Merging upstream version 1.66.0+dfsg1.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

136 files changed, 4131 insertions, 2786 deletions

diff --git a/library/std/Cargo.toml b/library/std/Cargo.toml
index 324ecc804..bc10b12ec 100644
--- a/library/std/Cargo.toml
+++ b/library/std/Cargo.toml
@@ -11,11 +11,11 @@ crate-type = ["dylib", "rlib"]
 
 [dependencies]
 alloc = { path = "../alloc" }
-cfg-if = { version = "0.1.8", features = ['rustc-dep-of-std'] }
+cfg-if = { version = "1.0", features = ['rustc-dep-of-std'] }
 panic_unwind = { path = "../panic_unwind", optional = true }
 panic_abort = { path = "../panic_abort" }
 core = { path = "../core" }
-libc = { version = "0.2.126", default-features = false, features = ['rustc-dep-of-std'] }
+libc = { version = "0.2.135", default-features = false, features = ['rustc-dep-of-std'] }
 compiler_builtins = { version = "0.1.73" }
 profiler_builtins = { path = "../profiler_builtins", optional = true }
 unwind = { path = "../unwind" }
diff --git a/library/std/src/backtrace.rs b/library/std/src/backtrace.rs
index 5cf6ec817..9cb74f951 100644
--- a/library/std/src/backtrace.rs
+++ b/library/std/src/backtrace.rs
@@ -14,8 +14,8 @@
 //! 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
@@ -39,7 +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
@@ -325,8 +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 {
@@ -469,8 +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 9845d1faf..708edc5de 100644
--- a/library/std/src/collections/hash/map.rs
+++ b/library/std/src/collections/hash/map.rs
@@ -9,7 +9,6 @@ use crate::borrow::Borrow;
 use crate::cell::Cell;
 use crate::collections::TryReserveError;
 use crate::collections::TryReserveErrorKind;
-#[cfg(not(bootstrap))]
 use crate::error::Error;
 use crate::fmt::{self, Debug};
 #[allow(deprecated)]
@@ -281,7 +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) }
     }
 
@@ -759,7 +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.
@@ -2160,7 +2160,6 @@ impl<'a, K: Debug, V: Debug> fmt::Display for OccupiedError<'a, K, V> {
     }
 }
 
-#[cfg(not(bootstrap))]
 #[unstable(feature = "map_try_insert", issue = "82766")]
 impl<'a, K: fmt::Debug, V: fmt::Debug> Error for OccupiedError<'a, K, V> {
     #[allow(deprecated)]
diff --git a/library/std/src/collections/hash/map/tests.rs b/library/std/src/collections/hash/map/tests.rs
index cb3032719..65634f206 100644
--- a/library/std/src/collections/hash/map/tests.rs
+++ b/library/std/src/collections/hash/map/tests.rs
@@ -1115,3 +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 5b6a415fa..cee884145 100644
--- a/library/std/src/collections/hash/set.rs
+++ b/library/std/src/collections/hash/set.rs
@@ -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 e45059595..05f8fd8de 100644
--- a/library/std/src/error.rs
+++ b/library/std/src/error.rs
@@ -4,242 +4,12 @@
 #[cfg(test)]
 mod tests;
 
-#[cfg(bootstrap)]
-use core::array;
-#[cfg(bootstrap)]
-use core::convert::Infallible;
-
-#[cfg(bootstrap)]
-use crate::alloc::{AllocError, LayoutError};
-#[cfg(bootstrap)]
-use crate::any::Demand;
-#[cfg(bootstrap)]
-use crate::any::{Provider, TypeId};
 use crate::backtrace::Backtrace;
-#[cfg(bootstrap)]
-use crate::borrow::Cow;
-#[cfg(bootstrap)]
-use crate::cell;
-#[cfg(bootstrap)]
-use crate::char;
-#[cfg(bootstrap)]
-use crate::fmt::Debug;
-#[cfg(bootstrap)]
-use crate::fmt::Display;
 use crate::fmt::{self, Write};
-#[cfg(bootstrap)]
-use crate::io;
-#[cfg(bootstrap)]
-use crate::mem::transmute;
-#[cfg(bootstrap)]
-use crate::num;
-#[cfg(bootstrap)]
-use crate::str;
-#[cfg(bootstrap)]
-use crate::string;
-#[cfg(bootstrap)]
-use crate::sync::Arc;
-#[cfg(bootstrap)]
-use crate::time;
 
