported/locks/rwlock.rs @@ -12,6 +12,7 @@ unsafe impl Sync for RwLock {} // no threads on this platform impl RwLock { #[inline] + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> RwLock { RwLock { mode: Cell::new(0) } } diff --git a/library/std/src/sys/unsupported/process.rs b/library/std/src/sys/unsupported/process.rs index 42a1ff730..633f17c05 100644 --- a/library/std/src/sys/unsupported/process.rs +++ b/library/std/src/sys/unsupported/process.rs @@ -200,6 +200,9 @@ impl<'a> Iterator for CommandArgs<'a> { fn next(&mut self) -> Option<&'a OsStr> { None } + fn size_hint(&self) -> (usize, Option<usize>) { + (0, Some(0)) + } } impl<'a> ExactSizeIterator for CommandArgs<'a> {} diff --git a/library/std/src/sys/unsupported/thread_local_dtor.rs b/library/std/src/sys/unsupported/thread_local_dtor.rs index 85d660983..84660ea58 100644 --- a/library/std/src/sys/unsupported/thread_local_dtor.rs +++ b/library/std/src/sys/unsupported/thread_local_dtor.rs @@ -1,5 +1,6 @@ #![unstable(feature = "thread_local_internals", issue = "none")] +#[cfg_attr(target_family = "wasm", allow(unused))] // unused on wasm32-unknown-unknown pub unsafe fn register_dtor(_t: *mut u8, _dtor: unsafe extern "C" fn(*mut u8)) { // FIXME: right now there is no concept of "thread exit", but this is likely // going to show up at some point in the form of an exported symbol that the diff --git a/library/std/src/sys/unsupported/thread_local_key.rs b/library/std/src/sys/unsupported/thread_local_key.rs index c31b61cbf..b6e5e4cd2 100644 --- a/library/std/src/sys/unsupported/thread_local_key.rs +++ b/library/std/src/sys/unsupported/thread_local_key.rs @@ -19,8 +19,3 @@ pub unsafe fn get(_key: Key) -> *mut u8 { pub unsafe fn destroy(_key: Key) { panic!("should not be used on this target"); } - -#[inline] -pub fn requires_synchronized_create() -> bool { - panic!("should not be used on this target"); -} diff --git a/library/std/src/sys/wasi/fs.rs b/library/std/src/sys/wasi/fs.rs index 6614ae397..d4866bbc3 100644 --- a/library/std/src/sys/wasi/fs.rs +++ b/library/std/src/sys/wasi/fs.rs @@ -1,9 +1,9 @@ #![deny(unsafe_op_in_unsafe_fn)] use super::fd::WasiFd; -use crate::ffi::{CStr, CString, OsStr, OsString}; +use crate::ffi::{CStr, OsStr, OsString}; use crate::fmt; -use crate::io::{self, IoSlice, IoSliceMut, ReadBuf, SeekFrom}; +use crate::io::{self, BorrowedCursor, IoSlice, IoSliceMut, SeekFrom}; use crate::iter; use crate::mem::{self, ManuallyDrop}; use crate::os::raw::c_int; @@ -12,6 +12,7 @@ use crate::os::wasi::io::{AsFd, AsRawFd, BorrowedFd, FromRawFd, IntoRawFd, RawFd use crate::path::{Path, PathBuf}; use crate::ptr; use crate::sync::Arc; +use crate::sys::common::small_c_string::run_path_with_cstr; use crate::sys::time::SystemTime; use crate::sys::unsupported; use crate::sys_common::{AsInner, FromInner, IntoInner}; @@ -65,8 +66,8 @@ pub struct FilePermissions { #[derive(Copy, Clone, Debug, Default)] pub struct FileTimes { - accessed: Option<wasi::Timestamp>, - modified: Option<wasi::Timestamp>, + accessed: Option<SystemTime>, + modified: Option<SystemTime>, } #[derive(PartialEq, Eq, Hash, Debug, Copy, Clone)] @@ -120,11 +121,11 @@ impl FilePermissions { impl FileTimes { pub fn set_accessed(&mut self, t: SystemTime) { - self.accessed = Some(t.to_wasi_timestamp_or_panic()); + self.accessed = Some(t); } pub fn set_modified(&mut self, t: SystemTime) { - self.modified = Some(t.to_wasi_timestamp_or_panic()); + self.modified = Some(t); } } @@ -439,8 +440,8 @@ impl File { true } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - crate::io::default_read_buf(|buf| self.read(buf), buf) + pub fn read_buf(&self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + crate::io::default_read_buf(|buf| self.read(buf), cursor) } pub fn write(&self, buf: &[u8]) -> io::Result<usize> { @@ -476,9 +477,16 @@ impl File { } pub fn set_times(&self, times: FileTimes) -> io::Result<()> { + let to_timestamp = |time: Option<SystemTime>| { + match time { + Some(time) if let Some(ts) = time.to_wasi_timestamp() => Ok(ts), + Some(_) => Err(io::const_io_error!(io::ErrorKind::InvalidInput, "timestamp is too large to set as a file time")), + None => Ok(0), + } + }; self.fd.filestat_set_times( - times.accessed.unwrap_or(0), - times.modified.unwrap_or(0), + to_timestamp(times.accessed)?, + to_timestamp(times.modified)?, times.accessed.map_or(0, |_| wasi::FSTFLAGS_ATIM) | times.modified.map_or(0, |_| wasi::FSTFLAGS_MTIM), ) @@ -687,51 +695,52 @@ fn open_at(fd: &WasiFd, path: &Path, opts: &OpenOptions) -> io::Result<File> { /// Note that this can fail if `p` doesn't look like it can be opened relative /// to any pre-opened file descriptor. fn open_parent(p: &Path) -> io::Result<(ManuallyDrop<WasiFd>, PathBuf)> { - let p = CString::new(p.as_os_str().as_bytes())?; - let mut buf = Vec::<u8>::with_capacity(512); - loop { - unsafe { - let mut relative_path = buf.as_ptr().cast(); - let mut abs_prefix = ptr::null(); - let fd = __wasilibc_find_relpath( - p.as_ptr(), - &mut abs_prefix, - &mut relative_path, - buf.capacity(), - ); - if fd == -1 { - if io::Error::last_os_error().raw_os_error() == Some(libc::ENOMEM) { - // Trigger the internal buffer resizing logic of `Vec` by requiring - // more space than the current capacity. - let cap = buf.capacity(); - buf.set_len(cap); - buf.reserve(1); - continue; - } - let msg = format!( - "failed to find a pre-opened file descriptor \ - through which {:?} could be opened", - p + run_path_with_cstr(p, |p| { + let mut buf = Vec::<u8>::with_capacity(512); + loop { + unsafe { + let mut relative_path = buf.as_ptr().cast(); + let mut abs_prefix = ptr::null(); + let fd = __wasilibc_find_relpath( + p.as_ptr(), + &mut abs_prefix, + &mut relative_path, + buf.capacity(), ); - return Err(io::Error::new(io::ErrorKind::Uncategorized, msg)); - } - let relative = CStr::from_ptr(relative_path).to_bytes().to_vec(); + if fd == -1 { + if io::Error::last_os_error().raw_os_error() == Some(libc::ENOMEM) { + // Trigger the internal buffer resizing logic of `Vec` by requiring + // more space than the current capacity. + let cap = buf.capacity(); + buf.set_len(cap); + buf.reserve(1); + continue; + } + let msg = format!( + "failed to find a pre-opened file descriptor \ + through which {:?} could be opened", + p + ); + return Err(io::Error::new(io::ErrorKind::Uncategorized, msg)); + } + let relative = CStr::from_ptr(relative_path).to_bytes().to_vec(); - return Ok(( - ManuallyDrop::new(WasiFd::from_raw_fd(fd as c_int)), - PathBuf::from(OsString::from_vec(relative)), - )); + return Ok(( + ManuallyDrop::new(WasiFd::from_raw_fd(fd as c_int)), + PathBuf::from(OsString::from_vec(relative)), + )); + } } - } - extern "C" { - pub fn __wasilibc_find_relpath( - path: *const libc::c_char, - abs_prefix: *mut *const libc::c_char, - relative_path: *mut *const libc::c_char, - relative_path_len: libc::size_t, - ) -> libc::c_int; - } + extern "C" { + pub fn __wasilibc_find_relpath( + path: *const libc::c_char, + abs_prefix: *mut *const libc::c_char, + relative_path: *mut *const libc::c_char, + relative_path_len: libc::size_t, + ) -> libc::c_int; + } + }) } pub fn osstr2str(f: &OsStr) -> io::Result<&str> { diff --git a/library/std/src/sys/wasi/io.rs b/library/std/src/sys/wasi/io.rs index ee017d13a..2cd45df88 100644 --- a/library/std/src/sys/wasi/io.rs +++ b/library/std/src/sys/wasi/io.rs @@ -1,6 +1,7 @@ #![deny(unsafe_op_in_unsafe_fn)] use crate::marker::PhantomData; +use crate::os::fd::{AsFd, AsRawFd}; use crate::slice; #[derive(Copy, Clone)] @@ -71,3 +72,8 @@ impl<'a> IoSliceMut<'a> { unsafe { slice::from_raw_parts_mut(self.vec.buf as *mut u8, self.vec.buf_len) } } } + +pub fn is_terminal(fd: &impl AsFd) -> bool { + let fd = fd.as_fd(); + unsafe { libc::isatty(fd.as_raw_fd()) != 0 } +} diff --git a/library/std/src/sys/wasi/mod.rs b/library/std/src/sys/wasi/mod.rs index 683a07a34..c8c47763a 100644 --- a/library/std/src/sys/wasi/mod.rs +++ b/library/std/src/sys/wasi/mod.rs @@ -25,6 +25,9 @@ pub mod cmath; pub mod env; pub mod fd; pub mod fs; +#[allow(unused)] +#[path = "../wasm/atomics/futex.rs"] +pub mod futex; pub mod io; #[path = "../unsupported/locks/mod.rs"] pub mod locks; diff --git a/library/std/src/sys/wasi/os.rs b/library/std/src/sys/wasi/os.rs index c5229a188..f5513e999 100644 --- a/library/std/src/sys/wasi/os.rs +++ b/library/std/src/sys/wasi/os.rs @@ -1,14 +1,15 @@ #![deny(unsafe_op_in_unsafe_fn)] -use crate::any::Any; use crate::error::Error as StdError; -use crate::ffi::{CStr, CString, OsStr, OsString}; +use crate::ffi::{CStr, OsStr, OsString}; use crate::fmt; use crate::io; use crate::marker::PhantomData; +use crate::ops::Drop; use crate::os::wasi::prelude::*; use crate::path::{self, PathBuf}; use crate::str; +use crate::sys::common::small_c_string::{run_path_with_cstr, run_with_cstr}; use crate::sys::memchr; use crate::sys::unsupported; use crate::vec; @@ -23,10 +24,26 @@ mod libc { } } -#[cfg(not(target_feature = "atomics"))] -pub unsafe fn env_lock() -> impl Any { - // No need for a lock if we're single-threaded, but this function will need - // to get implemented for multi-threaded scenarios +cfg_if::cfg_if! { + if #[cfg(target_feature = "atomics")] { + // Access to the environment must be protected by a lock in multi-threaded scenarios. + use crate::sync::{PoisonError, RwLock}; + static ENV_LOCK: RwLock<()> = RwLock::new(()); + pub fn env_read_lock() -> impl Drop { + ENV_LOCK.read().unwrap_or_else(PoisonError::into_inner) + } + pub fn env_write_lock() -> impl Drop { + ENV_LOCK.write().unwrap_or_else(PoisonError::into_inner) + } + } else { + // No need for a lock if we are single-threaded. + pub fn env_read_lock() -> impl Drop { + Box::new(()) + } + pub fn env_write_lock() -> impl Drop { + Box::new(()) + } + } } pub fn errno() -> i32 { @@ -77,13 +94,10 @@ pub fn getcwd() -> io::Result<PathBuf> { } pub fn chdir(p: &path::Path) -> io::Result<()> { - let p: &OsStr = p.as_ref(); - let p = CString::new(p.as_bytes())?; - unsafe { - match libc::chdir(p.as_ptr()) == (0 as libc::c_int) { - true => Ok(()), - false => Err(io::Error::last_os_error()), - } + let result = run_path_with_cstr(p, |p| unsafe { Ok(libc::chdir(p.as_ptr())) })?; + match result == (0 as libc::c_int) { + true => Ok(()), + false => Err(io::Error::last_os_error()), } } @@ -146,7 +160,7 @@ impl Iterator for Env { pub fn env() -> Env { unsafe { - let _guard = env_lock(); + let _guard = env_read_lock(); let mut environ = libc::environ; let mut result = Vec::new(); if !environ.is_null() { @@ -176,35 +190,32 @@ pub fn env() -> Env { } pub fn getenv(k: &OsStr) -> Option<OsString> { - let k = CString::new(k.as_bytes()).ok()?; - unsafe { - let _guard = env_lock(); - let s = libc::getenv(k.as_ptr()) as *const libc::c_char; - if s.is_null() { - None - } else { - Some(OsStringExt::from_vec(CStr::from_ptr(s).to_bytes().to_vec())) - } + let s = run_with_cstr(k.as_bytes(), |k| unsafe { + let _guard = env_read_lock(); + Ok(libc::getenv(k.as_ptr()) as *const libc::c_char) + }) + .ok()?; + if s.is_null() { + None + } else { + Some(OsStringExt::from_vec(unsafe { CStr::from_ptr(s) }.to_bytes().to_vec())) } } pub fn setenv(k: &OsStr, v: &OsStr) -> io::Result<()> { - let k = CString::new(k.as_bytes())?; - let v = CString::new(v.as_bytes())?; - - unsafe { - let _guard = env_lock(); - cvt(libc::setenv(k.as_ptr(), v.as_ptr(), 1)).map(drop) - } + run_with_cstr(k.as_bytes(), |k| { + run_with_cstr(v.as_bytes(), |v| unsafe { + let _guard = env_write_lock(); + cvt(libc::setenv(k.as_ptr(), v.as_ptr(), 1)).map(drop) + }) + }) } pub fn unsetenv(n: &OsStr) -> io::Result<()> { - let nbuf = CString::new(n.as_bytes())?; - - unsafe { - let _guard = env_lock(); + run_with_cstr(n.as_bytes(), |nbuf| unsafe { + let _guard = env_write_lock(); cvt(libc::unsetenv(nbuf.as_ptr())).map(drop) - } + }) } pub fn temp_dir() -> PathBuf { diff --git a/library/std/src/sys/wasi/time.rs b/library/std/src/sys/wasi/time.rs index 3d326e491..016b06efb 100644 --- a/library/std/src/sys/wasi/time.rs +++ b/library/std/src/sys/wasi/time.rs @@ -47,8 +47,8 @@ impl SystemTime { SystemTime(Duration::from_nanos(ts)) } - pub fn to_wasi_timestamp_or_panic(&self) -> wasi::Timestamp { - self.0.as_nanos().try_into().expect("time does not fit in WASI timestamp") + pub fn to_wasi_timestamp(&self) -> Option<wasi::Timestamp> { + self.0.as_nanos().try_into().ok() } pub fn sub_time(&self, other: &SystemTime) -> Result<Duration, Duration> { diff --git a/library/std/src/sys/wasm/mod.rs b/library/std/src/sys/wasm/mod.rs index 4159efe2a..93838390b 100644 --- a/library/std/src/sys/wasm/mod.rs +++ b/library/std/src/sys/wasm/mod.rs @@ -57,7 +57,7 @@ cfg_if::cfg_if! { mod futex_rwlock; pub(crate) use futex_condvar::{Condvar, MovableCondvar}; pub(crate) use futex_mutex::{Mutex, MovableMutex}; - pub(crate) use futex_rwlock::{RwLock, MovableRwLock}; + pub(crate) use futex_rwlock::MovableRwLock; } #[path = "atomics/futex.rs"] pub mod futex; diff --git a/library/std/src/sys/windows/alloc.rs b/library/std/src/sys/windows/alloc.rs index fdc81cdea..d53ea1600 100644 --- a/library/std/src/sys/windows/alloc.rs +++ b/library/std/src/sys/windows/alloc.rs @@ -16,6 +16,7 @@ mod tests; // Flag to indicate that the memory returned by `HeapAlloc` should be zeroed. const HEAP_ZERO_MEMORY: c::DWORD = 0x00000008; +#[link(name = "kernel32")] extern "system" { // Get a handle to the default heap of the current process, or null if the operation fails. // @@ -168,7 +169,7 @@ unsafe fn allocate(layout: Layout, zeroed: bool) -> *mut u8 { // SAFETY: Because the size and alignment of a header is <= `MIN_ALIGN` and `aligned` // is aligned to at least `MIN_ALIGN` and has at least `MIN_ALIGN` bytes of padding before // it, it is safe to write a header directly before it. - unsafe { ptr::write((aligned as *mut Header).offset(-1), Header(ptr)) }; + unsafe { ptr::write((aligned as *mut Header).sub(1), Header(ptr)) }; // SAFETY: The returned pointer does not point to the to the start of an allocated block, // but there is a header readable directly before it containing the location of the start @@ -213,7 +214,7 @@ unsafe impl GlobalAlloc for System { // SAFETY: Because of the contract of `System`, `ptr` is guaranteed to be non-null // and have a header readable directly before it. - unsafe { ptr::read((ptr as *mut Header).offset(-1)).0 } + unsafe { ptr::read((ptr as *mut Header).sub(1)).0 } } }; diff --git a/library/std/src/sys/windows/c.rs b/library/std/src/sys/windows/c.rs index 478068c73..be6fc2ebb 100644 --- a/library/std/src/sys/windows/c.rs +++ b/library/std/src/sys/windows/c.rs @@ -66,10 +66,12 @@ pub type LPSYSTEM_INFO = *mut SYSTEM_INFO; pub type LPWSABUF = *mut WSABUF; pub type LPWSAOVERLAPPED = *mut c_void; pub type LPWSAOVERLAPPED_COMPLETION_ROUTINE = *mut c_void; +pub type BCRYPT_ALG_HANDLE = LPVOID; pub type PCONDITION_VARIABLE = *mut CONDITION_VARIABLE; pub type PLARGE_INTEGER = *mut c_longlong; pub type PSRWLOCK = *mut SRWLOCK; +pub type LPINIT_ONCE = *mut INIT_ONCE; pub type SOCKET = crate::os::windows::raw::SOCKET; pub type socklen_t = c_int; @@ -125,6 +127,10 @@ pub const SECURITY_SQOS_PRESENT: DWORD = 0x00100000; pub const FIONBIO: c_ulong = 0x8004667e; +pub const MAX_PATH: usize = 260; + +pub const FILE_TYPE_PIPE: u32 = 3; + #[repr(C)] #[derive(Copy)] pub struct WIN32_FIND_DATAW { @@ -193,6 +199,9 @@ pub const DUPLICATE_SAME_ACCESS: DWORD = 0x00000002; pub const CONDITION_VARIABLE_INIT: CONDITION_VARIABLE = CONDITION_VARIABLE { ptr: ptr::null_mut() }; pub const SRWLOCK_INIT: SRWLOCK = SRWLOCK { ptr: ptr::null_mut() }; +pub const INIT_ONCE_STATIC_INIT: INIT_ONCE = INIT_ONCE { ptr: ptr::null_mut() }; + +pub const INIT_ONCE_INIT_FAILED: DWORD = 0x00000004; pub const DETACHED_PROCESS: DWORD = 0x00000008; pub const CREATE_NEW_PROCESS_GROUP: DWORD = 0x00000200; @@ -285,6 +294,8 @@ pub fn nt_success(status: NTSTATUS) -> bool { status >= 0 } +// "RNG\0" +pub const BCRYPT_RNG_ALGORITHM: &[u16] = &[b'R' as u16, b'N' as u16, b'G' as u16, 0]; pub const BCRYPT_USE_SYSTEM_PREFERRED_RNG: DWORD = 0x00000002; #[repr(C)] @@ -455,6 +466,12 @@ pub enum FILE_INFO_BY_HANDLE_CLASS { } #[repr(C)] +pub struct FILE_ATTRIBUTE_TAG_INFO { + pub FileAttributes: DWORD, + pub ReparseTag: DWORD, +} + +#[repr(C)] pub struct FILE_DISPOSITION_INFO { pub DeleteFile: BOOLEAN, } @@ -501,6 +518,8 @@ pub struct FILE_END_OF_FILE_INFO { pub EndOfFile: LARGE_INTEGER, } +/// NB: Use carefully! In general using this as a reference is likely to get the +/// provenance wrong for the `rest` field! #[repr(C)] pub struct REPARSE_DATA_BUFFER { pub ReparseTag: c_uint, @@ -509,6 +528,8 @@ pub struct REPARSE_DATA_BUFFER { pub rest: (), } +/// NB: Use carefully! In general using this as a reference is likely to get the +/// provenance wrong for the `PathBuffer` field! #[repr(C)] pub struct SYMBOLIC_LINK_REPARSE_BUFFER { pub SubstituteNameOffset: c_ushort, @@ -519,6 +540,14 @@ pub struct SYMBOLIC_LINK_REPARSE_BUFFER { pub PathBuffer: WCHAR, } +/// NB: Use carefully! In general using this as a reference is likely to get the +/// provenance wrong for the `PathBuffer` field! +#[repr(C)] +pub struct FILE_NAME_INFO { + pub FileNameLength: DWORD, + pub FileName: [WCHAR; 1], +} + #[repr(C)] pub struct MOUNT_POINT_REPARSE_BUFFER { pub SubstituteNameOffset: c_ushort, @@ -550,6 +579,10 @@ pub struct CONDITION_VARIABLE { pub struct SRWLOCK { pub ptr: LPVOID, } +#[repr(C)] +pub struct INIT_ONCE { + pub ptr: LPVOID, +} #[repr(C)] pub struct REPARSE_MOUNTPOINT_DATA_BUFFER { @@ -802,10 +835,6 @@ if #[cfg(not(target_vendor = "uwp"))] { #[link(name = "advapi32")] extern "system" { - // Forbidden when targeting UWP - #[link_name = "SystemFunction036"] - pub fn RtlGenRandom(RandomBuffer: *mut u8, RandomBufferLength: ULONG) -> BOOLEAN; - // Allowed but unused by UWP pub fn OpenProcessToken( ProcessHandle: HANDLE, @@ -944,6 +973,7 @@ extern "system" { pub fn TlsAlloc() -> DWORD; pub fn TlsGetValue(dwTlsIndex: DWORD) -> LPVOID; pub fn TlsSetValue(dwTlsIndex: DWORD, lpTlsvalue: LPVOID) -> BOOL; + pub fn TlsFree(dwTlsIndex: DWORD) -> BOOL; pub fn GetLastError() -> DWORD; pub fn QueryPerformanceFrequency(lpFrequency: *mut LARGE_INTEGER) -> BOOL; pub fn QueryPerformanceCounter(lpPerformanceCount: *mut LARGE_INTEGER) -> BOOL; @@ -1086,6 +1116,7 @@ extern "system" { lpFileInformation: LPVOID, dwBufferSize: DWORD, ) -> BOOL; + pub fn GetFileType(hfile: HANDLE) -> DWORD; pub fn SleepConditionVariableSRW( ConditionVariable: PCONDITION_VARIABLE, SRWLock: PSRWLOCK, @@ -1103,6 +1134,14 @@ extern "system" { pub fn TryAcquireSRWLockExclusive(SRWLock: PSRWLOCK) -> BOOLEAN; pub fn TryAcquireSRWLockShared(SRWLock: PSRWLOCK) -> BOOLEAN; + pub fn InitOnceBeginInitialize( + lpInitOnce: LPINIT_ONCE, + dwFlags: DWORD, + fPending: LPBOOL, + lpContext: *mut LPVOID, + ) -> BOOL; + pub fn InitOnceComplete(lpInitOnce: LPINIT_ONCE, dwFlags: DWORD, lpContext: LPVOID) -> BOOL; + pub fn CompareStringOrdinal( lpString1: LPCWSTR, cchCount1: c_int, @@ -1217,11 +1256,18 @@ extern "system" { // >= Vista / Server 2008 // https://docs.microsoft.com/en-us/windows/win32/api/bcrypt/nf-bcrypt-bcryptgenrandom pub fn BCryptGenRandom( - hAlgorithm: LPVOID, + hAlgorithm: BCRYPT_ALG_HANDLE, pBuffer: *mut u8, cbBuffer: ULONG, dwFlags: ULONG, ) -> NTSTATUS; + pub fn BCryptOpenAlgorithmProvider( + phalgorithm: *mut BCRYPT_ALG_HANDLE, + pszAlgId: LPCWSTR, + pszimplementation: LPCWSTR, + dwflags: ULONG, + ) -> NTSTATUS; + pub fn BCryptCloseAlgorithmProvider(hAlgorithm: BCRYPT_ALG_HANDLE, dwFlags: ULONG) -> NTSTATUS; } // Functions that aren't available on every version of Windows that we support, @@ -1251,17 +1297,14 @@ compat_fn_with_fallback! { } compat_fn_optional! { - pub static SYNCH_API: &CStr = ansi_str!