-#[cfg(not(bootstrap))]
 #[stable(feature = "rust1", since = "1.0.0")]
 pub use core::error::Error;
 
-/// `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 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.
-#[stable(feature = "rust1", since = "1.0.0")]
-#[cfg_attr(not(test), rustc_diagnostic_item = "Error")]
-#[cfg(bootstrap)]
-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, demand: &mut Demand<'a>) {
-    ///         demand
-    ///             .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, demand: &mut Demand<'a>) {}
-}
-
-#[cfg(bootstrap)]
-#[unstable(feature = "error_generic_member_access", issue = "99301")]
-impl<'b> Provider for dyn Error + 'b {
-    fn provide<'a>(&'a self, demand: &mut Demand<'a>) {
-        self.provide(demand)
-    }
-}
-
 mod private {
     // This is a hack to prevent `type_id` from being overridden by `Error`
     // implementations, since that can enable unsound downcasting.
@@ -248,799 +18,6 @@ mod private {
     pub struct Internal;
 }
 
-#[cfg(bootstrap)]
-#[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)
-    }
-}
-
-#[cfg(bootstrap)]
-#[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)
-    }
-}
-
-#[cfg(bootstrap)]
-#[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))
-    }
-}
-
-#[cfg(bootstrap)]
-#[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
-    }
-}
-
-#[cfg(bootstrap)]
-#[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))
-    }
-}
-
-#[cfg(bootstrap)]
-#[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))
-    }
-}
-
-#[cfg(bootstrap)]
-#[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))
-    }
-}
-
-#[cfg(bootstrap)]
-#[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))
-    }
-}
-
-#[cfg(bootstrap)]
-#[unstable(feature = "never_type", issue = "35121")]
-impl Error for ! {}
-
-#[cfg(bootstrap)]
-#[unstable(
-    feature = "allocator_api",
-    reason = "the precise API and guarantees it provides may be tweaked.",
-    issue = "32838"
-)]
-impl Error for AllocError {}
-
-#[cfg(bootstrap)]
-#[stable(feature = "alloc_layout", since = "1.28.0")]
-impl Error for LayoutError {}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for str::ParseBoolError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "failed to parse bool"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for str::Utf8Error {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "invalid utf-8: corrupt contents"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for num::ParseIntError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        self.__description()
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "try_from", since = "1.34.0")]
-impl Error for num::TryFromIntError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        self.__description()
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "try_from", since = "1.34.0")]
-impl Error for array::TryFromSliceError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        self.__description()
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for num::ParseFloatError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        self.__description()
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for string::FromUtf8Error {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "invalid utf-8"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for string::FromUtf16Error {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "invalid utf-16"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "str_parse_error2", since = "1.8.0")]
-impl Error for Infallible {
-    fn description(&self) -> &str {
-        match *self {}
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "decode_utf16", since = "1.9.0")]
-impl Error for char::DecodeUtf16Error {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "unpaired surrogate found"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "u8_from_char", since = "1.59.0")]
-impl Error for char::TryFromCharError {}
-
-#[cfg(bootstrap)]
-#[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"
-    }
-}
-
-#[cfg(bootstrap)]
-#[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"
-    }
-}
-
-#[cfg(bootstrap)]
-#[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)
-    }
-}
-
-#[cfg(bootstrap)]
-#[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()
-    }
-}
-
-#[cfg(bootstrap)]
-#[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, demand: &mut Demand<'b>) {
-        Error::provide(&**self, demand);
-    }
-}
-
-#[cfg(bootstrap)]
-#[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, demand: &mut Demand<'a>) {
-        Error::provide(&**self, demand);
-    }
-}
-
-#[cfg(bootstrap)]
-#[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"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "try_borrow", since = "1.13.0")]