("api-ms-win-core-synch-l1-2-0"); - - // >= Windows 8 / Server 2012 - // https://docs.microsoft.com/en-us/windows/win32/api/synchapi/nf-synchapi-waitonaddress + crate::sys::compat::load_synch_functions(); pub fn WaitOnAddress( Address: LPVOID, CompareAddress: LPVOID, AddressSize: SIZE_T, dwMilliseconds: DWORD - ) -> BOOL; - pub fn WakeByAddressSingle(Address: LPVOID) -> (); + ); + pub fn WakeByAddressSingle(Address: LPVOID); } compat_fn_with_fallback! { diff --git a/library/std/src/sys/windows/cmath.rs b/library/std/src/sys/windows/cmath.rs index 1a5421fac..43ab8c7ee 100644 --- a/library/std/src/sys/windows/cmath.rs +++ b/library/std/src/sys/windows/cmath.rs @@ -44,7 +44,7 @@ mod shims { } // On 32-bit x86 MSVC these functions aren't defined, so we just define shims -// which promote everything fo f64, perform the calculation, and then demote +// which promote everything to f64, perform the calculation, and then demote // back to f32. While not precisely correct should be "correct enough" for now. #[cfg(all(target_env = "msvc", target_arch = "x86"))] mod shims { diff --git a/library/std/src/sys/windows/compat.rs b/library/std/src/sys/windows/compat.rs index ccc90177a..7dff81ecb 100644 --- a/library/std/src/sys/windows/compat.rs +++ b/library/std/src/sys/windows/compat.rs @@ -7,52 +7,66 @@ //! `GetModuleHandle` and `GetProcAddress` to look up DLL entry points at //! runtime. //! -//! This implementation uses a static initializer to look up the DLL entry -//! points. The CRT (C runtime) executes static initializers before `main` -//! is called (for binaries) and before `DllMain` is called (for DLLs). -//! This is the ideal time to look up DLL imports, because we are guaranteed -//! that no other threads will attempt to call these entry points. Thus, -//! we can look up the imports and store them in `static mut` fields -//! without any synchronization. +//! This is implemented simply by storing a function pointer in an atomic. +//! Loading and calling this function will have little or no overhead +//! compared with calling any other dynamically imported function. //! -//! This has an additional advantage: Because the DLL import lookup happens -//! at module initialization, the cost of these lookups is deterministic, -//! and is removed from the code paths that actually call the DLL imports. -//! That is, there is no unpredictable "cache miss" that occurs when calling -//! a DLL import. For applications that benefit from predictable delays, -//! this is a benefit. This also eliminates the comparison-and-branch -//! from the hot path. -//! -//! Currently, the standard library uses only a small number of dynamic -//! DLL imports. If this number grows substantially, then the cost of -//! performing all of the lookups at initialization time might become -//! substantial. -//! -//! The mechanism of registering a static initializer with the CRT is -//! documented in -//! [CRT Initialization](https://docs.microsoft.com/en-us/cpp/c-runtime-library/crt-initialization?view=msvc-160). -//! It works by contributing a global symbol to the `.CRT$XCU` section. -//! The linker builds a table of all static initializer functions. -//! The CRT startup code then iterates that table, calling each -//! initializer function. -//! -//! # **WARNING!!* -//! The environment that a static initializer function runs in is highly -//! constrained. There are **many** restrictions on what static initializers -//! can safely do. Static initializer functions **MUST NOT** do any of the -//! following (this list is not comprehensive): -//! * touch any other static field that is used by a different static -//! initializer, because the order that static initializers run in -//! is not defined. -//! * call `LoadLibrary` or any other function that acquires the DLL -//! loader lock. -//! * call any Rust function or CRT function that touches any static -//! (global) state. +//! The stored function pointer starts out as an importer function which will +//! swap itself with the real function when it's called for the first time. If +//! the real function can't be imported then a fallback function is used in its +//! place. While this is low cost for the happy path (where the function is +//! already loaded) it does mean there's some overhead the first time the +//! function is called. In the worst case, multiple threads may all end up +//! importing the same function unnecessarily. use crate::ffi::{c_void, CStr}; use crate::ptr::NonNull; +use crate::sync::atomic::Ordering; use crate::sys::c; +// This uses a static initializer to preload some imported functions. +// The CRT (C runtime) executes static initializers before `main` +// is called (for binaries) and before `DllMain` is called (for DLLs). +// +// It works by contributing a global symbol to the `.CRT$XCT` section. +// The linker builds a table of all static initializer functions. +// The CRT startup code then iterates that table, calling each +// initializer function. +// +// NOTE: User code should instead use .CRT$XCU to reliably run after std's initializer. +// If you're reading this and would like a guarantee here, please +// file an issue for discussion; currently we don't guarantee any functionality +// before main. +// See https://docs.microsoft.com/en-us/cpp/c-runtime-library/crt-initialization?view=msvc-170 +#[used] +#[link_section = ".CRT$XCT"] +static INIT_TABLE_ENTRY: unsafe extern "C" fn() = init; + +/// Preload some imported functions. +/// +/// Note that any functions included here will be unconditionally loaded in +/// the final binary, regardless of whether or not they're actually used. +/// +/// Therefore, this should be limited to `compat_fn_optional` functions which +/// must be preloaded or any functions where lazier loading demonstrates a +/// negative performance impact in practical situations. +/// +/// Currently we only preload `WaitOnAddress` and `WakeByAddressSingle`. +unsafe extern "C" fn init() { + // In an exe this code is executed before main() so is single threaded. + // In a DLL the system's loader lock will be held thereby synchronizing + // access. So the same best practices apply here as they do to running in DllMain: + // https://docs.microsoft.com/en-us/windows/win32/dlls/dynamic-link-library-best-practices + // + // DO NOT do anything interesting or complicated in this function! DO NOT call + // any Rust functions or CRT functions if those functions touch any global state, + // because this function runs during global initialization. For example, DO NOT + // do any dynamic allocation, don't call LoadLibrary, etc. + + // Attempt to preload the synch functions. + load_synch_functions(); +} + /// Helper macro for creating CStrs from literals and symbol names. macro_rules! ansi_str { (sym $ident:ident) => {{ @@ -85,39 +99,6 @@ pub(crate) const fn const_cstr_from_bytes(bytes: &'static [u8]) -> &'static CStr unsafe { crate::ffi::CStr::from_bytes_with_nul_unchecked(bytes) } } -#[used] -#[link_section = ".CRT$XCU"] -static INIT_TABLE_ENTRY: unsafe extern "C" fn() = init; - -/// This is where the magic preloading of symbols happens. -/// -/// Note that any functions included here will be unconditionally included in -/// the final binary, regardless of whether or not they're actually used. -/// -/// Therefore, this is limited to `compat_fn_optional` functions which must be -/// preloaded and any functions which may be more time sensitive, even for the first call. -unsafe extern "C" fn init() { - // There is no locking here. This code is executed before main() is entered, and - // is guaranteed to be single-threaded. - // - // DO NOT do anything interesting or complicated in this function! DO NOT call - // any Rust functions or CRT functions if those functions touch any global state, - // because this function runs during global initialization. For example, DO NOT - // do any dynamic allocation, don't call LoadLibrary, etc. - - if let Some(synch) = Module::new(c::SYNCH_API) { - // These are optional and so we must manually attempt to load them - // before they can be used. - c::WaitOnAddress::preload(synch); - c::WakeByAddressSingle::preload(synch); - } - - if let Some(kernel32) = Module::new(c::KERNEL32) { - // Preloading this means getting a precise time will be as fast as possible. - c::GetSystemTimePreciseAsFileTime::preload(kernel32); - } -} - /// Represents a loaded module. /// /// Note that the modules std depends on must not be unloaded. @@ -151,7 +132,7 @@ impl Module { macro_rules! compat_fn_with_fallback { (pub static $module:ident: &CStr = $name:expr; $( $(#[$meta:meta])* - pub fn $symbol:ident($($argname:ident: $argtype:ty),*) -> $rettype:ty $fallback_body:block + $vis:vis fn $symbol:ident($($argname:ident: $argtype:ty),*) -> $rettype:ty $fallback_body:block )*) => ( pub static $module: &CStr = $name; $( @@ -196,11 +177,6 @@ macro_rules! compat_fn_with_fallback { $fallback_body } - #[allow(unused)] - pub(in crate::sys) fn preload(module: Module) { - load_from_module(Some(module)); - } - #[inline(always)] pub unsafe fn call($($argname: $argtype),*) -> $rettype { let func: F = mem::transmute(PTR.load(Ordering::Relaxed)); @@ -208,66 +184,60 @@ macro_rules! compat_fn_with_fallback { } } $(#[$meta])* - pub use $symbol::call as $symbol; + $vis use $symbol::call as $symbol; )*) } -/// A function that either exists or doesn't. +/// Optionally loaded functions. /// -/// NOTE: Optional functions must be preloaded in the `init` function above, or they will always be None. +/// Actual loading of the function defers to $load_functions. macro_rules! compat_fn_optional { - (pub static $module:ident: &CStr = $name:expr; $( - $(#[$meta:meta])* - pub fn $symbol:ident($($argname:ident: $argtype:ty),*) -> $rettype:ty; - )*) => ( - pub static $module: &CStr = $name; + ($load_functions:expr; $( - $(#[$meta])* - pub mod $symbol { - #[allow(unused_imports)] - use super::*; - use crate::mem; - use crate::sync::atomic::{AtomicPtr, Ordering}; - use crate::sys::compat::Module; - use crate::ptr::{self, NonNull}; - - type F = unsafe extern "system" fn($($argtype),*) -> $rettype; - - /// `PTR` will either be `null()` or set to the loaded function. - static PTR: AtomicPtr<c_void> = AtomicPtr::new(ptr::null_mut()); - - /// Only allow access to the function if it has loaded successfully. - #[inline(always)] - #[cfg(not(miri))] - pub fn option() -> Option<F> { - unsafe { - NonNull::new(PTR.load(Ordering::Relaxed)).map(|f| mem::transmute(f)) + $(#[$meta:meta])* + $vis:vis fn $symbol:ident($($argname:ident: $argtype:ty),*) $(-> $rettype:ty)?; + )+) => ( + $( + pub mod $symbol { + use super::*; + use crate::ffi::c_void; + use crate::mem; + use crate::ptr::{self, NonNull}; + use crate::sync::atomic::{AtomicPtr, Ordering}; + + pub(in crate::sys) static PTR: AtomicPtr<c_void> = AtomicPtr::new(ptr::null_mut()); + + type F = unsafe extern "system" fn($($argtype),*) $(-> $rettype)?; + + #[inline(always)] + pub fn option() -> Option<F> { + // Miri does not understand the way we do preloading + // therefore load the function here instead. + #[cfg(miri)] $load_functions; + NonNull::new(PTR.load(Ordering::Relaxed)).map(|f| unsafe { mem::transmute(f) }) } } + )+ + ) +} - // Miri does not understand the way we do preloading - // therefore load the function here instead. - #[cfg(miri)] - pub fn option() -> Option<F> { - let mut func = NonNull::new(PTR.load(Ordering::Relaxed)); - if func.is_none() { - unsafe { Module::new($module).map(preload) }; - func = NonNull::new(PTR.load(Ordering::Relaxed)); - } - unsafe { - func.map(|f| mem::transmute(f)) - } - } +/// Load all needed functions from "api-ms-win-core-synch-l1-2-0". +pub(super) fn load_synch_functions() { + fn try_load() -> Option<()> { + const MODULE_NAME: &CStr = ansi_str!("api-ms-win-core-synch-l1-2-0"); + const WAIT_ON_ADDRESS: &CStr = ansi_str!("WaitOnAddress"); + const WAKE_BY_ADDRESS_SINGLE: &CStr = ansi_str!("WakeByAddressSingle"); + + // Try loading the library and all the required functions. + // If any step fails, then they all fail. + let library = unsafe { Module::new(MODULE_NAME) }?; + let wait_on_address = library.proc_address(WAIT_ON_ADDRESS)?; + let wake_by_address_single = library.proc_address(WAKE_BY_ADDRESS_SINGLE)?; + + c::WaitOnAddress::PTR.store(wait_on_address.as_ptr(), Ordering::Relaxed); + c::WakeByAddressSingle::PTR.store(wake_by_address_single.as_ptr(), Ordering::Relaxed); + Some(()) + } - #[allow(unused)] - pub(in crate::sys) fn preload(module: Module) { - unsafe { - static SYMBOL_NAME: &CStr = ansi_str!(sym $symbol); - if let Some(f) = module.proc_address(SYMBOL_NAME) { - PTR.store(f.as_ptr(), Ordering::Relaxed); - } - } - } - } - )*) + try_load(); } diff --git a/library/std/src/sys/windows/fs.rs b/library/std/src/sys/windows/fs.rs index aed082b3e..378098038 100644 --- a/library/std/src/sys/windows/fs.rs +++ b/library/std/src/sys/windows/fs.rs @@ -1,9 +1,10 @@ use crate::os::windows::prelude::*; +use crate::borrow::Cow; use crate::ffi::OsString; use crate::fmt; -use crate::io::{self, Error, IoSlice, IoSliceMut, ReadBuf, SeekFrom}; -use crate::mem; +use crate::io::{self, BorrowedCursor, Error, IoSlice, IoSliceMut, SeekFrom}; +use crate::mem::{self, MaybeUninit}; use crate::os::windows::io::{AsHandle, BorrowedHandle}; use crate::path::{Path, PathBuf}; use crate::ptr; @@ -11,7 +12,7 @@ use crate::slice; use crate::sync::Arc; use crate::sys::handle::Handle; use crate::sys::time::SystemTime; -use crate::sys::{c, cvt}; +use crate::sys::{c, cvt, Align8}; use crate::sys_common::{AsInner, FromInner, IntoInner}; use crate::thread; @@ -326,9 +327,15 @@ impl File { cvt(c::GetFileInformationByHandle(self.handle.as_raw_handle(), &mut info))?; let mut reparse_tag = 0; if info.dwFileAttributes & c::FILE_ATTRIBUTE_REPARSE_POINT != 0 { - let mut b = [0; c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE]; - if let Ok((_, buf)) = self.reparse_point(&mut b) { - reparse_tag = buf.ReparseTag; + let mut attr_tag: c::FILE_ATTRIBUTE_TAG_INFO = mem::zeroed(); + cvt(c::GetFileInformationByHandleEx( + self.handle.as_raw_handle(), + c::FileAttributeTagInfo, + ptr::addr_of_mut!(attr_tag).cast(), + mem::size_of::<c::FILE_ATTRIBUTE_TAG_INFO>().try_into().unwrap(), + ))?; + if attr_tag.FileAttributes & c::FILE_ATTRIBUTE_REPARSE_POINT != 0 { + reparse_tag = attr_tag.ReparseTag; } } Ok(FileAttr { @@ -389,9 +396,15 @@ impl File { attr.file_size = info.AllocationSize as u64; attr.number_of_links = Some(info.NumberOfLinks); if attr.file_type().is_reparse_point() { - let mut b = [0; c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE]; - if let Ok((_, buf)) = self.reparse_point(&mut b) { - attr.reparse_tag = buf.ReparseTag; + let mut attr_tag: c::FILE_ATTRIBUTE_TAG_INFO = mem::zeroed(); + cvt(c::GetFileInformationByHandleEx( + self.handle.as_raw_handle(), + c::FileAttributeTagInfo, + ptr::addr_of_mut!(attr_tag).cast(), + mem::size_of::<c::FILE_ATTRIBUTE_TAG_INFO>().try_into().unwrap(), + ))?; + if attr_tag.FileAttributes & c::FILE_ATTRIBUTE_REPARSE_POINT != 0 { + attr.reparse_tag = attr_tag.ReparseTag; } } Ok(attr) @@ -415,8 +428,8 @@ impl File { self.handle.read_at(buf, offset) } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - self.handle.read_buf(buf) + pub fn read_buf(&self, cursor: BorrowedCursor<'_>) -> io::Result<()> { + self.handle.read_buf(cursor) } pub fn write(&self, buf: &[u8]) -> io::Result<usize> { @@ -458,38 +471,46 @@ impl File { Ok(Self { handle: self.handle.try_clone()? }) } - fn reparse_point<'a>( + // NB: returned pointer is derived from `space`, and has provenance to + // match. A raw pointer is returned rather than a reference in order to + // avoid narrowing provenance to the actual `REPARSE_DATA_BUFFER`. + fn reparse_point( &self, - space: &'a mut [u8; c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE], - ) -> io::Result<(c::DWORD, &'a c::REPARSE_DATA_BUFFER)> { + space: &mut Align8<[MaybeUninit<u8>]>, + ) -> io::Result<(c::DWORD, *const c::REPARSE_DATA_BUFFER)> { unsafe { let mut bytes = 0; cvt({ + // Grab this in advance to avoid it invalidating the pointer + // we get from `space.0.as_mut_ptr()`. + let len = space.0.len(); c::DeviceIoControl( self.handle.as_raw_handle(), c::FSCTL_GET_REPARSE_POINT, ptr::null_mut(), 0, - space.as_mut_ptr() as *mut _, - space.len() as c::DWORD, + space.0.as_mut_ptr().cast(), + len as c::DWORD, &mut bytes, ptr::null_mut(), ) })?; - Ok((bytes, &*(space.as_ptr() as *const c::REPARSE_DATA_BUFFER))) + const _: () = assert!(core::mem::align_of::<c::REPARSE_DATA_BUFFER>() <= 8); + Ok((bytes, space.0.as_ptr().cast::<c::REPARSE_DATA_BUFFER>())) } } fn readlink(&self) -> io::Result<PathBuf> { - let mut space = [0u8; c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE]; + let mut space = Align8([MaybeUninit::<u8>::uninit(); c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE]); let (_bytes, buf) = self.reparse_point(&mut space)?; unsafe { - let (path_buffer, subst_off, subst_len, relative) = match buf.ReparseTag { + let (path_buffer, subst_off, subst_len, relative) = match (*buf).ReparseTag { c::IO_REPARSE_TAG_SYMLINK => { let info: *const c::SYMBOLIC_LINK_REPARSE_BUFFER = - &buf.rest as *const _ as *const _; + ptr::addr_of!((*buf).rest).cast(); + assert!(info.is_aligned()); ( - &(*info).PathBuffer as *const _ as *const u16, + ptr::addr_of!