-impl Error for cell::BorrowError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "already mutably borrowed"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "try_borrow", since = "1.13.0")]
-impl Error for cell::BorrowMutError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "already borrowed"
-    }
-}
-
-#[cfg(bootstrap)]
-#[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`"
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "char_from_str", since = "1.20.0")]
-impl Error for char::ParseCharError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        self.__description()
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "try_reserve", since = "1.57.0")]
-impl Error for alloc::collections::TryReserveError {}
-
-#[cfg(bootstrap)]
-#[unstable(feature = "duration_checked_float", issue = "83400")]
-impl Error for time::FromFloatSecsError {}
-
-#[cfg(bootstrap)]
-#[stable(feature = "rust1", since = "1.0.0")]
-impl Error for alloc::ffi::NulError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        "nul byte found in data"
-    }
-}
-
-#[cfg(bootstrap)]
-#[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")
-    }
-}
-
-#[cfg(bootstrap)]
-#[stable(feature = "frombyteswithnulerror_impls", since = "1.17.0")]
-impl Error for core::ffi::FromBytesWithNulError {
-    #[allow(deprecated)]
-    fn description(&self) -> &str {
-        self.__description()
-    }
-}
-
-#[cfg(bootstrap)]
-#[unstable(feature = "cstr_from_bytes_until_nul", issue = "95027")]
-impl Error for core::ffi::FromBytesUntilNulError {}
-
-#[cfg(bootstrap)]
-#[stable(feature = "cstring_from_vec_with_nul", since = "1.58.0")]
-impl Error for alloc::ffi::FromVecWithNulError {}
-
-#[cfg(bootstrap)]
-#[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())
-    }
-}
-
-#[cfg(bootstrap)]
-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`.
-#[cfg(bootstrap)]
-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
-        }
-    }
-}
-
-#[cfg(bootstrap)]
-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)
-    }
-}
-
-#[cfg(bootstrap)]
-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)
-    }
-}
-
-#[cfg(bootstrap)]
-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.sources();
-    ///
-    /// 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 sources(&self) -> Sources<'_> {
-        // You may think this method would be better in the Error trait, and you'd be right.
-        // Unfortunately that doesn't work, not because of the object safety rules but because we
-        // save a reference to self in Sources below as a trait object. If this method was
-        // declared in Error, then self would have the type &T where T is some concrete type which
-        // implements Error. We would need to coerce self to have type &dyn Error, but that requires
-        // that Self has a known size (i.e., Self: Sized). We can't put that bound on Error
-        // since that would forbid Error trait objects, and we can't put that bound on the method
-        // because that means the method can't be called on trait objects (we'd also need the
-        // 'static bound, but that isn't allowed because methods with bounds on Self other than
-        // Sized are not object-safe). Requiring an Unsize bound is not backwards compatible.
-
-        Sources { 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)]
-#[cfg(bootstrap)]
-pub struct Sources<'a> {
-    current: Option<&'a (dyn Error + 'static)>,
-}
-
-#[cfg(bootstrap)]
-#[unstable(feature = "error_iter", issue = "58520")]
-impl<'a> Iterator for Sources<'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
-    }
-}
-
-#[cfg(bootstrap)]
-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)
-        })
-    }
-}
-
-#[cfg(bootstrap)]
-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 sources, either entirely on a
diff --git a/library/std/src/fs.rs b/library/std/src/fs.rs
index c6c78dc39..188ff00e1 100644
--- a/library/std/src/fs.rs
+++ b/library/std/src/fs.rs
@@ -1365,6 +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
@@ -1390,8 +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
     ///
@@ -1405,7 +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/error.rs b/library/std/src/io/error.rs
index 29b09fcc5..3cabf2449 100644
--- a/library/std/src/io/error.rs
+++ b/library/std/src/io/error.rs
@@ -76,7 +76,6 @@ impl fmt::Debug for Error {
     }
 }
 
-#[cfg(not(bootstrap))]
 #[stable(feature = "rust1", since = "1.0.0")]
 impl From<alloc::ffi::NulError> for Error {
     /// Converts a [`alloc::ffi::NulError`] into a [`Error`].
@@ -388,7 +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",
@@ -432,6 +431,7 @@ impl ErrorKind {
             WouldBlock => "operation would block",
             WriteZero => "write zero",
         }
+        // tidy-alphabetical-end
     }
 }
 