((*info).PathBuffer).cast::<u16>(), (*info).SubstituteNameOffset / 2, (*info).SubstituteNameLength / 2, (*info).Flags & c::SYMLINK_FLAG_RELATIVE != 0, @@ -497,9 +518,10 @@ impl File { } c::IO_REPARSE_TAG_MOUNT_POINT => { let info: *const c::MOUNT_POINT_REPARSE_BUFFER = - &buf.rest as *const _ as *const _; + ptr::addr_of!((*buf).rest).cast(); + assert!(info.is_aligned()); ( - &(*info).PathBuffer as *const _ as *const u16, + ptr::addr_of!((*info).PathBuffer).cast::<u16>(), (*info).SubstituteNameOffset / 2, (*info).SubstituteNameLength / 2, false, @@ -512,7 +534,7 @@ impl File { )); } }; - let subst_ptr = path_buffer.offset(subst_off as isize); + let subst_ptr = path_buffer.add(subst_off.into()); let mut subst = slice::from_raw_parts(subst_ptr, subst_len as usize); // Absolute paths start with an NT internal namespace prefix `\??\` // We should not let it leak through. @@ -551,6 +573,14 @@ impl File { "Cannot set file timestamp to 0", )); } + let is_max = + |t: c::FILETIME| t.dwLowDateTime == c::DWORD::MAX && t.dwHighDateTime == c::DWORD::MAX; + if times.accessed.map_or(false, is_max) || times.modified.map_or(false, is_max) { + return Err(io::const_io_error!( + io::ErrorKind::InvalidInput, + "Cannot set file timestamp to 0xFFFF_FFFF_FFFF_FFFF", + )); + } cvt(unsafe { c::SetFileTime(self.as_handle(), None, times.accessed.as_ref(), times.modified.as_ref()) })?; @@ -649,27 +679,31 @@ impl File { /// A buffer for holding directory entries. struct DirBuff { - buffer: Vec<u8>, + buffer: Box<Align8<[MaybeUninit<u8>; Self::BUFFER_SIZE]>>, } impl DirBuff { + const BUFFER_SIZE: usize = 1024; fn new() -> Self { - const BUFFER_SIZE: usize = 1024; - Self { buffer: vec![0_u8; BUFFER_SIZE] } + Self { + // Safety: `Align8<[MaybeUninit<u8>; N]>` does not need + // initialization. + buffer: unsafe { Box::new_uninit().assume_init() }, + } } fn capacity(&self) -> usize { - self.buffer.len() + self.buffer.0.len() } fn as_mut_ptr(&mut self) -> *mut u8 { - self.buffer.as_mut_ptr().cast() + self.buffer.0.as_mut_ptr().cast() } /// Returns a `DirBuffIter`. fn iter(&self) -> DirBuffIter<'_> { DirBuffIter::new(self) } } -impl AsRef<[u8]> for DirBuff { - fn as_ref(&self) -> &[u8] { - &self.buffer +impl AsRef<[MaybeUninit<u8>]> for DirBuff { + fn as_ref(&self) -> &[MaybeUninit<u8>] { + &self.buffer.0 } } @@ -677,7 +711,7 @@ impl AsRef<[u8]> for DirBuff { /// /// Currently only returns file names (UTF-16 encoded). struct DirBuffIter<'a> { - buffer: Option<&'a [u8]>, + buffer: Option<&'a [MaybeUninit<u8>]>, cursor: usize, } impl<'a> DirBuffIter<'a> { @@ -686,23 +720,34 @@ impl<'a> DirBuffIter<'a> { } } impl<'a> Iterator for DirBuffIter<'a> { - type Item = (&'a [u16], bool); + type Item = (Cow<'a, [u16]>, bool); fn next(&mut self) -> Option<Self::Item> { use crate::mem::size_of; let buffer = &self.buffer?[self.cursor..]; // Get the name and next entry from the buffer. - // SAFETY: The buffer contains a `FILE_ID_BOTH_DIR_INFO` struct but the - // last field (the file name) is unsized. So an offset has to be - // used to get the file name slice. + // SAFETY: + // - The buffer contains a `FILE_ID_BOTH_DIR_INFO` struct but the last + // field (the file name) is unsized. So an offset has to be used to + // get the file name slice. + // - The OS has guaranteed initialization of the fields of + // `FILE_ID_BOTH_DIR_INFO` and the trailing filename (for at least + // `FileNameLength` bytes) let (name, is_directory, next_entry) = unsafe { let info = buffer.as_ptr().cast::<c::FILE_ID_BOTH_DIR_INFO>(); - let next_entry = (*info).NextEntryOffset as usize; - let name = crate::slice::from_raw_parts( - (*info).FileName.as_ptr().cast::<u16>(), - (*info).FileNameLength as usize / size_of::<u16>(), + // While this is guaranteed to be aligned in documentation for + // https://docs.microsoft.com/en-us/windows/win32/api/winbase/ns-winbase-file_id_both_dir_info + // it does not seem that reality is so kind, and assuming this + // caused crashes in some cases (https://github.com/rust-lang/rust/issues/104530) + // presumably, this can be blamed on buggy filesystem drivers, but who knows. + let next_entry = ptr::addr_of!((*info).NextEntryOffset).read_unaligned() as usize; + let length = ptr::addr_of!((*info).FileNameLength).read_unaligned() as usize; + let attrs = ptr::addr_of!((*info).FileAttributes).read_unaligned(); + let name = from_maybe_unaligned( + ptr::addr_of!((*info).FileName).cast::<u16>(), + length / size_of::<u16>(), ); - let is_directory = ((*info).FileAttributes & c::FILE_ATTRIBUTE_DIRECTORY) != 0; + let is_directory = (attrs & c::FILE_ATTRIBUTE_DIRECTORY) != 0; (name, is_directory, next_entry) }; @@ -715,13 +760,21 @@ impl<'a> Iterator for DirBuffIter<'a> { // Skip `.` and `..` pseudo entries. const DOT: u16 = b'.' as u16; - match name { + match &name[..] { [DOT] | [DOT, DOT] => self.next(), _ => Some((name, is_directory)), } } } +unsafe fn from_maybe_unaligned<'a>(p: *const u16, len: usize) -> Cow<'a, [u16]> { + if p.is_aligned() { + Cow::Borrowed(crate::slice::from_raw_parts(p, len)) + } else { + Cow::Owned((0..len).map(|i| p.add(i).read_unaligned()).collect()) + } +} + /// Open a link relative to the parent directory, ensure no symlinks are followed. fn open_link_no_reparse(parent: &File, name: &[u16], access: u32) -> io::Result<File> { // This is implemented using the lower level `NtCreateFile` function as @@ -1077,13 +1130,13 @@ fn remove_dir_all_iterative(f: &File, delete: fn(&File) -> io::Result<()>) -> io if is_directory { let child_dir = open_link_no_reparse( &dir, - name, + &name, c::SYNCHRONIZE | c::DELETE | c::FILE_LIST_DIRECTORY, )?; dirlist.push(child_dir); } else { for i in 1..=MAX_RETRIES { - let result = open_link_no_reparse(&dir, name, c::SYNCHRONIZE | c::DELETE); + let result = open_link_no_reparse(&dir, &name, c::SYNCHRONIZE | c::DELETE); match result { Ok(f) => delete(&f)?, // Already deleted, so skip. @@ -1337,18 +1390,19 @@ fn symlink_junction_inner(original: &Path, junction: &Path) -> io::Result<()> { let h = f.as_inner().as_raw_handle(); unsafe { - let mut data = [0u8; c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE]; - let db = data.as_mut_ptr() as *mut c::REPARSE_MOUNTPOINT_DATA_BUFFER; - let buf = &mut (*db).ReparseTarget as *mut c::WCHAR; + let mut data = Align8([MaybeUninit::<u8>::uninit(); c::MAXIMUM_REPARSE_DATA_BUFFER_SIZE]); + let data_ptr = data.0.as_mut_ptr(); + let db = data_ptr.cast::<c::REPARSE_MOUNTPOINT_DATA_BUFFER>(); + let buf = ptr::addr_of_mut!((*db).ReparseTarget).cast::<c::WCHAR>(); let mut i = 0; // FIXME: this conversion is very hacky let v = br"\??\"; let v = v.iter().map(|x| *x as u16); for c in v.chain(original.as_os_str().encode_wide()) { - *buf.offset(i) = c; + *buf.add(i) = c; i += 1; } - *buf.offset(i) = 0; + *buf.add(i) = 0; i += 1; (*db).ReparseTag = c::IO_REPARSE_TAG_MOUNT_POINT; (*db).ReparseTargetMaximumLength = (i * 2) as c::WORD; @@ -1359,7 +1413,7 @@ fn symlink_junction_inner(original: &Path, junction: &Path) -> io::Result<()> { cvt(c::DeviceIoControl( h as *mut _, c::FSCTL_SET_REPARSE_POINT, - data.as_ptr() as *mut _, + data_ptr.cast(), (*db).ReparseDataLength + 8, ptr::null_mut(), 0, diff --git a/library/std/src/sys/windows/handle.rs b/library/std/src/sys/windows/handle.rs index e24b09cc9..ae33d48c6 100644 --- a/library/std/src/sys/windows/handle.rs +++ b/library/std/src/sys/windows/handle.rs @@ -4,7 +4,7 @@ mod tests; use crate::cmp; -use crate::io::{self, ErrorKind, IoSlice, IoSliceMut, Read, ReadBuf}; +use crate::io::{self, BorrowedCursor, ErrorKind, IoSlice, IoSliceMut, Read}; use crate::mem; use crate::os::windows::io::{ AsHandle, AsRawHandle, BorrowedHandle, FromRawHandle, IntoRawHandle, OwnedHandle, RawHandle, @@ -112,18 +112,16 @@ impl Handle { } } - pub fn read_buf(&self, buf: &mut ReadBuf<'_>) -> io::Result<()> { - let res = unsafe { - self.synchronous_read(buf.unfilled_mut().as_mut_ptr(), buf.remaining(), None) - }; + pub fn read_buf(&self, mut cursor: BorrowedCursor<'_>) -> io::Result<()> { + let res = + unsafe { self.synchronous_read(cursor.as_mut().as_mut_ptr(), cursor.capacity(), None) }; match res { Ok(read) => { // Safety: `read` bytes were written to the initialized portion of the buffer unsafe { - buf.assume_init(read as usize); + cursor.advance(read as usize); } - buf.add_filled(read as usize); Ok(()) } diff --git a/library/std/src/sys/windows/io.rs b/library/std/src/sys/windows/io.rs index fb06df1f8..2cc34c986 100644 --- a/library/std/src/sys/windows/io.rs +++ b/library/std/src/sys/windows/io.rs @@ -1,6 +1,10 @@ use crate::marker::PhantomData; +use crate::mem::size_of; +use crate::os::windows::io::{AsHandle, AsRawHandle, BorrowedHandle}; use crate::slice; -use crate::sys::c; +use crate::sys::{c, Align8}; +use core; +use libc; #[derive(Copy, Clone)] #[repr(transparent)] @@ -78,3 +82,73 @@ impl<'a> IoSliceMut<'a> { unsafe { slice::from_raw_parts_mut(self.vec.buf as *mut u8, self.vec.len as usize) } } } + +pub fn is_terminal(h: &impl AsHandle) -> bool { + unsafe { handle_is_console(h.as_handle()) } +} + +unsafe fn handle_is_console(handle: BorrowedHandle<'_>) -> bool { + let handle = handle.as_raw_handle(); + + // A null handle means the process has no console. + if handle.is_null() { + return false; + } + + let mut out = 0; + if c::GetConsoleMode(handle, &mut out) != 0 { + // False positives aren't possible. If we got a console then we definitely have a console. + return true; + } + + // At this point, we *could* have a false negative. We can determine that this is a true + // negative if we can detect the presence of a console on any of the standard I/O streams. If + // another stream has a console, then we know we're in a Windows console and can therefore + // trust the negative. + for std_handle in [c::STD_INPUT_HANDLE, c::STD_OUTPUT_HANDLE, c::STD_ERROR_HANDLE] { + let std_handle = c::GetStdHandle(std_handle); + if !std_handle.is_null() + && std_handle != handle + && c::GetConsoleMode(std_handle, &mut out) != 0 + { + return false; + } + } + + // Otherwise, we fall back to an msys hack to see if we can detect the presence of a pty. + msys_tty_on(handle) +} + +unsafe fn msys_tty_on(handle: c::HANDLE) -> bool { + // Early return if the handle is not a pipe. + if c::GetFileType(handle) != c::FILE_TYPE_PIPE { + return false; + } + + const SIZE: usize = size_of::<c::FILE_NAME_INFO>() + c::MAX_PATH * size_of::<c::WCHAR>(); + let mut name_info_bytes = Align8([0u8; SIZE]); + let res = c::GetFileInformationByHandleEx( + handle, + c::FileNameInfo, + name_info_bytes.0.as_mut_ptr() as *mut libc::c_void, + SIZE as u32, + ); + if res == 0 { + return false; + } + let name_info: &c::FILE_NAME_INFO = &*(name_info_bytes.0.as_ptr() as *const c::FILE_NAME_INFO); + let name_len = name_info.FileNameLength as usize / 2; + // Offset to get the `FileName` field. + let name_ptr = name_info_bytes.0.as_ptr().offset(size_of::<c::DWORD>() as isize).cast::<u16>(); + let s = core::slice::from_raw_parts(name_ptr, name_len); + let name = String::from_utf16_lossy(s); + // Get the file name only. + let name = name.rsplit('\\').next().unwrap_or(&name); + // This checks whether 'pty' exists in the file name, which indicates that + // a pseudo-terminal is attached. To mitigate against false positives + // (e.g., an actual file name that contains 'pty'), we also require that + // the file name begins with either the strings 'msys-' or 'cygwin-'.) + let is_msys = name.starts_with("msys-") || name.starts_with("cygwin-"); + let is_pty = name.contains("-pty"); + is_msys && is_pty +} diff --git a/library/std/src/sys/windows/locks/mod.rs b/library/std/src/sys/windows/locks/mod.rs index d412ff152..602a2d623 100644 --- a/library/std/src/sys/windows/locks/mod.rs +++ b/library/std/src/sys/windows/locks/mod.rs @@ -3,4 +3,4 @@ mod mutex; mod rwlock; pub use condvar::{Condvar, MovableCondvar}; pub use mutex::{MovableMutex, Mutex}; -pub use rwlock::{MovableRwLock, RwLock}; +pub use rwlock::MovableRwLock; diff --git a/library/std/src/sys/windows/locks/mutex.rs b/library/std/src/sys/windows/locks/mutex.rs index f91e8f9f5..91207f5f4 100644 --- a/library/std/src/sys/windows/locks/mutex.rs +++ b/library/std/src/sys/windows/locks/mutex.rs @@ -37,8 +37,6 @@ impl Mutex { pub const fn new() -> Mutex { Mutex { srwlock: UnsafeCell::new(c::SRWLOCK_INIT) } } - #[inline] - pub unsafe fn init(&mut self) {} #[inline] pub unsafe fn lock(&self) { diff --git a/library/std/src/sys/windows/mod.rs b/library/std/src/sys/windows/mod.rs index b3f6d2d0a..eab9b9612 100644 --- a/library/std/src/sys/windows/mod.rs +++ b/library/std/src/sys/windows/mod.rs @@ -2,6 +2,7 @@ use crate::ffi::{CStr, OsStr, OsString}; use crate::io::ErrorKind; +use crate::mem::MaybeUninit; use crate::os::windows::ffi::{OsStrExt, OsStringExt}; use crate::path::PathBuf; use crate::time::Duration; @@ -47,7 +48,7 @@ cfg_if::cfg_if! { // SAFETY: must be called only once during runtime initialization. // NOTE: this is not guaranteed to run, for example when Rust code is called externally. -pub unsafe fn init(_argc: isize, _argv: *const *const u8) { +pub unsafe fn init(_argc: isize, _argv: *const *const u8, _sigpipe: u8) { stack_overflow::init(); // Normally, `thread::spawn` will call `Thread::set_name` but since this thread already @@ -204,8 +205,8 @@ where // This initial size also works around `GetFullPathNameW` returning // incorrect size hints for some short paths: // https://github.com/dylni/normpath/issues/5 - let mut stack_buf = [0u16; 512]; - let mut heap_buf = Vec::new(); + let mut stack_buf: [MaybeUninit<u16>; 512] = MaybeUninit::uninit_array(); + let mut heap_buf: Vec<MaybeUninit<u16>> = Vec::new(); unsafe { let mut n = stack_buf.len(); loop { @@ -214,6 +215,11 @@ where } else { let extra = n - heap_buf.len(); heap_buf.reserve(extra); + // We used `reserve` and not `reserve_exact`, so in theory we + // may have gotten more than requested. If so, we'd like to use + // it... so long as we won't cause overflow. + n = heap_buf.capacity().min(c::DWORD::MAX as usize); + // Safety: MaybeUninit<u16> does not need initialization heap_buf.set_len(n); &mut heap_buf[..] }; @@ -228,13 +234,13 @@ where // error" is still 0 then we interpret it as a 0 length buffer and // not an actual error. c::SetLastError(0); - let k = match f1(buf.as_mut_ptr(), n as c::DWORD) { + let k = match f1(buf.as_mut_ptr().cast::<u16>(), n as c::DWORD) { 0 if c::GetLastError() == 0 => 0, 0 => return Err(crate::io::Error::last_os_error()), n => n, } as usize; if k == n && c::GetLastError() == c::ERROR_INSUFFICIENT_BUFFER { - n *= 2; + n = n.saturating_mul(2).min(c::DWORD::MAX as usize); } else if k > n { n = k; } else if k == n { @@ -244,7 +250,9 @@ where // Therefore k never equals n. unreachable!(); } else { - return Ok(f2(&buf[..k])); + // Safety: First `k` values are initialized. + let slice: &[u16] = MaybeUninit::slice_assume_init_ref(&buf[..k]); + return Ok(f2(slice)); } } } @@ -321,3 +329,11 @@ pub fn abort_internal() -> ! { } crate::intrinsics::abort(); } + +/// Align the inner value to 8 bytes. +/// +/// This is enough for almost all of the buffers we're likely to work with in +/// the Windows APIs we use. +#[repr(C, align(8))] +#[derive(Copy, Clone)] +pub(crate) struct Align8<T: ?Sized>(pub T); diff --git a/library/std/src/sys/windows/os.rs b/library/std/src/sys/windows/os.rs index bcac996c0..352337ba3 100644 --- a/library/std/src/sys/windows/os.rs +++ b/library/std/src/sys/windows/os.rs @@ -99,11 +99,11 @@ impl Iterator for Env { } let p = self.cur as *const u16; let mut len = 0; - while *p.offset(len) != 0 { + while *p.add(len) != 0 { len += 1; } - let s = slice::from_raw_parts(p, len as usize); - self.cur = self.cur.offset(len + 1); + let s = slice::from_raw_parts(p, len); + self.cur = self.cur.add(len + 1); // Windows allows environment variables to start with an equals // symbol (in any other position, this is the separator between diff --git a/library/std/src/sys/windows/os_str.rs b/library/std/src/sys/windows/os_str.rs index 11883f150..4bdd8c505 100644 --- a/library/std/src/sys/windows/os_str.rs +++ b/library/std/src/sys/windows/os_str.rs @@ -164,9 +164,7 @@ impl Slice { } pub fn to_owned(&self) -> Buf { - let mut buf = Wtf8Buf::with_capacity(self.inner.len()); - buf.push_wtf8(&self.inner); - Buf { inner: buf } + Buf { inner: self.inner.to_owned() } } pub fn clone_into(&self, buf: &mut Buf) { diff --git a/library/std/src/sys/windows/path/tests.rs b/library/std/src/sys/windows/path/tests.rs index 6eab38cab..623c62361 100644 --- a/library/std/src/sys/windows/path/tests.rs +++ b/library/std/src/sys/windows/path/tests.rs @@ -105,7 +105,7 @@ fn test_parse_prefix_verbatim_device() { assert_eq!(prefix, parse_prefix(r"\\?/C:\windows\system32\notepad.exe")); } -// See #93586 for more infomation. +// See #93586 for more information. #[test] fn test_windows_prefix_components() { use crate::path::Path; diff --git a/library/std/src/sys/windows/process.rs b/library/std/src/sys/windows/process.rs index 02d5af471..9cbb4ef19 100644 --- a/library/std/src/sys/windows/process.rs +++ b/library/std/src/sys/windows/process.rs @@ -16,6 +16,7 @@ use crate::os::windows::ffi::{OsStrExt, OsStringExt}; use crate::os::windows::io::{AsHandle, AsRawHandle, BorrowedHandle, FromRawHandle, IntoRawHandle}; use crate::path::{Path, PathBuf}; use crate::ptr; +use crate::sync::Mutex; use crate::sys::args::{self, Arg}; use crate::sys::c; use crate::sys::c::NonZeroDWORD; @@ -25,7 +26,6 @@ use crate::sys::handle::Handle; use crate::sys::path; use crate::sys::pipe::{self, AnonPipe}; use crate::sys::stdio; -use crate::sys_common::mutex::StaticMutex; use crate::sys_common::process::{CommandEnv, CommandEnvs}; use crate::sys_common::IntoInner; @@ -301,9 +301,9 @@ impl Command { // // For more information, msdn also has an article about this race: // https://support.microsoft.com/kb/315939 - static CREATE_PROCESS_LOCK: StaticMutex = StaticMutex::new(); + static CREATE_PROCESS_LOCK: Mutex<()> = Mutex::new(()); - let _guard = unsafe { CREATE_PROCESS_LOCK.lock() }; + let _guard = CREATE_PROCESS_LOCK.lock(); let mut pipes = StdioPipes { stdin: None, stdout: None, stderr: None }; let null = Stdio::Null; diff --git a/library/std/src/sys/windows/rand.rs b/library/std/src/sys/windows/rand.rs index f8fd93a73..b5a49489d 100644 --- a/library/std/src/sys/windows/rand.rs +++ b/library/std/src/sys/windows/rand.rs @@ -1,35 +1,106 @@ -use crate::io; +//! # Random key generation +//! +//! This module wraps the RNG provided by the OS. There are a few different +//! ways to interface with the OS RNG so it's worth exploring each of the options. +//! Note that at the time of writing these all go through the (undocumented) +//! `bcryptPrimitives.dll` but they use different route to get there. +//! +//! Originally we were using [`RtlGenRandom`], however that function is +//! deprecated and warns it "may be altered or unavailable in subsequent versions". +//! +//! So we switched to [`BCryptGenRandom`] with the `BCRYPT_USE_SYSTEM_PREFERRED_RNG` +//! flag to query and find the system configured RNG. However, this change caused a small +//! but significant number of users to experience panics caused by a failure of +//! this function. See [#94098]. +//! +//! The current version falls back to using `BCryptOpenAlgorithmProvider` if +//! `BCRYPT_USE_SYSTEM_PREFERRED_RNG` fails for any reason. +//! +//! [#94098]: https://github.com/rust-lang/rust/issues/94098 +//! [`RtlGenRandom`]: https://docs.microsoft.com/en-us/windows/win32/api/ntsecapi/nf-ntsecapi-rtlgenrandom +//! [`BCryptGenRandom`]: https://docs.microsoft.com/en-us/windows/win32/api/bcrypt/nf-bcrypt-bcryptgenrandom use crate::mem; use crate::ptr; use crate::sys::c; +/// Generates high quality secure random keys for use by [`HashMap`]. +/// +/// This is used to seed the default [`RandomState`]. +/// +/// [`HashMap`]: crate::collections::HashMap +/// [`RandomState`]: crate::collections::hash_map::RandomState pub fn hashmap_random_keys() -> (u64, u64) { - let mut v = (0, 0); - let ret = unsafe { - c::BCryptGenRandom( - ptr::null_mut(), - &mut v as *mut _ as *mut u8, - mem::size_of_val(&v) as c::ULONG, - c::BCRYPT_USE_SYSTEM_PREFERRED_RNG, - ) - }; - if ret != 0 { fallback_rng() } else { v } + Rng::SYSTEM.gen_random_keys().unwrap_or_else(fallback_rng) } -/// Generate random numbers using the fallback RNG function (RtlGenRandom) -#[cfg(not(target_vendor = "uwp"))] -#[inline(never)] -fn fallback_rng() -> (u64, u64) { - let mut v = (0, 0); - let ret = - unsafe { c::RtlGenRandom(&mut v as *mut _ as *mut u8, mem::size_of_val(&v) as c::ULONG) }; +struct Rng { + algorithm: c::BCRYPT_ALG_HANDLE, + flags: u32, +} +impl Rng { + const SYSTEM: Self = unsafe { Self::new(ptr::null_mut(), c::BCRYPT_USE_SYSTEM_PREFERRED_RNG) }; + + /// Create the RNG from an existing algorithm handle. + /// + /// # Safety + /// + /// The handle must either be null or a valid algorithm handle. + const unsafe fn new(algorithm: c::BCRYPT_ALG_HANDLE, flags: u32) -> Self { + Self { algorithm, flags } + } + + /// Open a handle to the RNG algorithm. + fn open() -> Result<Self, c::NTSTATUS> { + use crate::sync::atomic::AtomicPtr; + use crate::sync::atomic::Ordering::{Acquire, Release}; + + // An atomic is used so we don't need to reopen the handle every time. + static HANDLE: AtomicPtr<crate::ffi::c_void> = AtomicPtr::new(ptr::null_mut()); + + let mut handle = HANDLE.load(Acquire); + if handle.is_null() { + let status = unsafe { + c::BCryptOpenAlgorithmProvider( + &mut handle, + c::BCRYPT_RNG_ALGORITHM.as_ptr(), + ptr::null(), + 0, + ) + }; + if c::nt_success(status) { + // If another thread opens a handle first then use that handle instead. + let result = HANDLE.compare_exchange(ptr::null_mut(), handle, Release, Acquire); + if let Err(previous_handle) = result { + // Close our handle and return the previous one. + unsafe { c::BCryptCloseAlgorithmProvider(handle, 0) }; + handle = previous_handle; + } + Ok(unsafe { Self::new(handle, 0) }) + } else { + Err(status) + } + } else { + Ok(unsafe { Self::new(handle, 0) }) + } + } - if ret != 0 { v } else { panic!("fallback RNG broken: {}", io::Error::last_os_error()) } + fn gen_random_keys(self) -> Result<(u64, u64), c::NTSTATUS> { + let mut v = (0, 0); + let status = unsafe { + let size = mem::size_of_val(&v).try_into().unwrap(); + c::BCryptGenRandom(self.algorithm, ptr::addr_of_mut!(v).cast(), size, self.flags) + }; + if c::nt_success(status) { Ok(v) } else { Err(status) } + } } -/// We can't use RtlGenRandom with UWP, so there is no fallback -#[cfg(target_vendor = "uwp")] +/// Generate random numbers using the fallback RNG function #[inline(never)] -fn fallback_rng() -> (u64, u64) { - panic!("fallback RNG broken: RtlGenRandom() not supported on UWP"); +fn fallback_rng(rng_status: c::NTSTATUS) -> (u64, u64) { + match Rng::open().and_then(|rng| rng.gen_random_keys()) { + Ok(keys) => keys, + Err(status) => { + panic!("RNG broken: {rng_status:#x}, fallback RNG broken: {status:#x}") + } + } } diff --git a/library/std/src/sys/windows/stdio.rs b/library/std/src/sys/windows/stdio.rs index a001d6b98..70c9b14a0 100644 --- a/library/std/src/sys/windows/stdio.rs +++ b/library/std/src/sys/windows/stdio.rs @@ -3,6 +3,7 @@ use crate::char::decode_utf16; use crate::cmp; use crate::io; +use crate::mem::MaybeUninit; use crate::os::windows::io::{FromRawHandle, IntoRawHandle}; use crate::ptr; use crate::str; @@ -169,13 +170,14 @@ fn write( } fn write_valid_utf8_to_console(handle: c::HANDLE, utf8: &str) -> io::Result<usize> { - let mut utf16 = [0u16; MAX_BUFFER_SIZE / 2]; + let mut utf16 = [MaybeUninit::<u16>::uninit(); MAX_BUFFER_SIZE / 2]; let mut len_utf16 = 0; for (chr, dest) in utf8.encode_utf16().zip(utf16.iter_mut()) { - *dest = chr; + *dest = MaybeUninit::new(chr); len_utf16 += 1; } - let utf16 = &utf16[..len_utf16]; + // Safety: We've initialized `len_utf16` values. + let utf16: &[u16] = unsafe { MaybeUninit::slice_assume_init_ref(&utf16[..len_utf16]) }; let mut written = write_u16s(handle, &utf16)?; @@ -250,11 +252,14 @@ impl io::Read for Stdin { return Ok(bytes_copied); } else if buf.len() - bytes_copied < 4 { // Not enough space to get a UTF-8 byte. We will use the incomplete UTF8. - let mut utf16_buf = [0u16; 1]; + let mut utf16_buf = [MaybeUninit::new(0); 1]; // Read one u16 character. let read = read_u16s_fixup_surrogates(handle, &mut utf16_buf, 1, &mut self.surrogate)?; // Read bytes, using the (now-empty) self.incomplete_utf8 as extra space. - let read_bytes = utf16_to_utf8(&utf16_buf[..read], &mut self.incomplete_utf8.bytes)?; + let read_bytes = utf16_to_utf8( + unsafe { MaybeUninit::slice_assume_init_ref(&utf16_buf[..read]) }, + &mut self.incomplete_utf8.bytes, + )?; // Read in the bytes from incomplete_utf8 until the buffer is full. self.incomplete_utf8.len = read_bytes as u8; @@ -262,15 +267,18 @@ impl io::Read for Stdin { bytes_copied += self.incomplete_utf8.read(&mut buf[bytes_copied..]); Ok(bytes_copied) } else { - let mut utf16_buf = [0u16; MAX_BUFFER_SIZE / 2]; + let mut utf16_buf = [MaybeUninit::<u16>::uninit(); MAX_BUFFER_SIZE / 2]; + // In the worst case, a UTF-8 string can take 3 bytes for every `u16` of a UTF-16. So // we can read at most a third of `buf.len()` chars and uphold the guarantee no data gets // lost. let amount = cmp::min(buf.len() / 3, utf16_buf.len()); let read = read_u16s_fixup_surrogates(handle, &mut utf16_buf, amount, &mut self.surrogate)?; - - match utf16_to_utf8(&utf16_buf[..read], buf) { + // Safety `read_u16s_fixup_surrogates` returns the number of items + // initialized. + let utf16s = unsafe { MaybeUninit::slice_assume_init_ref(&utf16_buf[..read]) }; + match utf16_to_utf8(utf16s, buf) { Ok(value) => return Ok(bytes_copied + value), Err(e) => return Err(e), } @@ -283,14 +291,14 @@ impl io::Read for Stdin { // This is a best effort, and might not work if we are not the only reader on Stdin. fn read_u16s_fixup_surrogates( handle: c::HANDLE, - buf: &mut [u16], + buf: &mut [MaybeUninit<u16>], mut amount: usize, surrogate: &mut u16, ) -> io::Result<usize> { // Insert possibly remaining unpaired surrogate from last read. let mut start = 0; if *surrogate != 0 { - buf[0] = *surrogate; + buf[0] = MaybeUninit::new(*surrogate); *surrogate = 0; start = 1; if amount == 1 { @@ -303,7 +311,10 @@ fn read_u16s_fixup_surrogates( let mut amount = read_u16s(handle, &mut buf[start..amount])? + start; if amount > 0 { - let last_char = buf[amount - 1]; + // Safety: The returned `amount` is the number of values initialized, + // and it is not 0, so we know that `buf[amount - 1]` have been + // initialized. + let last_char = unsafe { buf[amount - 1].assume_init() }; if last_char >= 0xD800 && last_char <= 0xDBFF { // high surrogate *surrogate = last_char; @@ -313,7 +324,8 @@ fn read_u16s_fixup_surrogates( Ok(amount) } -fn read_u16s(handle: c::HANDLE, buf: &mut [u16]) -> io::Result<usize> { +// Returns `Ok(n)` if it initialized `n` values in `buf`. +fn read_u16s(handle: c::HANDLE, buf: &mut [MaybeUninit<u16>]) -> io::Result<usize> { // Configure the `pInputControl` parameter to not only return on `\r\n` but also Ctrl-Z, the // traditional DOS method to indicate end of character stream / user input (SUB). // See #38274 and https://stackoverflow.com/questions/43836040/win-api-readconsole. @@ -346,8 +358,9 @@ fn read_u16s(handle: c::HANDLE, buf: &mut [u16]) -> io::Result<usize> { } break; } - - if amount > 0 && buf[amount as usize - 1] == CTRL_Z { + // Safety: if `amount > 0`, then that many bytes were written, so + // `buf[amount as usize - 1]` has been initialized. + if amount > 0 && unsafe { buf[amount as usize - 1].assume_init() } == CTRL_Z { amount -= 1; } Ok(amount as usize) diff --git a/library/std/src/sys/windows/thread_local_dtor.rs b/library/std/src/sys/windows/thread_local_dtor.rs index 25d1c6e8e..9707a95df 100644 --- a/library/std/src/sys/windows/thread_local_dtor.rs +++ b/library/std/src/sys/windows/thread_local_dtor.rs @@ -8,10 +8,14 @@ #[thread_local] static mut DESTRUCTORS: Vec<(*mut u8, unsafe extern "C" fn(*mut u8))> = Vec::new(); +// Ensure this can never be inlined because otherwise this may break in dylibs. +// See #44391. +#[inline(never)] pub unsafe fn register_dtor(t: *mut u8, dtor: unsafe extern "C" fn(*mut u8)) { DESTRUCTORS.push((t, dtor)); } +#[inline(never)] // See comment above /// Runs destructors. This should not be called until thread exit. pub unsafe fn run_keyless_dtors() { // Drop all the destructors. diff --git a/library/std/src/sys/windows/thread_local_key.rs b/library/std/src/sys/windows/thread_local_key.rs index ec670238e..17628b757 100644 --- a/library/std/src/sys/windows/thread_local_key.rs +++ b/library/std/src/sys/windows/thread_local_key.rs @@ -1,11 +1,16 @@ -use crate::mem::ManuallyDrop; +use crate::cell::UnsafeCell; use crate::ptr; -use crate::sync::atomic::AtomicPtr; -use crate::sync::atomic::Ordering::SeqCst; +use crate::sync::atomic::{ + AtomicPtr, AtomicU32, + Ordering::{AcqRel, Acquire, Relaxed, Release}, +}; use crate::sys::c; -pub type Key = c::DWORD; -pub type Dtor = unsafe extern "C" fn(*mut u8); +#[cfg(test)] +mod tests; + +type Key = c::DWORD; +type Dtor = unsafe extern "C" fn(*mut u8); // Turns out, like pretty much everything, Windows is pretty close the // functionality that Unix provides, but slightly different! In the case of @@ -22,60 +27,109 @@ pub type Dtor = unsafe extern "C" fn(*mut u8); // To accomplish this feat, we perform a number of threads, all contained // within this module: // -// * All TLS destructors are tracked by *us*, not the windows runtime. This +// * All TLS destructors are tracked by *us*, not the Windows runtime. This // means that we have a global list of destructors for each TLS key that // we know about. // * When a thread exits, we run over the entire list and run dtors for all // non-null keys. This attempts to match Unix semantics in this regard. // -// This ends up having the overhead of using a global list, having some -// locks here and there, and in general just adding some more code bloat. We -// attempt to optimize runtime by forgetting keys that don't have -// destructors, but this only gets us so far. -// // For more details and nitty-gritty, see the code sections below! // // [1]: https://www.codeproject.com/Articles/8113/Thread-Local-Storage-The-C-Way -// [2]: https://github.com/ChromiumWebApps/chromium/blob/master/base -// /threading/thread_local_storage_win.cc#L42 +// [2]: https://github.com/ChromiumWebApps/chromium/blob/master/base/threading/thread_local_storage_win.cc#L42 -// ------------------------------------------------------------------------- -// Native bindings -// -// This section is just raw bindings to the native functions that Windows -// provides, There's a few extra calls to deal with destructors. +pub struct StaticKey { + /// The key value shifted up by one. Since TLS_OUT_OF_INDEXES == DWORD::MAX + /// is not a valid key value, this allows us to use zero as sentinel value + /// without risking overflow. + key: AtomicU32, + dtor: Option<Dtor>, + next: AtomicPtr<StaticKey>, + /// Currently, destructors cannot be unregistered, so we cannot use racy + /// initialization for keys. Instead, we need synchronize initialization. + /// Use the Windows-provided `Once` since it does not require TLS. + once: UnsafeCell<c::INIT_ONCE>, +} -#[inline] -pub unsafe fn create(dtor: Option<Dtor>) -> Key { - let key = c::TlsAlloc(); - assert!(key != c::TLS_OUT_OF_INDEXES); - if let Some(f) = dtor { - register_dtor(key, f); +impl StaticKey { + #[inline] + pub const fn new(dtor: Option<Dtor>) -> StaticKey { + StaticKey { + key: AtomicU32::new(0), + dtor, + next: AtomicPtr::new(ptr::null_mut()), + once: UnsafeCell::new(c::INIT_ONCE_STATIC_INIT), + } } - key -} -#[inline] -pub unsafe fn set(key: Key, value: *mut u8) { - let r = c::TlsSetValue(key, value as c::LPVOID); - debug_assert!(r != 0); -} + #[inline] + pub unsafe fn set(&'static self, val: *mut u8) { + let r = c::TlsSetValue(self.key(), val.cast()); + debug_assert_eq!(r, c::TRUE); + } -#[inline] -pub unsafe fn get(key: Key) -> *mut u8 { - c::TlsGetValue(key) as *mut u8 -} + #[inline] + pub unsafe fn get(&'static self) -> *mut u8 { + c::TlsGetValue(self.key()).cast() + } -#[inline] -pub unsafe fn destroy(_key: Key) { - rtabort!("can't destroy tls keys on windows") -} + #[inline] + unsafe fn key(&'static self) -> Key { + match self.key.load(Acquire) { + 0 => self.init(), + key => key - 1, + } + } + + #[cold] + unsafe fn init(&'static self) -> Key { + if self.dtor.is_some() { + let mut pending = c::FALSE; + let r = c::InitOnceBeginInitialize(self.once.get(), 0, &mut pending, ptr::null_mut()); + assert_eq!(r, c::TRUE); -#[inline] -pub fn requires_synchronized_create() -> bool { - true + if pending == c::FALSE { + // Some other thread initialized the key, load it. + self.key.load(Relaxed) - 1 + } else { + let key = c::TlsAlloc(); + if key == c::TLS_OUT_OF_INDEXES { + // Wakeup the waiting threads before panicking to avoid deadlock. + c::InitOnceComplete(self.once.get(), c::INIT_ONCE_INIT_FAILED, ptr::null_mut()); + panic!("out of TLS indexes"); + } + + self.key.store(key + 1, Release); + register_dtor(self); + + let r = c::InitOnceComplete(self.once.get(), 0, ptr::null_mut()); + debug_assert_eq!(r, c::TRUE); + + key + } + } else { + // If there is no destructor to clean up, we can use racy initialization. + + let key = c::TlsAlloc(); + assert_ne!(key, c::TLS_OUT_OF_INDEXES, "out of TLS indexes"); + + match self.key.compare_exchange(0, key + 1, AcqRel, Acquire) { + Ok(_) => key, + Err(new) => { + // Some other thread completed initialization first, so destroy + // our key and use theirs. + let r = c::TlsFree(key); + debug_assert_eq!(r, c::TRUE); + new - 1 + } + } + } + } } +unsafe impl Send for StaticKey {} +unsafe impl Sync for StaticKey {} + // ------------------------------------------------------------------------- // Dtor registration // @@ -96,29 +150,21 @@ pub fn requires_synchronized_create() -> bool { // Typically processes have a statically known set of TLS keys which is pretty // small, and we'd want to keep this memory alive for the whole process anyway // really. -// -// Perhaps one day we can fold the `Box` here into a static allocation, -// expanding the `StaticKey` structure to contain not only a slot for the TLS -// key but also a slot for the destructor queue on windows. An optimization for -// another day! - -static DTORS: AtomicPtr<Node> = AtomicPtr::new(ptr::null_mut()); - -struct Node { - dtor: Dtor, - key: Key, - next: *mut Node, -} -unsafe fn register_dtor(key: Key, dtor: Dtor) { - let mut node = ManuallyDrop::new(Box::new(Node { key, dtor, next: ptr::null_mut() })); +static DTORS: AtomicPtr<StaticKey> = AtomicPtr::new(ptr::null_mut()); - let mut head = DTORS.load(SeqCst); +/// Should only be called once per key, otherwise loops or breaks may occur in +/// the linked list. +unsafe fn register_dtor(key: &'static StaticKey) { + let this = <*const StaticKey>::cast_mut(key); + // Use acquire ordering to pass along the changes done by the previously + // registered keys when we store the new head with release ordering. + let mut head = DTORS.load(Acquire); loop { - node.next = head; - match DTORS.compare_exchange(head, &mut **node, SeqCst, SeqCst) { - Ok(_) => return, // nothing to drop, we successfully added the node to the list - Err(cur) => head = cur, + key.next.store(head, Relaxed); + match DTORS.compare_exchange_weak(head, this, Release, Acquire) { + Ok(_) => break, + Err(new) => head = new, } } } @@ -214,25 +260,29 @@ unsafe extern "system" fn on_tls_callback(h: c::LPVOID, dwReason: c::DWORD, pv: unsafe fn reference_tls_used() {} } -#[allow(dead_code)] // actually called above +#[allow(dead_code)] // actually called below unsafe fn run_dtors() { - let mut any_run = true; for _ in 0..5 { - if !any_run { - break; - } - any_run = false; - let mut cur = DTORS.load(SeqCst); + let mut any_run = false; + + // Use acquire ordering to observe key initialization. + let mut cur = DTORS.load(Acquire); while !cur.is_null() { - let ptr = c::TlsGetValue((*cur).key); + let key = (*cur).key.load(Relaxed) - 1; + let dtor = (*cur).dtor.unwrap(); + let ptr = c::TlsGetValue(key); if !ptr.is_null() { - c::TlsSetValue((*cur).key, ptr::null_mut()); - ((*cur).dtor)(ptr as *mut _); + c::TlsSetValue(key, ptr::null_mut()); + dtor(ptr as *mut _); any_run = true; } - cur = (*cur).next; + cur = (*cur).next.load(Relaxed); + } + + if !any_run { + break; } } } diff --git a/library/std/src/sys/windows/thread_local_key/tests.rs b/library/std/src/sys/windows/thread_local_key/tests.rs new file mode 100644 index 000000000..c95f383fb --- /dev/null +++ b/library/std/src/sys/windows/thread_local_key/tests.rs @@ -0,0 +1,53 @@ +use super::StaticKey; +use crate::ptr; + +#[test] +fn smoke() { + static K1: StaticKey = StaticKey::new(None); + static K2: StaticKey = StaticKey::new(None); + + unsafe { + assert!(K1.get().is_null()); + assert!(K2.get().is_null()); + K1.set(ptr::invalid_mut(1)); + K2.set(ptr::invalid_mut(2)); + assert_eq!