@@ -482,6 +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`.
     ///
@@ -496,7 +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")]
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/mod.rs b/library/std/src/io/mod.rs
index eeace2c43..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")]
@@ -580,7 +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.
     ///
@@ -756,7 +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
diff --git a/library/std/src/io/readbuf.rs b/library/std/src/io/readbuf.rs
index b1a84095f..4800eeda0 100644
--- a/library/std/src/io/readbuf.rs
+++ b/library/std/src/io/readbuf.rs
@@ -3,10 +3,10 @@
 #[cfg(test)]
 mod tests;
 
-use crate::cmp;
 use crate::fmt::{self, Debug, Formatter};
 use crate::io::{Result, Write};
 use crate::mem::{self, MaybeUninit};
+use crate::{cmp, ptr};
 
 /// A borrowed byte buffer which is incrementally filled and initialized.
 ///
@@ -250,8 +250,11 @@ impl<'a> BorrowedCursor<'a> {
     /// Initializes all bytes in the cursor.
     #[inline]
     pub fn ensure_init(&mut self) -> &mut Self {
-        for byte in self.uninit_mut() {
-            byte.write(0);
+        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();
 
diff --git a/library/std/src/io/stdio.rs b/library/std/src/io/stdio.rs
index 2dc12a18a..1141a957d 100644
--- a/library/std/src/io/stdio.rs
+++ b/library/std/src/io/stdio.rs
@@ -7,6 +7,7 @@ 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::sync::atomic::{AtomicBool, Ordering};
 use crate::sync::{Arc, Mutex, MutexGuard, OnceLock};
@@ -999,7 +1000,18 @@ 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
@@ -1009,16 +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/keyword_docs.rs b/library/std/src/keyword_docs.rs
index a4b0522b0..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
 ///
@@ -1914,12 +1918,12 @@ 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 information.
 ///
@@ -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 {}
 
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 bc4f1b27c..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
@@ -251,11 +251,11 @@
 #![feature(doc_notable_trait)]
 #![feature(dropck_eyepatch)]
 #![feature(exhaustive_patterns)]
+#![feature(if_let_guard)]
 #![feature(intra_doc_pointers)]
-#![cfg_attr(bootstrap, feature(label_break_value))]
+#![feature(is_terminal)]
 #![feature(lang_items)]
 #![feature(let_chains)]
-#![cfg_attr(bootstrap, feature(let_else))]
 #![feature(linkage)]
 #![feature(link_cfg)]
 #![feature(min_specialization)]
@@ -280,11 +280,10 @@
 #![feature(core_intrinsics)]
 #![feature(cstr_from_bytes_until_nul)]
 #![feature(cstr_internals)]
-#![feature(duration_checked_float)]
 #![feature(duration_constants)]
-#![cfg_attr(not(bootstrap), feature(error_generic_member_access))]
-#![cfg_attr(not(bootstrap), feature(error_in_core))]
-#![cfg_attr(not(bootstrap), feature(error_iter))]
+#![feature(error_generic_member_access)]
+#![feature(error_in_core)]
+#![feature(error_iter)]
 #![feature(exact_size_is_empty)]
 #![feature(exclusive_wrapper)]
 #![feature(extend_one)]
@@ -293,10 +292,9 @@
 #![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)]
@@ -315,6 +313,7 @@
 #![feature(strict_provenance)]
 #![feature(maybe_uninit_uninit_array)]
 #![feature(const_maybe_uninit_uninit_array)]
+#![feature(const_waker)]
 //
 // Library features (alloc):
 #![feature(alloc_layout_extra)]
@@ -350,9 +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)]
@@ -530,9 +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"]
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 71e33fb9e..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")))]
@@ -192,6 +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
@@ -206,10 +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"))]
@@ -387,3 +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 1b3d11042..f92a05066 100644
--- a/library/std/src/os/fd/raw.rs
+++ b/library/std/src/os/fd/raw.rs
@@ -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.
@@ -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"))]
@@ -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/mod.rs b/library/std/src/os/mod.rs
index 18c64b510..42773805c 100644
--- a/library/std/src/os/mod.rs
+++ b/library/std/src/os/mod.rs
@@ -145,9 +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/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/wasi/io/fd.rs b/library/std/src/os/wasi/io/fd.rs
index 75703af6a..930aca887 100644
--- a/library/std/src/os/wasi/io/fd.rs
+++ b/library/std/src/os/wasi/io/fd.rs
@@ -1,10 +1,9 @@
 //! Owned and borrowed file descriptors.
 