(K1.get() as usize, 1); + assert_eq!(K2.get() as usize, 2); + } +} + +#[test] +fn destructors() { + use crate::mem::ManuallyDrop; + use crate::sync::Arc; + use crate::thread; + + unsafe extern "C" fn destruct(ptr: *mut u8) { + drop(Arc::from_raw(ptr as *const ())); + } + + static KEY: StaticKey = StaticKey::new(Some(destruct)); + + let shared1 = Arc::new(()); + let shared2 = Arc::clone(&shared1); + + unsafe { + assert!(KEY.get().is_null()); + KEY.set(Arc::into_raw(shared1) as *mut u8); + } + + thread::spawn(move || unsafe { + assert!(KEY.get().is_null()); + KEY.set(Arc::into_raw(shared2) as *mut u8); + }) + .join() + .unwrap(); + + // Leak the Arc, let the TLS destructor clean it up. + let shared1 = unsafe { ManuallyDrop::new(Arc::from_raw(KEY.get() as *const ())) }; + assert_eq!( + Arc::strong_count(&shared1), + 1, + "destructor should have dropped the other reference on thread exit" + ); +} diff --git a/library/std/src/sys/windows/thread_parker.rs b/library/std/src/sys/windows/thread_parker.rs index d876e0f6f..2f7ae863b 100644 --- a/library/std/src/sys/windows/thread_parker.rs +++ b/library/std/src/sys/windows/thread_parker.rs @@ -197,19 +197,17 @@ impl Parker { // purpose, to make sure every unpark() has a release-acquire ordering // with park(). if self.state.swap(NOTIFIED, Release) == PARKED { - if let Some(wake_by_address_single) = c::WakeByAddressSingle::option() { - unsafe { + unsafe { + if let Some(wake_by_address_single) = c::WakeByAddressSingle::option() { wake_by_address_single(self.ptr()); - } - } else { - // If we run NtReleaseKeyedEvent before the waiting thread runs - // NtWaitForKeyedEvent, this (shortly) blocks until we can wake it up. - // If the waiting thread wakes up before we run NtReleaseKeyedEvent - // (e.g. due to a timeout), this blocks until we do wake up a thread. - // To prevent this thread from blocking indefinitely in that case, - // park_impl() will, after seeing the state set to NOTIFIED after - // waking up, call NtWaitForKeyedEvent again to unblock us. - unsafe { + } else { + // If we run NtReleaseKeyedEvent before the waiting thread runs + // NtWaitForKeyedEvent, this (shortly) blocks until we can wake it up. + // If the waiting thread wakes up before we run NtReleaseKeyedEvent + // (e.g. due to a timeout), this blocks until we do wake up a thread. + // To prevent this thread from blocking indefinitely in that case, + // park_impl() will, after seeing the state set to NOTIFIED after + // waking up, call NtWaitForKeyedEvent again to unblock us. c::NtReleaseKeyedEvent(keyed_event_handle(), self.ptr(), 0, ptr::null_mut()); } } diff --git a/library/std/src/sys_common/backtrace.rs b/library/std/src/sys_common/backtrace.rs index 31164afdc..8807077cb 100644 --- a/library/std/src/sys_common/backtrace.rs +++ b/library/std/src/sys_common/backtrace.rs @@ -7,15 +7,14 @@ use crate::fmt; use crate::io; use crate::io::prelude::*; use crate::path::{self, Path, PathBuf}; -use crate::sys_common::mutex::StaticMutex; +use crate::sync::{Mutex, PoisonError}; /// Max number of frames to print. const MAX_NB_FRAMES: usize = 100; -// SAFETY: Don't attempt to lock this reentrantly. -pub unsafe fn lock() -> impl Drop { - static LOCK: StaticMutex = StaticMutex::new(); - LOCK.lock() +pub fn lock() -> impl Drop { + static LOCK: Mutex<()> = Mutex::new(()); + LOCK.lock().unwrap_or_else(PoisonError::into_inner) } /// Prints the current backtrace. diff --git a/library/std/src/sys_common/condvar.rs b/library/std/src/sys_common/condvar.rs index f3ac1061b..8bc5b2411 100644 --- a/library/std/src/sys_common/condvar.rs +++ b/library/std/src/sys_common/condvar.rs @@ -15,6 +15,7 @@ pub struct Condvar { impl Condvar { /// Creates a new condition variable for use. #[inline] + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> Self { Self { inner: imp::MovableCondvar::new(), check: CondvarCheck::new() } } diff --git a/library/std/src/sys_common/condvar/check.rs b/library/std/src/sys_common/condvar/check.rs index ce8f36704..4ac9e62bf 100644 --- a/library/std/src/sys_common/condvar/check.rs +++ b/library/std/src/sys_common/condvar/check.rs @@ -50,6 +50,7 @@ pub struct NoCheck; #[allow(dead_code)] impl NoCheck { + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> Self { Self } diff --git a/library/std/src/sys_common/mod.rs b/library/std/src/sys_common/mod.rs index 80f56bf75..8c19f9332 100644 --- a/library/std/src/sys_common/mod.rs +++ b/library/std/src/sys_common/mod.rs @@ -27,17 +27,25 @@ pub mod io; pub mod lazy_box; pub mod memchr; pub mod mutex; +pub mod once; pub mod process; pub mod remutex; pub mod rwlock; pub mod thread; pub mod thread_info; pub mod thread_local_dtor; -pub mod thread_local_key; pub mod thread_parker; pub mod wtf8; cfg_if::cfg_if! { + if #[cfg(target_os = "windows")] { + pub use crate::sys::thread_local_key; + } else { + pub mod thread_local_key; + } +} + +cfg_if::cfg_if! { if #[cfg(any(target_os = "l4re", target_os = "hermit", feature = "restricted-std", diff --git a/library/std/src/sys_common/mutex.rs b/library/std/src/sys_common/mutex.rs index 48479f5bd..98046f20f 100644 --- a/library/std/src/sys_common/mutex.rs +++ b/library/std/src/sys_common/mutex.rs @@ -1,49 +1,5 @@ use crate::sys::locks as imp; -/// An OS-based mutual exclusion lock, meant for use in static variables. -/// -/// This mutex has a const constructor ([`StaticMutex::new`]), does not -/// implement `Drop` to cleanup resources, and causes UB when used reentrantly. -/// -/// This mutex does not implement poisoning. -/// -/// This is a wrapper around `imp::Mutex` that does *not* call `init()` and -/// `destroy()`. -pub struct StaticMutex(imp::Mutex); - -unsafe impl Sync for StaticMutex {} - -impl StaticMutex { - /// Creates a new mutex for use. - #[inline] - pub const fn new() -> Self { - Self(imp::Mutex::new()) - } - - /// Calls raw_lock() and then returns an RAII guard to guarantee the mutex - /// will be unlocked. - /// - /// It is undefined behaviour to call this function while locked by the - /// same thread. - #[inline] - pub unsafe fn lock(&'static self) -> StaticMutexGuard { - self.0.lock(); - StaticMutexGuard(&self.0) - } -} - -#[must_use] -pub struct StaticMutexGuard(&'static imp::Mutex); - -impl Drop for StaticMutexGuard { - #[inline] - fn drop(&mut self) { - unsafe { - self.0.unlock(); - } - } -} - /// An OS-based mutual exclusion lock. /// /// This mutex cleans up its resources in its `Drop` implementation, may safely @@ -61,6 +17,7 @@ unsafe impl Sync for MovableMutex {} impl MovableMutex { /// Creates a new mutex. #[inline] + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> Self { Self(imp::MovableMutex::new()) } diff --git a/library/std/src/sys_common/net.rs b/library/std/src/sys_common/net.rs index 33d336c43..fad4a6333 100644 --- a/library/std/src/sys_common/net.rs +++ b/library/std/src/sys_common/net.rs @@ -2,15 +2,16 @@ mod tests; use crate::cmp; -use crate::ffi::CString; +use crate::convert::{TryFrom, TryInto}; use crate::fmt; use crate::io::{self, ErrorKind, IoSlice, IoSliceMut}; use crate::mem; use crate::net::{Ipv4Addr, Ipv6Addr, Shutdown, SocketAddr}; use crate::ptr; +use crate::sys::common::small_c_string::run_with_cstr; use crate::sys::net::netc as c; use crate::sys::net::{cvt, cvt_gai, cvt_r, init, wrlen_t, Socket}; -use crate::sys_common::{FromInner, IntoInner}; +use crate::sys_common::{AsInner, FromInner, IntoInner}; use crate::time::Duration; use libc::{c_int, c_void}; @@ -197,14 +198,15 @@ impl<'a> TryFrom<(&'a str, u16)> for LookupHost { fn try_from((host, port): (&'a str, u16)) -> io::Result<LookupHost> { init(); - let c_host = CString::new(host)?; - let mut hints: c::addrinfo = unsafe { mem::zeroed() }; - hints.ai_socktype = c::SOCK_STREAM; - let mut res = ptr::null_mut(); - unsafe { - cvt_gai(c::getaddrinfo(c_host.as_ptr(), ptr::null(), &hints, &mut res)) - .map(|_| LookupHost { original: res, cur: res, port }) - } + run_with_cstr(host.as_bytes(), |c_host| { + let mut hints: c::addrinfo = unsafe { mem::zeroed() }; + hints.ai_socktype = c::SOCK_STREAM; + let mut res = ptr::null_mut(); + unsafe { + cvt_gai(c::getaddrinfo(c_host.as_ptr(), ptr::null(), &hints, &mut res)) + .map(|_| LookupHost { original: res, cur: res, port }) + } + }) } } @@ -345,6 +347,12 @@ impl TcpStream { } } +impl AsInner<Socket> for TcpStream { + fn as_inner(&self) -> &Socket { + &self.inner + } +} + impl FromInner<Socket> for TcpStream { fn from_inner(socket: Socket) -> TcpStream { TcpStream { inner: socket } diff --git a/library/std/src/sys_common/once/futex.rs b/library/std/src/sys_common/once/futex.rs new file mode 100644 index 000000000..5c7e6c013 --- /dev/null +++ b/library/std/src/sys_common/once/futex.rs @@ -0,0 +1,134 @@ +use crate::cell::Cell; +use crate::sync as public; +use crate::sync::atomic::{ + AtomicU32, + Ordering::{Acquire, Relaxed, Release}, +}; +use crate::sys::futex::{futex_wait, futex_wake_all}; + +// On some platforms, the OS is very nice and handles the waiter queue for us. +// This means we only need one atomic value with 5 states: + +/// No initialization has run yet, and no thread is currently using the Once. +const INCOMPLETE: u32 = 0; +/// Some thread has previously attempted to initialize the Once, but it panicked, +/// so the Once is now poisoned. There are no other threads currently accessing +/// this Once. +const POISONED: u32 = 1; +/// Some thread is currently attempting to run initialization. It may succeed, +/// so all future threads need to wait for it to finish. +const RUNNING: u32 = 2; +/// Some thread is currently attempting to run initialization and there are threads +/// waiting for it to finish. +const QUEUED: u32 = 3; +/// Initialization has completed and all future calls should finish immediately. +const COMPLETE: u32 = 4; + +// Threads wait by setting the state to QUEUED and calling `futex_wait` on the state +// variable. When the running thread finishes, it will wake all waiting threads using +// `futex_wake_all`. + +pub struct OnceState { + poisoned: bool, + set_state_to: Cell<u32>, +} + +impl OnceState { + #[inline] + pub fn is_poisoned(&self) -> bool { + self.poisoned + } + + #[inline] + pub fn poison(&self) { + self.set_state_to.set(POISONED); + } +} + +struct CompletionGuard<'a> { + state: &'a AtomicU32, + set_state_on_drop_to: u32, +} + +impl<'a> Drop for CompletionGuard<'a> { + fn drop(&mut self) { + // Use release ordering to propagate changes to all threads checking + // up on the Once. `futex_wake_all` does its own synchronization, hence + // we do not need `AcqRel`. + if self.state.swap(self.set_state_on_drop_to, Release) == QUEUED { + futex_wake_all(&self.state); + } + } +} + +pub struct Once { + state: AtomicU32, +} + +impl Once { + #[inline] + pub const fn new() -> Once { + Once { state: AtomicU32::new(INCOMPLETE) } + } + + #[inline] + pub fn is_completed(&self) -> bool { + // Use acquire ordering to make all initialization changes visible to the + // current thread. + self.state.load(Acquire) == COMPLETE + } + + // This uses FnMut to match the API of the generic implementation. As this + // implementation is quite light-weight, it is generic over the closure and + // so avoids the cost of dynamic dispatch. + #[cold] + #[track_caller] + pub fn call(&self, ignore_poisoning: bool, f: &mut impl FnMut(&public::OnceState)) { + let mut state = self.state.load(Acquire); + loop { + match state { + POISONED if !ignore_poisoning => { + // Panic to propagate the poison. + panic!("Once instance has previously been poisoned"); + } + INCOMPLETE | POISONED => { + // Try to register the current thread as the one running. + if let Err(new) = + self.state.compare_exchange_weak(state, RUNNING, Acquire, Acquire) + { + state = new; + continue; + } + // `waiter_queue` will manage other waiting threads, and + // wake them up on drop. + let mut waiter_queue = + CompletionGuard { state: &self.state, set_state_on_drop_to: POISONED }; + // Run the function, letting it know if we're poisoned or not. + let f_state = public::OnceState { + inner: OnceState { + poisoned: state == POISONED, + set_state_to: Cell::new(COMPLETE), + }, + }; + f(&f_state); + waiter_queue.set_state_on_drop_to = f_state.inner.set_state_to.get(); + return; + } + RUNNING | QUEUED => { + // Set the state to QUEUED if it is not already. + if state == RUNNING + && let Err(new) = self.state.compare_exchange_weak(RUNNING, QUEUED, Relaxed, Acquire) + { + state = new; + continue; + } + + futex_wait(&self.state, QUEUED, None); + state = self.state.load(Acquire); + } + COMPLETE => return, + _ => unreachable!("state is never set to invalid values"), + } + } + } +} diff --git a/library/std/src/sys_common/once/generic.rs b/library/std/src/sys_common/once/generic.rs new file mode 100644 index 000000000..acf5f2471 --- /dev/null +++ b/library/std/src/sys_common/once/generic.rs @@ -0,0 +1,282 @@ +// Each `Once` has one word of atomic state, and this state is CAS'd on to +// determine what to do. There are four possible state of a `Once`: +// +// * Incomplete - no initialization has run yet, and no thread is currently +// using the Once. +// * Poisoned - some thread has previously attempted to initialize the Once, but +// it panicked, so the Once is now poisoned. There are no other +// threads currently accessing this Once. +// * Running - some thread is currently attempting to run initialization. It may +// succeed, so all future threads need to wait for it to finish. +// Note that this state is accompanied with a payload, described +// below. +// * Complete - initialization has completed and all future calls should finish +// immediately. +// +// With 4 states we need 2 bits to encode this, and we use the remaining bits +// in the word we have allocated as a queue of threads waiting for the thread +// responsible for entering the RUNNING state. This queue is just a linked list +// of Waiter nodes which is monotonically increasing in size. Each node is +// allocated on the stack, and whenever the running closure finishes it will +// consume the entire queue and notify all waiters they should try again. +// +// You'll find a few more details in the implementation, but that's the gist of +// it! +// +// Atomic orderings: +// When running `Once` we deal with multiple atomics: +// `Once.state_and_queue` and an unknown number of `Waiter.signaled`. +// * `state_and_queue` is used (1) as a state flag, (2) for synchronizing the +// result of the `Once`, and (3) for synchronizing `Waiter` nodes. +// - At the end of the `call` function we have to make sure the result +// of the `Once` is acquired. So every load which can be the only one to +// load COMPLETED must have at least acquire ordering, which means all +// three of them. +// - `WaiterQueue::drop` is the only place that may store COMPLETED, and +// must do so with release ordering to make the result available. +// - `wait` inserts `Waiter` nodes as a pointer in `state_and_queue`, and +// needs to make the nodes available with release ordering. The load in +// its `compare_exchange` can be relaxed because it only has to compare +// the atomic, not to read other data. +// - `WaiterQueue::drop` must see the `Waiter` nodes, so it must load +// `state_and_queue` with acquire ordering. +// - There is just one store where `state_and_queue` is used only as a +// state flag, without having to synchronize data: switching the state +// from INCOMPLETE to RUNNING in `call`. This store can be Relaxed, +// but the read has to be Acquire because of the requirements mentioned +// above. +// * `Waiter.signaled` is both used as a flag, and to protect a field with +// interior mutability in `Waiter`. `Waiter.thread` is changed in +// `WaiterQueue::drop` which then sets `signaled` with release ordering. +// After `wait` loads `signaled` with acquire ordering and sees it is true, +// it needs to see the changes to drop the `Waiter` struct correctly. +// * There is one place where the two atomics `Once.state_and_queue` and +// `Waiter.signaled` come together, and might be reordered by the compiler or +// processor. Because both use acquire ordering such a reordering is not +// allowed, so no need for `SeqCst`. + +use crate::cell::Cell; +use crate::fmt; +use crate::ptr; +use crate::sync as public; +use crate::sync::atomic::{AtomicBool, AtomicPtr, Ordering}; +use crate::thread::{self, Thread}; + +type Masked = (); + +pub struct Once { + state_and_queue: AtomicPtr<Masked>, +} + +pub struct OnceState { + poisoned: bool, + set_state_on_drop_to: Cell<*mut Masked>, +} + +// Four states that a Once can be in, encoded into the lower bits of +// `state_and_queue` in the Once structure. +const INCOMPLETE: usize = 0x0; +const POISONED: usize = 0x1; +const RUNNING: usize = 0x2; +const COMPLETE: usize = 0x3; + +// Mask to learn about the state. All other bits are the queue of waiters if +// this is in the RUNNING state. +const STATE_MASK: usize = 0x3; + +// Representation of a node in the linked list of waiters, used while in the +// RUNNING state. +// Note: `Waiter` can't hold a mutable pointer to the next thread, because then +// `wait` would both hand out a mutable reference to its `Waiter` node, and keep +// a shared reference to check `signaled`. Instead we hold shared references and +// use interior mutability. +#[repr(align(4))] // Ensure the two lower bits are free to use as state bits. +struct Waiter { + thread: Cell<Option<Thread>>, + signaled: AtomicBool, + next: *const Waiter, +} + +// Head of a linked list of waiters. +// Every node is a struct on the stack of a waiting thread. +// Will wake up the waiters when it gets dropped, i.e. also on panic. +struct WaiterQueue<'a> { + state_and_queue: &'a AtomicPtr<Masked>, + set_state_on_drop_to: *mut Masked, +} + +impl Once { + #[inline] + pub const fn new() -> Once { + Once { state_and_queue: AtomicPtr::new(ptr::invalid_mut(INCOMPLETE)) } + } + + #[inline] + pub fn is_completed(&self) -> bool { + // An `Acquire` load is enough because that makes all the initialization + // operations visible to us, and, this being a fast path, weaker + // ordering helps with performance. This `Acquire` synchronizes with + // `Release` operations on the slow path. + self.state_and_queue.load(Ordering::Acquire).addr() == COMPLETE + } + + // This is a non-generic function to reduce the monomorphization cost of + // using `call_once` (this isn't exactly a trivial or small implementation). + // + // Additionally, this is tagged with `#[cold]` as it should indeed be cold + // and it helps let LLVM know that calls to this function should be off the + // fast path. Essentially, this should help generate more straight line code + // in LLVM. + // + // Finally, this takes an `FnMut` instead of a `FnOnce` because there's + // currently no way to take an `FnOnce` and call it via virtual dispatch + // without some allocation overhead. + #[cold] + #[track_caller] + pub fn call(&self, ignore_poisoning: bool, init: &mut dyn FnMut(&public::OnceState)) { + let mut state_and_queue = self.state_and_queue.load(Ordering::Acquire); + loop { + match state_and_queue.addr() { + COMPLETE => break, + POISONED if !ignore_poisoning => { + // Panic to propagate the poison. + panic!("Once instance has previously been poisoned"); + } + POISONED | INCOMPLETE => { + // Try to register this thread as the one RUNNING. + let exchange_result = self.state_and_queue.compare_exchange( + state_and_queue, + ptr::invalid_mut(RUNNING), + Ordering::Acquire, + Ordering::Acquire, + ); + if let Err(old) = exchange_result { + state_and_queue = old; + continue; + } + // `waiter_queue` will manage other waiting threads, and + // wake them up on drop. + let mut waiter_queue = WaiterQueue { + state_and_queue: &self.state_and_queue, + set_state_on_drop_to: ptr::invalid_mut(POISONED), + }; + // Run the initialization function, letting it know if we're + // poisoned or not. + let init_state = public::OnceState { + inner: OnceState { + poisoned: state_and_queue.addr() == POISONED, + set_state_on_drop_to: Cell::new(ptr::invalid_mut(COMPLETE)), + }, + }; + init(&init_state); + waiter_queue.set_state_on_drop_to = init_state.inner.set_state_on_drop_to.get(); + break; + } + _ => { + // All other values must be RUNNING with possibly a + // pointer to the waiter queue in the more significant bits. + assert!(state_and_queue.addr() & STATE_MASK == RUNNING); + wait(&self.state_and_queue, state_and_queue); + state_and_queue = self.state_and_queue.load(Ordering::Acquire); + } + } + } + } +} + +fn wait(state_and_queue: &AtomicPtr<Masked>, mut current_state: *mut Masked) { + // Note: the following code was carefully written to avoid creating a + // mutable reference to `node` that gets aliased. + loop { + // Don't queue this thread if the status is no longer running, + // otherwise we will not be woken up. + if current_state.addr() & STATE_MASK != RUNNING { + return; + } + + // Create the node for our current thread. + let node = Waiter { + thread: Cell::new(Some(thread::current())), + signaled: AtomicBool::new(false), + next: current_state.with_addr(current_state.addr() & !STATE_MASK) as *const Waiter, + }; + let me = &node as *const Waiter as *const Masked as *mut Masked; + + // Try to slide in the node at the head of the linked list, making sure + // that another thread didn't just replace the head of the linked list. + let exchange_result = state_and_queue.compare_exchange( + current_state, + me.with_addr(me.addr() | RUNNING), + Ordering::Release, + Ordering::Relaxed, + ); + if let Err(old) = exchange_result { + current_state = old; + continue; + } + + // We have enqueued ourselves, now lets wait. + // It is important not to return before being signaled, otherwise we + // would drop our `Waiter` node and leave a hole in the linked list + // (and a dangling reference). Guard against spurious wakeups by + // reparking ourselves until we are signaled. + while !node.signaled.load(Ordering::Acquire) { + // If the managing thread happens to signal and unpark us before we + // can park ourselves, the result could be this thread never gets + // unparked. Luckily `park` comes with the guarantee that if it got + // an `unpark` just before on an unparked thread it does not park. + thread::park(); + } + break; + } +} + +#[stable(feature = "std_debug", since = "1.16.0")] +impl fmt::Debug for Once { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("Once").finish_non_exhaustive() + } +} + +impl Drop for WaiterQueue<'_> { + fn drop(&mut self) { + // Swap out our state with however we finished. + let state_and_queue = + self.state_and_queue.swap(self.set_state_on_drop_to, Ordering::AcqRel); + + // We should only ever see an old state which was RUNNING. + assert_eq!(state_and_queue.addr() & STATE_MASK, RUNNING); + + // Walk the entire linked list of waiters and wake them up (in lifo + // order, last to register is first to wake up). + unsafe { + // Right after setting `node.signaled = true` the other thread may + // free `node` if there happens to be has a spurious wakeup. + // So we have to take out the `thread` field and copy the pointer to + // `next` first. + let mut queue = + state_and_queue.with_addr(state_and_queue.addr() & !STATE_MASK) as *const Waiter; + while !queue.is_null() { + let next = (*queue).next; + let thread = (*queue).thread.take().unwrap(); + (*queue).signaled.store(true, Ordering::Release); + // ^- FIXME (maybe): This is another case of issue #55005 + // `store()` has a potentially dangling ref to `signaled`. + queue = next; + thread.unpark(); + } + } + } +} + +impl OnceState { + #[inline] + pub fn is_poisoned(&self) -> bool { + self.poisoned + } + + #[inline] + pub fn poison(&self) { + self.set_state_on_drop_to.set(ptr::invalid_mut(POISONED)); + } +} diff --git a/library/std/src/sys_common/once/mod.rs b/library/std/src/sys_common/once/mod.rs new file mode 100644 index 000000000..8742e68cc --- /dev/null +++ b/library/std/src/sys_common/once/mod.rs @@ -0,0 +1,43 @@ +// A "once" is a relatively simple primitive, and it's also typically provided +// by the OS as well (see `pthread_once` or `InitOnceExecuteOnce`). The OS +// primitives, however, tend to have surprising restrictions, such as the Unix +// one doesn't allow an argument to be passed to the function. +// +// As a result, we end up implementing it ourselves in the standard library. +// This also gives us the opportunity to optimize the implementation a bit which +// should help the fast path on call sites. +// +// So to recap, the guarantees of a Once are that it will call the +// initialization closure at most once, and it will never return until the one +// that's running has finished running. This means that we need some form of +// blocking here while the custom callback is running at the very least. +// Additionally, we add on the restriction of **poisoning**. Whenever an +// initialization closure panics, the Once enters a "poisoned" state which means +// that all future calls will immediately panic as well. +// +// So to implement this, one might first reach for a `Mutex`, but those cannot +// be put into a `static`. It also gets a lot harder with poisoning to figure +// out when the mutex needs to be deallocated because it's not after the closure +// finishes, but after the first successful closure finishes. +// +// All in all, this is instead implemented with atomics and lock-free +// operations! Whee! + +cfg_if::cfg_if! { + if #[cfg(any( + target_os = "linux", + target_os = "android", + all(target_arch = "wasm32", target_feature = "atomics"), + target_os = "freebsd", + target_os = "openbsd", + target_os = "dragonfly", + target_os = "fuchsia", + target_os = "hermit", + ))] { + mod futex; + pub use futex::{Once, OnceState}; + } else { + mod generic; + pub use generic::{Once, OnceState}; + } +} diff --git a/library/std/src/sys_common/remutex.rs b/library/std/src/sys_common/remutex.rs index 8921af311..b448ae3a9 100644 --- a/library/std/src/sys_common/remutex.rs +++ b/library/std/src/sys_common/remutex.rs @@ -1,13 +1,11 @@ #[cfg(all(test, not(target_os = "emscripten")))] mod tests; +use super::mutex as sys; use crate::cell::UnsafeCell; -use crate::marker::PhantomPinned; use crate::ops::Deref; use crate::panic::{RefUnwindSafe, UnwindSafe}; -use crate::pin::Pin; use crate::sync::atomic::{AtomicUsize, Ordering::Relaxed}; -use crate::sys::locks as sys; /// A re-entrant mutual exclusion /// @@ -41,11 +39,10 @@ use crate::sys::locks as sys; /// synchronization is left to the mutex, making relaxed memory ordering for /// the `owner` field fine in all cases. pub struct ReentrantMutex<T> { - mutex: sys::Mutex, + mutex: sys::MovableMutex, owner: AtomicUsize, lock_count: UnsafeCell<u32>, data: T, - _pinned: PhantomPinned, } unsafe impl<T: Send> Send for ReentrantMutex<T> {} @@ -68,39 +65,22 @@ impl<T> RefUnwindSafe for ReentrantMutex<T> {} /// guarded data. #[must_use = "if unused the ReentrantMutex will immediately unlock"] pub struct ReentrantMutexGuard<'a, T: 'a> { - lock: Pin<&'a ReentrantMutex<T>>, + lock: &'a ReentrantMutex<T>, } impl<T> !Send for ReentrantMutexGuard<'_, T> {} impl<T> ReentrantMutex<T> { /// Creates a new reentrant mutex in an unlocked state. - /// - /// # Unsafety - /// - /// This function is unsafe because it is required that `init` is called - /// once this mutex is in its final resting place, and only then are the - /// lock/unlock methods safe. - pub const unsafe fn new(t: T) -> ReentrantMutex<T> { + pub const fn new(t: T) -> ReentrantMutex<T> { ReentrantMutex { - mutex: sys::Mutex::new(), + mutex: sys::MovableMutex::new(), owner: AtomicUsize::new(0), lock_count: UnsafeCell::new(0), data: t, - _pinned: PhantomPinned, } } - /// Initializes this mutex so it's ready for use. - /// - /// # Unsafety - /// - /// Unsafe to call more than once, and must be called after this will no - /// longer move in memory. - pub unsafe fn init(self: Pin<&mut Self>) { - self.get_unchecked_mut().mutex.init() - } - /// Acquires a mutex, blocking the current thread until it is able to do so. /// /// This function will block the caller until it is available to acquire the mutex. @@ -113,15 +93,14 @@ impl<T> ReentrantMutex<T> { /// If another user of this mutex panicked while holding the mutex, then /// this call will return failure if the mutex would otherwise be /// acquired. - pub fn lock(self: Pin<&Self>) -> ReentrantMutexGuard<'_, T> { + pub fn lock(&self) -> ReentrantMutexGuard<'_, T> { let this_thread = current_thread_unique_ptr(); - // Safety: We only touch lock_count when we own the lock, - // and since self is pinned we can safely call the lock() on the mutex. + // Safety: We only touch lock_count when we own the lock. unsafe { if self.owner.load(Relaxed) == this_thread { self.increment_lock_count(); } else { - self.mutex.lock(); + self.mutex.raw_lock(); self.owner.store(this_thread, Relaxed); debug_assert_eq!(*self.lock_count.get(), 0); *self.lock_count.get() = 1; @@ -142,10 +121,9 @@ impl<T> ReentrantMutex<T> { /// If another user of this mutex panicked while holding the mutex, then /// this call will return failure if the mutex would otherwise be /// acquired. - pub fn try_lock(self: Pin<&Self>) -> Option<ReentrantMutexGuard<'_, T>> { + pub fn try_lock(&self) -> Option<ReentrantMutexGuard<'_, T>> { let this_thread = current_thread_unique_ptr(); - // Safety: We only touch lock_count when we own the lock, - // and since self is pinned we can safely call the try_lock on the mutex. + // Safety: We only touch lock_count when we own the lock. unsafe { if self.owner.load(Relaxed) == this_thread { self.increment_lock_count(); @@ -179,12 +157,12 @@ impl<T> Deref for ReentrantMutexGuard<'_, T> { impl<T> Drop for ReentrantMutexGuard<'_, T> { #[inline] fn drop(&mut self) { - // Safety: We own the lock, and the lock is pinned. + // Safety: We own the lock. unsafe { *self.lock.lock_count.get() -= 1; if *self.lock.lock_count.get() == 0 { self.lock.owner.store(0, Relaxed); - self.lock.mutex.unlock(); + self.lock.mutex.raw_unlock(); } } } diff --git a/library/std/src/sys_common/remutex/tests.rs b/library/std/src/sys_common/remutex/tests.rs index 64873b850..8e97ce11c 100644 --- a/library/std/src/sys_common/remutex/tests.rs +++ b/library/std/src/sys_common/remutex/tests.rs @@ -1,18 +1,11 @@ -use crate::boxed::Box; use crate::cell::RefCell; -use crate::pin::Pin; use crate::sync::Arc; use crate::sys_common::remutex::{ReentrantMutex, ReentrantMutexGuard}; use crate::thread; #[test] fn smoke() { - let m = unsafe { - let mut m = Box::pin(ReentrantMutex::new(())); - m.as_mut().init(); - m - }; - let m = m.as_ref(); + let m = ReentrantMutex::new(()); { let a = m.lock(); { @@ -29,20 +22,15 @@ fn smoke() { #[test] fn is_mutex() { - let m = unsafe { - // FIXME: Simplify this if Arc gets an Arc::get_pin_mut. - let mut m = Arc::new(ReentrantMutex::new(RefCell::new(0))); - Pin::new_unchecked(Arc::get_mut_unchecked(&mut m)).init(); - Pin::new_unchecked(m) - }; + let m = Arc::new(ReentrantMutex::new(RefCell::new(0))); let m2 = m.clone(); - let lock = m.as_ref().lock(); + let lock = m.lock(); let child = thread::spawn(move || { - let lock = m2.as_ref().lock(); + let lock = m2.lock(); assert_eq!(*lock.borrow(), 4950); }); for i in 0..100 { - let lock = m.as_ref().lock(); + let lock = m.lock(); *lock.borrow_mut() += i; } drop(lock); @@ -51,22 +39,17 @@ fn is_mutex() { #[test] fn trylock_works() { - let m = unsafe { - // FIXME: Simplify this if Arc gets an Arc::get_pin_mut. - let mut m = Arc::new(ReentrantMutex::new(())); - Pin::new_unchecked(Arc::get_mut_unchecked(&mut m)).init(); - Pin::new_unchecked(m) - }; + let m = Arc::new(ReentrantMutex::new(())); let m2 = m.clone(); - let _lock = m.as_ref().try_lock(); - let _lock2 = m.as_ref().try_lock(); + let _lock = m.try_lock(); + let _lock2 = m.try_lock(); thread::spawn(move || { - let lock = m2.as_ref().try_lock(); + let lock = m2.try_lock(); assert!(lock.is_none()); }) .join() .unwrap(); - let _lock3 = m.as_ref().try_lock(); + let _lock3 = m.try_lock(); } pub struct Answer<'a>(pub ReentrantMutexGuard<'a, RefCell<u32>>); diff --git a/library/std/src/sys_common/rwlock.rs b/library/std/src/sys_common/rwlock.rs index ba56f3a8f..042981dac 100644 --- a/library/std/src/sys_common/rwlock.rs +++ b/library/std/src/sys_common/rwlock.rs @@ -1,65 +1,5 @@ use crate::sys::locks as imp; -/// An OS-based reader-writer lock, meant for use in static variables. -/// -/// This rwlock does not implement poisoning. -/// -/// This rwlock has a const constructor ([`StaticRwLock::new`]), does not -/// implement `Drop` to cleanup resources. -pub struct StaticRwLock(imp::RwLock); - -impl StaticRwLock { - /// Creates a new rwlock for use. - #[inline] - pub const fn new() -> Self { - Self(imp::RwLock::new()) - } - - /// Acquires shared access to the underlying lock, blocking the current - /// thread to do so. - /// - /// The lock is automatically unlocked when the returned guard is dropped. - #[inline] - pub fn read(&'static self) -> StaticRwLockReadGuard { - unsafe { self.0.read() }; - StaticRwLockReadGuard(&self.0) - } - - /// Acquires write access to the underlying lock, blocking the current thread - /// to do so. - /// - /// The lock is automatically unlocked when the returned guard is dropped. - #[inline] - pub fn write(&'static self) -> StaticRwLockWriteGuard { - unsafe { self.0.write() }; - StaticRwLockWriteGuard(&self.0) - } -} - -#[must_use] -pub struct StaticRwLockReadGuard(&'static imp::RwLock); - -impl Drop for StaticRwLockReadGuard { - #[inline] - fn drop(&mut self) { - unsafe { - self.0.read_unlock(); - } - } -} - -#[must_use] -pub struct StaticRwLockWriteGuard(&'static imp::RwLock); - -impl Drop for StaticRwLockWriteGuard { - #[inline] - fn drop(&mut self) { - unsafe { - self.0.write_unlock(); - } - } -} - /// An OS-based reader-writer lock. /// /// This rwlock cleans up its resources in its `Drop` implementation and may @@ -75,6 +15,7 @@ pub struct MovableRwLock(imp::MovableRwLock); impl MovableRwLock { /// Creates a new reader-writer lock for use. #[inline] + #[rustc_const_stable(feature = "const_locks", since = "1.63.0")] pub const fn new() -> Self { Self(imp::MovableRwLock::new()) } diff --git a/library/std/src/sys_common/thread_local_key.rs b/library/std/src/sys_common/thread_local_key.rs index 70beebe86..747579f17 100644 --- a/library/std/src/sys_common/thread_local_key.rs +++ b/library/std/src/sys_common/thread_local_key.rs @@ -53,7 +53,6 @@ mod tests; use crate::sync::atomic::{self, AtomicUsize, Ordering}; use crate::sys::thread_local_key as imp; -use crate::sys_common::mutex::StaticMutex; /// A type for TLS keys that are statically allocated. /// @@ -69,8 +68,10 @@ use crate::sys_common::mutex::StaticMutex; /// ```ignore (cannot-doctest-private-modules) /// use tls::os::{StaticKey, INIT}; /// +/// // Use a regular global static to store the key. /// static KEY: StaticKey = INIT; /// +/// // The state provided via `get` and `set` is thread-local. /// unsafe { /// assert!(KEY.get().is_null()); /// KEY.set(1 as *mut u8); @@ -149,25 +150,6 @@ impl StaticKey { } unsafe fn lazy_init(&self) -> usize { - // Currently the Windows implementation of TLS is pretty hairy, and - // it greatly simplifies creation if we just synchronize everything. - // - // Additionally a 0-index of a tls key hasn't been seen on windows, so - // we just simplify the whole branch. - if imp::requires_synchronized_create() { - // We never call `INIT_LOCK.init()`, so it is UB to attempt to - // acquire this mutex reentrantly! - static INIT_LOCK: StaticMutex = StaticMutex::new(); - let _guard = INIT_LOCK.lock(); - let mut key = self.key.load(Ordering::SeqCst); - if key == 0 { - key = imp::create(self.dtor) as usize; - self.key.store(key, Ordering::SeqCst); - } - rtassert!(key != 0); - return key; - } - // POSIX allows the key created here to be 0, but the compare_exchange // below relies on using 0 as a sentinel value to check who won the // race to set the shared TLS key. As far as I know, there is no @@ -230,8 +212,6 @@ impl Key { impl Drop for Key { fn drop(&mut self) { - // Right now Windows doesn't support TLS key destruction, but this also - // isn't used anywhere other than tests, so just leak the TLS key. - // unsafe { imp::destroy(self.key) } + unsafe { imp::destroy(self.key) } } } diff --git a/library/std/src/sys_common/thread_local_key/tests.rs b/library/std/src/sys_common/thread_local_key/tests.rs index 968738a41..6f32b858f 100644 --- a/library/std/src/sys_common/thread_local_key/tests.rs +++ b/library/std/src/sys_common/thread_local_key/tests.rs @@ -1,4 +1,5 @@ use super::{Key, StaticKey}; +use core::ptr; fn assert_sync<T: Sync>() {} fn assert_send<T: Send>() {} @@ -12,8 +13,8 @@ fn smoke() { let k2 = Key::new(None); assert!(k1.get().is_null()); assert!(k2.get().is_null()); - k1.set(1 as *mut _); - k2.set(2 as *mut _); + k1.set(ptr::invalid_mut(1)); + k2.set(ptr::invalid_mut(2)); assert_eq!(k1.get() as usize, 1); assert_eq!(k2.get() as usize, 2); } @@ -26,8 +27,8 @@ fn statik() { unsafe { assert!