-#![stable(feature = "io_safety_wasi", since = "1.65.0")]
+#![unstable(feature = "wasi_ext", issue = "71213")]
 
 // Tests for this module
 #[cfg(test)]
 mod tests;
 
-#[stable(feature = "io_safety_wasi", since = "1.65.0")]
 pub use crate::os::fd::owned::*;
diff --git a/library/std/src/os/wasi/io/mod.rs b/library/std/src/os/wasi/io/mod.rs
index 4f5cfbf9a..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)]
-#![stable(feature = "io_safety_wasi", since = "1.65.0")]
+#![stable(feature = "io_safety", since = "1.63.0")]
 
-mod fd;
-mod raw;
-
-#[stable(feature = "io_safety_wasi", since = "1.65.0")]
-pub use fd::*;
-#[stable(feature = "io_safety_wasi", since = "1.65.0")]
-pub use raw::*;
+#[stable(feature = "io_safety", since = "1.63.0")]
+pub use crate::os::fd::*;
diff --git a/library/std/src/os/wasi/io/raw.rs b/library/std/src/os/wasi/io/raw.rs
index 4ac792ee8..da3b36ada 100644
--- a/library/std/src/os/wasi/io/raw.rs
+++ b/library/std/src/os/wasi/io/raw.rs
@@ -1,6 +1,20 @@
 //! WASI-specific extensions to general I/O primitives.
 
-#![stable(feature = "io_safety_wasi", since = "1.65.0")]
+#![unstable(feature = "wasi_ext", issue = "71213")]
 
-#[stable(feature = "io_safety_wasi", since = "1.65.0")]
+// NOTE: despite the fact that this module is unstable,
+// stable Rust had the capability to access the stable
+// re-exported items from os::fd::raw through this
+// unstable module.
+// In PR #95956 the stability checker was changed to check
+// all path segments of an item rather than just the last,
+// which caused the aforementioned stable usage to regress
+// (see issue #99502).
+// As a result, the items in os::fd::raw were given the
+// rustc_allowed_through_unstable_modules attribute.
+// No regression tests were added to ensure this property,
+// as CI is not configured to test wasm32-wasi.
+// If this module is stabilized,
+// you may want to remove those attributes
+// (assuming no other unstable modules need them).
 pub use crate::os::fd::raw::*;
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/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/personality/dwarf/eh.rs b/library/std/src/personality/dwarf/eh.rs
index 8799137b7..27b50c13b 100644
--- a/library/std/src/personality/dwarf/eh.rs
+++ b/library/std/src/personality/dwarf/eh.rs
@@ -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/std/src/primitive_docs.rs b/library/std/src/primitive_docs.rs
index 242f44ade..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:
 ///
 /// ```
diff --git a/library/std/src/process.rs b/library/std/src/process.rs
index d91d4fa64..400d25beb 100644
--- a/library/std/src/process.rs
+++ b/library/std/src/process.rs
@@ -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 98f6cc7aa..9c2f0c1dd 100644
--- a/library/std/src/rt.rs
+++ b/library/std/src/rt.rs
@@ -89,7 +89,7 @@ macro_rules! rtunwrap {
 // `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 3 values. These are documented in
+// 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, sigpipe: u8) {
@@ -160,15 +160,12 @@ fn lang_start<T: crate::process::Termination + 'static>(
     main: fn() -> T,
     argc: isize,
     argv: *const *const u8,
-    #[cfg(not(bootstrap))] sigpipe: 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,
-        #[cfg(bootstrap)]
-        2, // Temporary inlining of sigpipe::DEFAULT until bootstrap stops being special
-        #[cfg(not(bootstrap))]
         sigpipe,
     );
     v
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/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()`]’s 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/rwlock.rs b/library/std/src/sync/rwlock.rs
index 9ab781561..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,
@@ -166,7 +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
@@ -180,9 +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
     ///
@@ -214,7 +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
@@ -227,13 +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
@@ -262,20 +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
     ///
@@ -302,7 +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
@@ -315,13 +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
@@ -421,10 +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
     ///
@@ -454,10 +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/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/fs.rs b/library/std/src/sys/hermit/fs.rs
index f921839cf..af297ff1e 100644
--- a/library/std/src/sys/hermit/fs.rs
+++ b/library/std/src/sys/hermit/fs.rs
@@ -1,3 +1,4 @@
+use crate::convert::TryFrom;
 use crate::ffi::{CStr, CString, OsString};
 use crate::fmt;
 use crate::hash::{Hash, Hasher};
@@ -5,6 +6,7 @@ use crate::io::{self, Error, ErrorKind};
 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);
 