(K1.get().is_null()); assert!(K2.get().is_null()); - K1.set(1 as *mut _); - K2.set(2 as *mut _); + K1.set(ptr::invalid_mut(1)); + K2.set(ptr::invalid_mut(2)); assert_eq!(K1.get() as usize, 1); assert_eq!(K2.get() as usize, 2); } diff --git a/library/std/src/sys_common/thread_parker/mod.rs b/library/std/src/sys_common/thread_parker/mod.rs index cbd7832eb..f86a9a555 100644 --- a/library/std/src/sys_common/thread_parker/mod.rs +++ b/library/std/src/sys_common/thread_parker/mod.rs @@ -7,6 +7,7 @@ cfg_if::cfg_if! { target_os = "openbsd", target_os = "dragonfly", target_os = "fuchsia", + target_os = "hermit", ))] { mod futex; pub use futex::Parker; diff --git a/library/std/src/sys_common/wtf8.rs b/library/std/src/sys_common/wtf8.rs index 57fa49893..dd53767d4 100644 --- a/library/std/src/sys_common/wtf8.rs +++ b/library/std/src/sys_common/wtf8.rs @@ -89,6 +89,24 @@ impl CodePoint { self.value } + /// Returns the numeric value of the code point if it is a leading surrogate. + #[inline] + pub fn to_lead_surrogate(&self) -> Option<u16> { + match self.value { + lead @ 0xD800..=0xDBFF => Some(lead as u16), + _ => None, + } + } + + /// Returns the numeric value of the code point if it is a trailing surrogate. + #[inline] + pub fn to_trail_surrogate(&self) -> Option<u16> { + match self.value { + trail @ 0xDC00..=0xDFFF => Some(trail as u16), + _ => None, + } + } + /// Optionally returns a Unicode scalar value for the code point. /// /// Returns `None` if the code point is a surrogate (from U+D800 to U+DFFF). @@ -117,6 +135,14 @@ impl CodePoint { #[derive(Eq, PartialEq, Ord, PartialOrd, Clone)] pub struct Wtf8Buf { bytes: Vec<u8>, + + /// Do we know that `bytes` holds a valid UTF-8 encoding? We can easily + /// know this if we're constructed from a `String` or `&str`. + /// + /// It is possible for `bytes` to have valid UTF-8 without this being + /// set, such as when we're concatenating `&Wtf8`'s and surrogates become + /// paired, as we don't bother to rescan the entire string. + is_known_utf8: bool, } impl ops::Deref for Wtf8Buf { @@ -147,13 +173,13 @@ impl Wtf8Buf { /// Creates a new, empty WTF-8 string. #[inline] pub fn new() -> Wtf8Buf { - Wtf8Buf { bytes: Vec::new() } + Wtf8Buf { bytes: Vec::new(), is_known_utf8: true } } /// Creates a new, empty WTF-8 string with pre-allocated capacity for `capacity` bytes. #[inline] pub fn with_capacity(capacity: usize) -> Wtf8Buf { - Wtf8Buf { bytes: Vec::with_capacity(capacity) } + Wtf8Buf { bytes: Vec::with_capacity(capacity), is_known_utf8: true } } /// Creates a WTF-8 string from a UTF-8 `String`. @@ -163,7 +189,7 @@ impl Wtf8Buf { /// Since WTF-8 is a superset of UTF-8, this always succeeds. #[inline] pub fn from_string(string: String) -> Wtf8Buf { - Wtf8Buf { bytes: string.into_bytes() } + Wtf8Buf { bytes: string.into_bytes(), is_known_utf8: true } } /// Creates a WTF-8 string from a UTF-8 `&str` slice. @@ -173,11 +199,12 @@ impl Wtf8Buf { /// Since WTF-8 is a superset of UTF-8, this always succeeds. #[inline] pub fn from_str(str: &str) -> Wtf8Buf { - Wtf8Buf { bytes: <[_]>::to_vec(str.as_bytes()) } + Wtf8Buf { bytes: <[_]>::to_vec(str.as_bytes()), is_known_utf8: true } } pub fn clear(&mut self) { - self.bytes.clear() + self.bytes.clear(); + self.is_known_utf8 = true; } /// Creates a WTF-8 string from a potentially ill-formed UTF-16 slice of 16-bit code units. @@ -193,9 +220,11 @@ impl Wtf8Buf { let surrogate = surrogate.unpaired_surrogate(); // Surrogates are known to be in the code point range. let code_point = unsafe { CodePoint::from_u32_unchecked(surrogate as u32) }; + // The string will now contain an unpaired surrogate. + string.is_known_utf8 = false; // Skip the WTF-8 concatenation check, // surrogate pairs are already decoded by decode_utf16 - string.push_code_point_unchecked(code_point) + string.push_code_point_unchecked(code_point); } } } @@ -203,7 +232,7 @@ impl Wtf8Buf { } /// Copied from String::push - /// This does **not** include the WTF-8 concatenation check. + /// This does **not** include the WTF-8 concatenation check or `is_known_utf8` check. fn push_code_point_unchecked(&mut self, code_point: CodePoint) { let mut bytes = [0; 4]; let bytes = char::encode_utf8_raw(code_point.value, &mut bytes); @@ -217,6 +246,9 @@ impl Wtf8Buf { #[inline] pub fn as_mut_slice(&mut self) -> &mut Wtf8 { + // Safety: `Wtf8` doesn't expose any way to mutate the bytes that would + // cause them to change from well-formed UTF-8 to ill-formed UTF-8, + // which would break the assumptions of the `is_known_utf8` field. unsafe { Wtf8::from_mut_bytes_unchecked(&mut self.bytes) } } @@ -236,7 +268,8 @@ impl Wtf8Buf { /// in the given `Wtf8Buf`. The `Wtf8Buf` may reserve more space to avoid /// frequent reallocations. After calling `try_reserve`, capacity will be /// greater than or equal to `self.len() + additional`. Does nothing if - /// capacity is already sufficient. + /// capacity is already sufficient. This method preserves the contents even + /// if an error occurs. /// /// # Errors /// @@ -313,7 +346,15 @@ impl Wtf8Buf { self.push_char(decode_surrogate_pair(lead, trail)); self.bytes.extend_from_slice(other_without_trail_surrogate); } - _ => self.bytes.extend_from_slice(&other.bytes), + _ => { + // If we'll be pushing a string containing a surrogate, we may + // no longer have UTF-8. + if other.next_surrogate(0).is_some() { + self.is_known_utf8 = false; + } + + self.bytes.extend_from_slice(&other.bytes); + } } } @@ -330,13 +371,19 @@ impl Wtf8Buf { /// like concatenating ill-formed UTF-16 strings effectively would. #[inline] pub fn push(&mut self, code_point: CodePoint) { - if let trail @ 0xDC00..=0xDFFF = code_point.to_u32() { + if let Some(trail) = code_point.to_trail_surrogate() { if let Some(lead) = (&*self).final_lead_surrogate() { let len_without_lead_surrogate = self.len() - 3; self.bytes.truncate(len_without_lead_surrogate); - self.push_char(decode_surrogate_pair(lead, trail as u16)); + self.push_char(decode_surrogate_pair(lead, trail)); return; } + + // We're pushing a trailing surrogate. + self.is_known_utf8 = false; + } else if code_point.to_lead_surrogate().is_some() { + // We're pushing a leading surrogate. + self.is_known_utf8 = false; } // No newly paired surrogates at the boundary. @@ -363,9 +410,10 @@ impl Wtf8Buf { /// (that is, if the string contains surrogates), /// the original WTF-8 string is returned instead. pub fn into_string(self) -> Result<String, Wtf8Buf> { - match self.next_surrogate(0) { - None => Ok(unsafe { String::from_utf8_unchecked(self.bytes) }), - Some(_) => Err(self), + if self.is_known_utf8 || self.next_surrogate(0).is_none() { + Ok(unsafe { String::from_utf8_unchecked(self.bytes) }) + } else { + Err(self) } } @@ -375,6 +423,11 @@ impl Wtf8Buf { /// /// Surrogates are replaced with `"\u{FFFD}"` (the replacement character āļæ½ā) pub fn into_string_lossy(mut self) -> String { + // Fast path: If we already have UTF-8, we can return it immediately. + if self.is_known_utf8 { + return unsafe { String::from_utf8_unchecked(self.bytes) }; + } + let mut pos = 0; loop { match self.next_surrogate(pos) { @@ -397,7 +450,7 @@ impl Wtf8Buf { /// Converts a `Box<Wtf8>` into a `Wtf8Buf`. pub fn from_box(boxed: Box<Wtf8>) -> Wtf8Buf { let bytes: Box<[u8]> = unsafe { mem::transmute(boxed) }; - Wtf8Buf { bytes: bytes.into_vec() } + Wtf8Buf { bytes: bytes.into_vec(), is_known_utf8: false } } } @@ -575,6 +628,11 @@ impl Wtf8 { } } + /// Creates an owned `Wtf8Buf` from a borrowed `Wtf8`. + pub fn to_owned(&self) -> Wtf8Buf { + Wtf8Buf { bytes: self.bytes.to_vec(), is_known_utf8: false } + } + /// Lossily converts the string to UTF-8. /// Returns a UTF-8 `&str` slice if the contents are well-formed in UTF-8. /// @@ -664,7 +722,8 @@ impl Wtf8 { } pub fn clone_into(&self, buf: &mut Wtf8Buf) { - self.bytes.clone_into(&mut buf.bytes) + buf.is_known_utf8 = false; + self.bytes.clone_into(&mut buf.bytes); } /// Boxes this `Wtf8`. @@ -704,12 +763,12 @@ impl Wtf8 { #[inline] pub fn to_ascii_lowercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_lowercase() } + Wtf8Buf { bytes: self.bytes.to_ascii_lowercase(), is_known_utf8: false } } #[inline] pub fn to_ascii_uppercase(&self) -> Wtf8Buf { - Wtf8Buf { bytes: self.bytes.to_ascii_uppercase() } + Wtf8Buf { bytes: self.bytes.to_ascii_uppercase(), is_known_utf8: false } } #[inline] diff --git a/library/std/src/sys_common/wtf8/tests.rs b/library/std/src/sys_common/wtf8/tests.rs index 931996791..1a302d646 100644 --- a/library/std/src/sys_common/wtf8/tests.rs +++ b/library/std/src/sys_common/wtf8/tests.rs @@ -20,6 +20,36 @@ fn code_point_to_u32() { } #[test] +fn code_point_to_lead_surrogate() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0).to_lead_surrogate(), None); + assert_eq!(c(0xE9).to_lead_surrogate(), None); + assert_eq!(c(0xD800).to_lead_surrogate(), Some(0xD800)); + assert_eq!(c(0xDBFF).to_lead_surrogate(), Some(0xDBFF)); + assert_eq!(c(0xDC00).to_lead_surrogate(), None); + assert_eq!(c(0xDFFF).to_lead_surrogate(), None); + assert_eq!(c(0x1F4A9).to_lead_surrogate(), None); + assert_eq!(c(0x10FFFF).to_lead_surrogate(), None); +} + +#[test] +fn code_point_to_trail_surrogate() { + fn c(value: u32) -> CodePoint { + CodePoint::from_u32(value).unwrap() + } + assert_eq!(c(0).to_trail_surrogate(), None); + assert_eq!(c(0xE9).to_trail_surrogate(), None); + assert_eq!(c(0xD800).to_trail_surrogate(), None); + assert_eq!(c(0xDBFF).to_trail_surrogate(), None); + assert_eq!(c(0xDC00).to_trail_surrogate(), Some(0xDC00)); + assert_eq!(c(0xDFFF).to_trail_surrogate(), Some(0xDFFF)); + assert_eq!(c(0x1F4A9).to_trail_surrogate(), None); + assert_eq!(c(0x10FFFF).to_trail_surrogate(), None); +} + +#[test] fn code_point_from_char() { assert_eq!(CodePoint::from_char('a').to_u32(), 0x61); assert_eq!(CodePoint::from_char('š©').to_u32(), 0x1F4A9); @@ -70,35 +100,66 @@ fn wtf8buf_from_string() { #[test] fn wtf8buf_from_wide() { - assert_eq!(Wtf8Buf::from_wide(&[]).bytes, b""); - assert_eq!( - Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]).bytes, - b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9" - ); + let buf = Wtf8Buf::from_wide(&[]); + assert_eq!(buf.bytes, b""); + assert!(buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xDCA9]); + assert_eq!(buf.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0x61, 0xE9, 0x20, 0xD83D, 0xD83D, 0xDCA9]); + assert_eq!(buf.bytes, b"a\xC3\xA9 \xED\xA0\xBD\xF0\x9F\x92\xA9"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xD800]); + assert_eq!(buf.bytes, b"\xED\xA0\x80"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xDBFF]); + assert_eq!(buf.bytes, b"\xED\xAF\xBF"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xDC00]); + assert_eq!(buf.bytes, b"\xED\xB0\x80"); + assert!(!buf.is_known_utf8); + + let buf = Wtf8Buf::from_wide(&[0xDFFF]); + assert_eq!(buf.bytes, b"\xED\xBF\xBF"); + assert!(!buf.is_known_utf8); } #[test] fn wtf8buf_push_str() { let mut string = Wtf8Buf::new(); assert_eq!(string.bytes, b""); + assert!(string.is_known_utf8); + string.push_str("aĆ© š©"); assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); } #[test] fn wtf8buf_push_char() { let mut string = Wtf8Buf::from_str("aĆ© "); assert_eq!(string.bytes, b"a\xC3\xA9 "); + assert!(string.is_known_utf8); + string.push_char('š©'); assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); } #[test] fn wtf8buf_push() { let mut string = Wtf8Buf::from_str("aĆ© "); assert_eq!(string.bytes, b"a\xC3\xA9 "); + assert!(string.is_known_utf8); + string.push(CodePoint::from_char('š©')); assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); fn c(value: u32) -> CodePoint { CodePoint::from_u32(value).unwrap() @@ -106,37 +167,46 @@ fn wtf8buf_push() { let mut string = Wtf8Buf::new(); string.push(c(0xD83D)); // lead + assert!(!string.is_known_utf8); string.push(c(0xDCA9)); // trail assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); // Magic! let mut string = Wtf8Buf::new(); string.push(c(0xD83D)); // lead + assert!(!string.is_known_utf8); string.push(c(0x20)); // not surrogate string.push(c(0xDCA9)); // trail assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); let mut string = Wtf8Buf::new(); string.push(c(0xD800)); // lead + assert!(!string.is_known_utf8); string.push(c(0xDBFF)); // lead assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); let mut string = Wtf8Buf::new(); string.push(c(0xD800)); // lead + assert!(!string.is_known_utf8); string.push(c(0xE000)); // not surrogate assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); let mut string = Wtf8Buf::new(); string.push(c(0xD7FF)); // not surrogate + assert!(string.is_known_utf8); string.push(c(0xDC00)); // trail + assert!(!string.is_known_utf8); assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); let mut string = Wtf8Buf::new(); string.push(c(0x61)); // not surrogate, < 3 bytes + assert!(string.is_known_utf8); string.push(c(0xDC00)); // trail + assert!(!string.is_known_utf8); assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); let mut string = Wtf8Buf::new(); string.push(c(0xDC00)); // trail + assert!(!string.is_known_utf8); assert_eq!(string.bytes, b"\xED\xB0\x80"); } @@ -146,6 +216,7 @@ fn wtf8buf_push_wtf8() { assert_eq!(string.bytes, b"a\xC3\xA9"); string.push_wtf8(Wtf8::from_str(" š©")); assert_eq!(string.bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); fn w(v: &[u8]) -> &Wtf8 { unsafe { Wtf8::from_bytes_unchecked(v) } @@ -161,37 +232,68 @@ fn wtf8buf_push_wtf8() { string.push_wtf8(w(b" ")); // not surrogate string.push_wtf8(w(b"\xED\xB2\xA9")); // trail assert_eq!(string.bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); + assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xA0\x80")); // lead string.push_wtf8(w(b"\xED\xAF\xBF")); // lead assert_eq!(string.bytes, b"\xED\xA0\x80\xED\xAF\xBF"); + assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xA0\x80")); // lead string.push_wtf8(w(b"\xEE\x80\x80")); // not surrogate assert_eq!(string.bytes, b"\xED\xA0\x80\xEE\x80\x80"); + assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\x9F\xBF")); // not surrogate string.push_wtf8(w(b"\xED\xB0\x80")); // trail assert_eq!(string.bytes, b"\xED\x9F\xBF\xED\xB0\x80"); + assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"a")); // not surrogate, < 3 bytes string.push_wtf8(w(b"\xED\xB0\x80")); // trail assert_eq!(string.bytes, b"\x61\xED\xB0\x80"); + assert!(!string.is_known_utf8); let mut string = Wtf8Buf::new(); string.push_wtf8(w(b"\xED\xB0\x80")); // trail assert_eq!(string.bytes, b"\xED\xB0\x80"); + assert!(!string.is_known_utf8); } #[test] fn wtf8buf_truncate() { let mut string = Wtf8Buf::from_str("aĆ©"); + assert!(string.is_known_utf8); + + string.truncate(3); + assert_eq!(string.bytes, b"a\xC3\xA9"); + assert!(string.is_known_utf8); + string.truncate(1); assert_eq!(string.bytes, b"a"); + assert!(string.is_known_utf8); + + string.truncate(0); + assert_eq!(string.bytes, b""); + assert!(string.is_known_utf8); +} + +#[test] +fn wtf8buf_truncate_around_non_bmp() { + let mut string = Wtf8Buf::from_str("š©"); + assert!(string.is_known_utf8); + + string.truncate(4); + assert_eq!(string.bytes, b"\xF0\x9F\x92\xA9"); + assert!(string.is_known_utf8); + + string.truncate(0); + assert_eq!(string.bytes, b""); + assert!(string.is_known_utf8); } #[test] @@ -209,10 +311,36 @@ fn wtf8buf_truncate_fail_longer() { } #[test] +#[should_panic] +fn wtf8buf_truncate_splitting_non_bmp3() { + let mut string = Wtf8Buf::from_str("š©"); + assert!(string.is_known_utf8); + string.truncate(3); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_splitting_non_bmp2() { + let mut string = Wtf8Buf::from_str("š©"); + assert!(string.is_known_utf8); + string.truncate(2); +} + +#[test] +#[should_panic] +fn wtf8buf_truncate_splitting_non_bmp1() { + let mut string = Wtf8Buf::from_str("š©"); + assert!(string.is_known_utf8); + string.truncate(1); +} + +#[test] fn wtf8buf_into_string() { let mut string = Wtf8Buf::from_str("aĆ© š©"); + assert!(string.is_known_utf8); assert_eq!(string.clone().into_string(), Ok(String::from("aĆ© š©"))); string.push(CodePoint::from_u32(0xD800).unwrap()); + assert!(!string.is_known_utf8); assert_eq!(string.clone().into_string(), Err(string)); } @@ -229,15 +357,33 @@ fn wtf8buf_from_iterator() { fn f(values: &[u32]) -> Wtf8Buf { values.iter().map(|&c| CodePoint::from_u32(c).unwrap()).collect::<Wtf8Buf>() } - assert_eq!(f(&[0x61, 0xE9, 0x20, 0x1F4A9]).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!( + f(&[0x61, 0xE9, 0x20, 0x1F4A9]), + Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } + ); assert_eq!(f(&[0xD83D, 0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! - assert_eq!(f(&[0xD83D, 0x20, 0xDCA9]).bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); - assert_eq!(f(&[0xD800, 0xDBFF]).bytes, b"\xED\xA0\x80\xED\xAF\xBF"); - assert_eq!(f(&[0xD800, 0xE000]).bytes, b"\xED\xA0\x80\xEE\x80\x80"); - assert_eq!(f(&[0xD7FF, 0xDC00]).bytes, b"\xED\x9F\xBF\xED\xB0\x80"); - assert_eq!(f(&[0x61, 0xDC00]).bytes, b"\x61\xED\xB0\x80"); - assert_eq!(f(&[0xDC00]).bytes, b"\xED\xB0\x80"); + assert_eq!( + f(&[0xD83D, 0x20, 0xDCA9]), + Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0xD800, 0xDBFF]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xED\xAF\xBF".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0xD800, 0xE000]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xEE\x80\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0xD7FF, 0xDC00]), + Wtf8Buf { bytes: b"\xED\x9F\xBF\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + f(&[0x61, 0xDC00]), + Wtf8Buf { bytes: b"\x61\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!(f(&[0xDC00]), Wtf8Buf { bytes: b"\xED\xB0\x80".to_vec(), is_known_utf8: false }); } #[test] @@ -251,15 +397,36 @@ fn wtf8buf_extend() { string } - assert_eq!(e(&[0x61, 0xE9], &[0x20, 0x1F4A9]).bytes, b"a\xC3\xA9 \xF0\x9F\x92\xA9"); + assert_eq!( + e(&[0x61, 0xE9], &[0x20, 0x1F4A9]), + Wtf8Buf { bytes: b"a\xC3\xA9 \xF0\x9F\x92\xA9".to_vec(), is_known_utf8: true } + ); assert_eq!(e(&[0xD83D], &[0xDCA9]).bytes, b"\xF0\x9F\x92\xA9"); // Magic! - assert_eq!(e(&[0xD83D, 0x20], &[0xDCA9]).bytes, b"\xED\xA0\xBD \xED\xB2\xA9"); - assert_eq!(e(&[0xD800], &[0xDBFF]).bytes, b"\xED\xA0\x80\xED\xAF\xBF"); - assert_eq!(e(&[0xD800], &[0xE000]).bytes, b"\xED\xA0\x80\xEE\x80\x80"); - assert_eq!(e(&[0xD7FF], &[0xDC00]).bytes, b"\xED\x9F\xBF\xED\xB0\x80"); - assert_eq!(e(&[0x61], &[0xDC00]).bytes, b"\x61\xED\xB0\x80"); - assert_eq!(e(&[], &[0xDC00]).bytes, b"\xED\xB0\x80"); + assert_eq!( + e(&[0xD83D, 0x20], &[0xDCA9]), + Wtf8Buf { bytes: b"\xED\xA0\xBD \xED\xB2\xA9".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0xD800], &[0xDBFF]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xED\xAF\xBF".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0xD800], &[0xE000]), + Wtf8Buf { bytes: b"\xED\xA0\x80\xEE\x80\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0xD7FF], &[0xDC00]), + Wtf8Buf { bytes: b"\xED\x9F\xBF\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[0x61], &[0xDC00]), + Wtf8Buf { bytes: b"\x61\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); + assert_eq!( + e(&[], &[0xDC00]), + Wtf8Buf { bytes: b"\xED\xB0\x80".to_vec(), is_known_utf8: false } + ); } #[test] @@ -407,3 +574,93 @@ fn wtf8_encode_wide_size_hint() { assert_eq!((0, Some(0)), iter.size_hint()); assert!(iter.next().is_none()); } + +#[test] +fn wtf8_clone_into() { + let mut string = Wtf8Buf::new(); + Wtf8::from_str("green").clone_into(&mut string); + assert_eq!(string.bytes, b"green"); + + let mut string = Wtf8Buf::from_str("green"); + Wtf8::from_str("").clone_into(&mut string); + assert_eq!(string.bytes, b""); + + let mut string = Wtf8Buf::from_str("red"); + Wtf8::from_str("green").clone_into(&mut string); + assert_eq!(string.bytes, b"green"); + + let mut string = Wtf8Buf::from_str("green"); + Wtf8::from_str("red").clone_into(&mut string); + assert_eq!(string.bytes, b"red"); + + let mut string = Wtf8Buf::from_str("green"); + assert!(string.is_known_utf8); + unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").clone_into(&mut string) }; + assert_eq!(string.bytes, b"\xED\xA0\x80"); + assert!(!string.is_known_utf8); +} + +#[test] +fn wtf8_to_ascii_lowercase() { + let lowercase = Wtf8::from_str("").to_ascii_lowercase(); + assert_eq!(lowercase.bytes, b""); + + let lowercase = Wtf8::from_str("GrEeN gRaPeS! š").to_ascii_lowercase(); + assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); + + let lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_lowercase() }; + assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); + assert!(!lowercase.is_known_utf8); +} + +#[test] +fn wtf8_to_ascii_uppercase() { + let uppercase = Wtf8::from_str("").to_ascii_uppercase(); + assert_eq!(uppercase.bytes, b""); + + let uppercase = Wtf8::from_str("GrEeN gRaPeS! š").to_ascii_uppercase(); + assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); + + let uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_ascii_uppercase() }; + assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); + assert!(!uppercase.is_known_utf8); +} + +#[test] +fn wtf8_make_ascii_lowercase() { + let mut lowercase = Wtf8Buf::from_str(""); + lowercase.make_ascii_lowercase(); + assert_eq!(lowercase.bytes, b""); + + let mut lowercase = Wtf8Buf::from_str("GrEeN gRaPeS! š"); + lowercase.make_ascii_lowercase(); + assert_eq!(lowercase.bytes, b"green grapes! \xf0\x9f\x8d\x87"); + + let mut lowercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + lowercase.make_ascii_lowercase(); + assert_eq!(lowercase.bytes, b"\xED\xA0\x80"); + assert!(!lowercase.is_known_utf8); +} + +#[test] +fn wtf8_make_ascii_uppercase() { + let mut uppercase = Wtf8Buf::from_str(""); + uppercase.make_ascii_uppercase(); + assert_eq!(uppercase.bytes, b""); + + let mut uppercase = Wtf8Buf::from_str("GrEeN gRaPeS! š"); + uppercase.make_ascii_uppercase(); + assert_eq!(uppercase.bytes, b"GREEN GRAPES! \xf0\x9f\x8d\x87"); + + let mut uppercase = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + uppercase.make_ascii_uppercase(); + assert_eq!(uppercase.bytes, b"\xED\xA0\x80"); + assert!(!uppercase.is_known_utf8); +} + +#[test] +fn wtf8_to_owned() { + let string = unsafe { Wtf8::from_bytes_unchecked(b"\xED\xA0\x80").to_owned() }; + assert_eq!(string.bytes, b"\xED\xA0\x80"); + assert!(!string.is_known_utf8); +} diff --git a/library/std/src/thread/local.rs b/library/std/src/thread/local.rs index f4750cdf7..5d267891b 100644 --- a/library/std/src/thread/local.rs +++ b/library/std/src/thread/local.rs @@ -95,6 +95,7 @@ use crate::fmt; /// [loader lock]: https://docs.microsoft.com/en-us/windows/win32/dlls/dynamic-link-library-best-practices /// [`JoinHandle::join`]: crate::thread::JoinHandle::join /// [`with`]: LocalKey::with +#[cfg_attr(not(test), rustc_diagnostic_item = "LocalKey")] #[stable(feature = "rust1", since = "1.0.0")] pub struct LocalKey<T: 'static> { // This outer `LocalKey<T>` type is what's going to be stored in statics, @@ -900,7 +901,7 @@ pub mod statik { } #[doc(hidden)] -#[cfg(target_thread_local)] +#[cfg(all(target_thread_local, not(all(target_family = "wasm", not(target_feature = "atomics"))),))] pub mod fast { use super::lazy::LazyKeyInner; use crate::cell::Cell; @@ -1036,6 +1037,10 @@ pub mod fast { } #[doc(hidden)] +#[cfg(all( + not(target_thread_local), + not(all(target_family = "wasm", not(target_feature = "atomics"))), +))] pub mod os { use super::lazy::LazyKeyInner; use crate::cell::Cell; @@ -1044,6 +1049,8 @@ pub mod os { use crate::ptr; use crate::sys_common::thread_local_key::StaticKey as OsStaticKey; + /// Use a regular global static to store this key; the state provided will then be + /// thread-local. pub struct Key<T> { // OS-TLS key that we'll use to key off. os: OsStaticKey, diff --git a/library/std/src/thread/mod.rs b/library/std/src/thread/mod.rs index 44c8a50fd..05023df1b 100644 --- a/library/std/src/thread/mod.rs +++ b/library/std/src/thread/mod.rs @@ -116,7 +116,7 @@ //! Threads are able to have associated names for identification purposes. By default, spawned //! threads are unnamed. To specify a name for a thread, build the thread with [`Builder`] and pass //! the desired thread name to [`Builder::name`]. To retrieve the thread name from within the -//! thread, use [`Thread::name`]. A couple examples of where the name of a thread gets used: +//! thread, use [`Thread::name`]. A couple of examples where the name of a thread gets used: //! //! * If a panic occurs in a named thread, the thread name will be printed in the panic message. //! * The thread name is provided to the OS where applicable (e.g., `pthread_setname_np` in @@ -150,6 +150,8 @@ #![stable(feature = "rust1", since = "1.0.0")] #![deny(unsafe_op_in_unsafe_fn)] +// Under `test`, `__FastLocalKeyInner` seems unused. +#![cfg_attr(test, allow(dead_code))] #[cfg(all(test, not(target_os = "emscripten")))] mod tests; @@ -160,7 +162,7 @@ use crate::ffi::{CStr, CString}; use crate::fmt; use crate::io; use crate::marker::PhantomData; -use crate::mem; +use crate::mem::{self, forget}; use crate::num::NonZeroU64; use crate::num::NonZeroUsize; use crate::panic; @@ -170,7 +172,6 @@ use crate::ptr::addr_of_mut; use crate::str; use crate::sync::Arc; use crate::sys::thread as imp; -use crate::sys_common::mutex; use crate::sys_common::thread; use crate::sys_common::thread_info; use crate::sys_common::thread_parker::Parker; @@ -193,22 +194,40 @@ pub use scoped::{scope, Scope, ScopedJoinHandle}; #[stable(feature = "rust1", since = "1.0.0")] pub use self::local::{AccessError, LocalKey}; -// The types used by the thread_local! macro to access TLS keys. Note that there -// are two types, the "OS" type and the "fast" type. The OS thread local key -// type is accessed via platform-specific API calls and is slow, while the fast +// Provide the type used by the thread_local! macro to access TLS keys. This +// needs to be kept in sync with the macro itself (in `local.rs`). +// There are three types: "static", "fast", "OS". The "OS" thread local key +// type is accessed via platform-specific API calls and is slow, while the "fast" // key type is accessed via code generated via LLVM, where TLS keys are set up -// by the elf linker. Note that the OS TLS type is always available: on macOS -// the standard library is compiled with support for older platform versions -// where fast TLS was not available; end-user code is compiled with fast TLS -// where available, but both are needed. +// by the elf linker. "static" is for single-threaded platforms where a global +// static is sufficient. #[unstable(feature = "libstd_thread_internals", issue = "none")] -#[cfg(target_thread_local)] +#[cfg(not(test))] +#[cfg(all( + target_thread_local, + not(all(target_family = "wasm", not(target_feature = "atomics"))), +))] #[doc(hidden)] pub use self::local::fast::Key as __FastLocalKeyInner; + +// when building for tests, use real std's type +#[unstable(feature = "libstd_thread_internals", issue = "none")] +#[cfg(test)] +#[cfg(all( + target_thread_local, + not(all(target_family = "wasm", not(target_feature = "atomics"))), +))] +pub use realstd::thread::__FastLocalKeyInner; + #[unstable(feature = "libstd_thread_internals", issue = "none")] +#[cfg(all( + not(target_thread_local), + not(all(target_family = "wasm", not(target_feature = "atomics"))), +))] #[doc(hidden)] pub use self::local::os::Key as __OsLocalKeyInner; + #[unstable(feature = "libstd_thread_internals", issue = "none")] #[cfg(all(target_family = "wasm", not(target_feature = "atomics")))] #[doc(hidden)] @@ -490,6 +509,31 @@ impl Builder { let output_capture = crate::io::set_output_capture(None); crate::io::set_output_capture(output_capture.clone()); + // Pass `f` in `MaybeUninit` because actually that closure might *run longer than the lifetime of `F`*. + // See <https://github.com/rust-lang/rust/issues/101983> for more details. + // To prevent leaks we use a wrapper that drops its contents. + #[repr(transparent)] + struct MaybeDangling<T>(mem::MaybeUninit<T>); + impl<T> MaybeDangling<T> { + fn new(x: T) -> Self { + MaybeDangling(mem::MaybeUninit::new(x)) + } + fn into_inner(self) -> T { + // SAFETY: we are always initiailized. + let ret = unsafe { self.0.assume_init_read() }; + // Make sure we don't drop. + mem::forget(self); + ret + } + } + impl<T> Drop for MaybeDangling<T> { + fn drop(&mut self) { + // SAFETY: we are always initiailized. + unsafe { self.0.assume_init_drop() }; + } + } + + let f = MaybeDangling::new(f); let main = move || { if let Some(name) = their_thread.cname() { imp::Thread::set_name(name); @@ -497,6 +541,8 @@ impl Builder { crate::io::set_output_capture(output_capture); + // SAFETY: we constructed `f` initialized. + let f = f.into_inner(); // SAFETY: the stack guard passed is the one for the current thread. // This means the current thread's stack and the new thread's stack // are properly set and protected from each other. @@ -509,6 +555,12 @@ impl Builder { // same `JoinInner` as this closure meaning the mutation will be // safe (not modify it and affect a value far away). unsafe { *their_packet.result.get() = Some(try_result) }; + // Here `their_packet` gets dropped, and if this is the last `Arc` for that packet that + // will call `decrement_num_running_threads` and therefore signal that this thread is + // done. + drop(their_packet); + // Here, the lifetime `'a` and even `'scope` can end. `main` keeps running for a bit + // after that before returning itself. }; if let Some(scope_data) = &my_packet.scope { @@ -770,6 +822,8 @@ pub fn panicking() -> bool { panicking::panicking() } +/// Use [`sleep`]. +/// /// Puts the current thread to sleep for at least the specified amount of time. /// /// The thread may sleep longer than the duration specified due to scheduling @@ -840,10 +894,22 @@ pub fn sleep(dur: Duration) { imp::Thread::sleep(dur) } +/// Used to ensure that `park` and `park_timeout` do not unwind, as that can +/// cause undefined behaviour if not handled correctly (see #102398 for context). +struct PanicGuard; + +impl Drop for PanicGuard { + fn drop(&mut self) { + rtabort!("an irrecoverable error occurred while synchronizing threads") + } +} + /// Blocks unless or until the current thread's token is made available. /// /// A call to `park` does not guarantee that the thread will remain parked -/// forever, and callers should be prepared for this possibility. +/// forever, and callers should be prepared for this possibility. However, +/// it is guaranteed that this function will not panic (it may abort the +/// process if the implementation encounters some rare errors). /// /// # park and unpark /// @@ -928,10 +994,13 @@ pub fn sleep(dur: Duration) { /// [`thread::park_timeout`]: park_timeout #[stable(feature = "rust1", since = "1.0.0")] pub fn park() { + let guard = PanicGuard; // SAFETY: park_timeout is called on the parker owned by this thread. unsafe { current().inner.as_ref().parker().park(); } + // No panic occurred, do not abort. + forget(guard); } /// Use [`park_timeout`]. @@ -992,10 +1061,13 @@ pub fn park_timeout_ms(ms: u32) { /// ``` #[stable(feature = "park_timeout", since = "1.4.0")] pub fn park_timeout(dur: Duration) { + let guard = PanicGuard; // SAFETY: park_timeout is called on the parker owned by this thread. unsafe { current().inner.as_ref().parker().park_timeout(dur); } + // No panic occurred, do not abort. + forget(guard); } //////////////////////////////////////////////////////////////////////////////// @@ -1033,24 +1105,45 @@ pub struct ThreadId(NonZeroU64); impl ThreadId { // Generate a new unique thread ID. fn new() -> ThreadId { - // It is UB to attempt to acquire this mutex reentrantly! - static GUARD: mutex::StaticMutex = mutex::StaticMutex::new(); - static mut COUNTER: u64 = 1; - - unsafe { - let guard = GUARD.lock(); - - // If we somehow use up all our bits, panic so that we're not - // covering up subtle bugs of IDs being reused. - if COUNTER == u64::MAX { - drop(guard); // in case the panic handler ends up calling `ThreadId::new()`, avoid reentrant lock acquire. - panic!("failed to generate unique thread ID: bitspace exhausted"); - } - - let id = COUNTER; - COUNTER += 1; + #[cold] + fn exhausted() -> ! { + panic!("failed to generate unique thread ID: bitspace exhausted") + } - ThreadId(NonZeroU64::new(id).unwrap()) + cfg_if::cfg_if! { + if #[cfg(target_has_atomic = "64")] { + use crate::sync::atomic::{AtomicU64, Ordering::Relaxed}; + + static COUNTER: AtomicU64 = AtomicU64::new(0); + + let mut last = COUNTER.load(Relaxed); + loop { + let Some(id) = last.checked_add(1) else { + exhausted(); + }; + + match COUNTER.compare_exchange_weak(last, id, Relaxed, Relaxed) { + Ok(_) => return ThreadId(NonZeroU64::new(id).unwrap()), + Err(id) => last = id, + } + } + } else { + use crate::sync::{Mutex, PoisonError}; + + static COUNTER: Mutex<u64> = Mutex::new(0); + + let mut counter = COUNTER.lock().unwrap_or_else(PoisonError::into_inner); + let Some(id) = counter.checked_add(1) else { + // in case the panic handler ends up calling `ThreadId::new()`, + // avoid reentrant lock acquire. + drop(counter); + exhausted(); + }; + + *counter = id; + drop(counter); + ThreadId(NonZeroU64::new(id).unwrap()) + } } } diff --git a/library/std/src/thread/tests.rs b/library/std/src/thread/tests.rs index ec68b5291..6c9ce6fa0 100644 --- a/library/std/src/thread/tests.rs +++ b/library/std/src/thread/tests.rs @@ -37,6 +37,37 @@ fn test_named_thread() { .unwrap(); } +#[cfg(any( + // Note: musl didn't add pthread_getname_np until 1.2.3 + all(target_os = "linux", target_env = "gnu"), + target_os = "macos", + target_os = "ios", + target_os = "watchos" +))] +#[test] +fn test_named_thread_truncation() { + use crate::ffi::CStr; + + let long_name = crate::iter::once("test_named_thread_truncation") + .chain(crate::iter::repeat(" yada").take(100)) + .collect::<String>(); + + let result = Builder::new().name(long_name.clone()).spawn(move || { + // Rust remembers the full thread name itself. + assert_eq!(thread::current().name(), Some(long_name.as_str())); + + // But the system is limited -- make sure we successfully set a truncation. + let mut buf = vec![0u8; long_name.len() + 1]; + unsafe { + libc::pthread_getname_np(libc::pthread_self(), buf.as_mut_ptr().cast(), buf.len()); + } + let cstr = CStr::from_bytes_until_nul(&buf).unwrap(); + assert!(cstr.to_bytes().len() > 0); + assert!(long_name.as_bytes().starts_with(cstr.to_bytes())); + }); + result.unwrap().join().unwrap(); +} + #[test] #[should_panic] fn test_invalid_named_thread() { @@ -245,6 +276,28 @@ fn test_try_panic_any_message_unit_struct() { } #[test] +fn test_park_unpark_before() { + for _ in 0..10 { + thread::current().unpark(); + thread::park(); + } +} + +#[test] +fn test_park_unpark_called_other_thread() { + for _ in 0..10 { + let th = thread::current(); + + let _guard = thread::spawn(move || { + super::sleep(Duration::from_millis(50)); + th.unpark(); + }); + + thread::park(); + } +} + +#[test] fn test_park_timeout_unpark_before() { for _ in 0..10 { thread::current().unpark(); @@ -329,3 +382,22 @@ fn test_scoped_threads_nll() { let x = 42_u8; foo(&x); } + +// Regression test for https://github.com/rust-lang/rust/issues/98498. +#[test] +#[cfg(miri)] // relies on Miri's data race detector +fn scope_join_race() { + for _ in 0..100 { + let a_bool = AtomicBool::new(false); + + thread::scope(|s| { + for _ in 0..5 { + s.spawn(|| a_bool.load(Ordering::Relaxed)); + } + + for _ in 0..5 { + s.spawn(|| a_bool.load(Ordering::Relaxed)); + } + }); + } +} diff --git a/library/std/src/time.rs b/library/std/src/time.rs index 759a59e1f..ecd06ebf7 100644 --- a/library/std/src/time.rs +++ b/library/std/src/time.rs @@ -43,8 +43,8 @@ use crate::sys_common::{FromInner, IntoInner}; #[stable(feature = "time", since = "1.3.0")] pub use core::time::Duration; -#[unstable(feature = "duration_checked_float", issue = "83400")] -pub use core::time::FromFloatSecsError; +#[stable(feature = "duration_checked_float", since = "1.66.0")] +pub use core::time::TryFromFloatSecsError; /// A measurement of a monotonically nondecreasing clock. /// Opaque and useful only with [`Duration`]. @@ -356,7 +356,7 @@ impl Instant { /// /// # Panics /// - /// Previous rust versions panicked when self was earlier than the current time. Currently this + /// Previous rust versions panicked when the current time was earlier than self. Currently this /// method returns a Duration of zero in that case. Future versions may reintroduce the panic. /// See [Monotonicity]. /// diff --git a/library/std/src/time/tests.rs b/library/std/src/time/tests.rs index d710a5744..6229556c8 100644 --- a/library/std/src/time/tests.rs +++ b/library/std/src/time/tests.rs @@ -31,7 +31,8 @@ fn instant_monotonic_concurrent() -> crate::thread::Result<()> { .map(|_| { crate::thread::spawn(|| { let mut old = Instant::now(); - for _ in 0..5_000_000 { + let count = if cfg!(miri) { 1_000 } else { 5_000_000 }; + for _ in 0..count { let new = Instant::now(); assert!(new >= old); old = new; |