@@ -272,8 +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> {
@@ -373,9 +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/mod.rs b/library/std/src/sys/hermit/mod.rs
index 827d82900..e6534df89 100644
--- a/library/std/src/sys/hermit/mod.rs
+++ b/library/std/src/sys/hermit/mod.rs
@@ -106,9 +106,7 @@ 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.
-pub unsafe fn cleanup() {
-    args::cleanup();
-}
+pub unsafe fn cleanup() {}
 
 #[cfg(not(test))]
 #[no_mangle]
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/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 5409bd177..0d934318c 100644
--- a/library/std/src/sys/sgx/abi/usercalls/alloc.rs
+++ b/library/std/src/sys/sgx/abi/usercalls/alloc.rs
@@ -316,9 +316,9 @@ where
 //   | small1 | Chunk smaller than 8 bytes
 //   +--------+
 fn region_as_aligned_chunks(ptr: *const u8, len: usize) -> (usize, usize, usize) {
-    let small0_size = if ptr as usize % 8 == 0 { 0 } else { 8 - ptr as usize % 8 };
-    let small1_size = (len - small0_size as usize) % 8;
-    let big_size = len - small0_size as usize - small1_size as 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)
 }
@@ -364,8 +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)
                 );
@@ -378,8 +378,8 @@ pub(crate) unsafe fn copy_to_userspace(src: *const u8, dst: *mut u8, len: usize)
     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
@@ -404,17 +404,17 @@ pub(crate) unsafe fn copy_to_userspace(src: *const u8, dst: *mut u8, len: usize)
 
         unsafe {
             // Copy small0
-            copy_bytewise_to_userspace(src, dst, small0_size as _);
+            copy_bytewise_to_userspace(src, dst, small0_size);
 
             // Copy big
-            let big_src = src.offset(small0_size as _);
-            let big_dst = dst.offset(small0_size as _);
-            copy_quadwords(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);
         }
     }
 }
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 969222253..6c66b93a3 100644
--- a/library/std/src/sys/solid/fs.rs
+++ b/library/std/src/sys/solid/fs.rs
@@ -175,15 +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) }))
     }
 }
 
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/fs.rs b/library/std/src/sys/unix/fs.rs
index cc347e358..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, 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 and HorizonOS do not support `futimens` at all and the behavior for those OS is therefore
-        // the same as for Redox.
-        #[cfg(any(target_os = "redox", target_os = "espidf", target_os = "horizon"))]
-        let omit = libc::timespec { tv_sec: 0, tv_nsec: 0 };
-        #[cfg(not(any(target_os = "redox", target_os = "espidf", target_os = "horizon")))]
-        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));
-                }
+                }));
             }
         }
     }
@@ -704,7 +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();
@@ -725,7 +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())
     }
@@ -829,7 +853,6 @@ impl DirEntry {
         target_os = "fuchsia",
         target_os = "redox"
     )))]
-    #[cfg_attr(miri, allow(unused))]
     fn name_cstr(&self) -> &CStr {
         unsafe { CStr::from_ptr(self.entry.d_name.as_ptr()) }
     }
@@ -841,7 +864,6 @@ impl DirEntry {
         target_os = "fuchsia",
         target_os = "redox"
     ))]
-    #[cfg_attr(miri, allow(unused))]
     fn name_cstr(&self) -> &CStr {
         &self.name
     }
@@ -931,8 +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> {
@@ -1084,6 +1105,17 @@ 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", target_os = "horizon"))] {
                 // Redox doesn't appear to support `UTIME_OMIT`.
@@ -1099,7 +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 {
@@ -1108,7 +1140,7 @@ impl File {
                                     tv_usec: (ts.tv_nsec / 1000) as _
                                 }
                             }
-                            let timevals = [ts_to_tv(&times.0[0]), ts_to_tv(&times.0[1])];
+                            let timevals = [ts_to_tv(&times[0]), ts_to_tv(&times[1])];
                             libc::futimes(self.as_raw_fd(), timevals.as_ptr())
                         }
                         // futimes requires even newer Android.
@@ -1121,7 +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(())
             }
         }
@@ -1134,9 +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) {
@@ -1144,10 +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
@@ -1198,7 +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());
@@ -1253,14 +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 {
@@ -1274,7 +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
@@ -1293,173 +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)> {
@@ -1609,9 +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() {
@@ -1655,9 +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<()> {
@@ -1666,16 +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;
@@ -1689,13 +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")),))]
@@ -1862,7 +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/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/mod.rs b/library/std/src/sys/unix/mod.rs
index c84e292ea..9055a011c 100644
--- a/library/std/src/sys/unix/mod.rs
+++ b/library/std/src/sys/unix/mod.rs
@@ -163,17 +163,27 @@ pub unsafe fn init(argc: isize, argv: *const *const u8, sigpipe: u8) {
             // 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 handler = match sigpipe {
-                sigpipe::INHERIT => None,
-                sigpipe::SIG_IGN => Some(libc::SIG_IGN),
-                sigpipe::SIG_DFL => Some(libc::SIG_DFL),
+            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);
             }
@@ -181,6 +191,26 @@ pub unsafe fn init(argc: isize, argv: *const *const u8, sigpipe: u8) {
     }
 }
 
+// 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() {
@@ -352,16 +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/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/process/process_common.rs b/library/std/src/sys/unix/process/process_common.rs
index 2834ee0ac..848adca78 100644
--- a/library/std/src/sys/unix/process/process_common.rs
+++ b/library/std/src/sys/unix/process/process_common.rs
@@ -39,10 +39,12 @@ 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::{
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 d176b3401..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]
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 26ae62817..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()
@@ -531,13 +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
@@ -822,14 +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/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 f6b627afc..c1d30dd9d 100644
--- a/library/std/src/sys/unix/thread.rs
+++ b/library/std/src/sys/unix/thread.rs
@@ -137,7 +137,9 @@ impl Thread {
         unsafe {
             // Available since glibc 2.12, musl 1.1.16, and uClibc 1.0.20.
             let name = truncate_cstr(name, TASK_COMM_LEN);
-            libc::pthread_setname_np(libc::pthread_self(), name.as_ptr());
+            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);
         }
     }
 
@@ -152,20 +154,22 @@ impl Thread {
     pub fn set_name(name: &CStr) {
         unsafe {
             let name = truncate_cstr(name, libc::MAXTHREADNAMESIZE);
-            libc::pthread_setname_np(name.as_ptr());
+            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);
         }
     }
 
@@ -178,9 +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);
         }
     }
 
@@ -785,6 +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/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
index e2453580d..35f1e68a8 100644
--- a/library/std/src/sys/unix/thread_parker/mod.rs
+++ b/library/std/src/sys/unix/thread_parker/mod.rs
@@ -11,7 +11,18 @@
 )))]
 
 cfg_if::cfg_if! {
-    if #[cfg(target_os = "netbsd")] {
+    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 {
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/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 2be0b34b9..87ea475c6 100644
--- a/library/std/src/sys/unsupported/locks/mutex.rs
+++ b/library/std/src/sys/unsupported/locks/mutex.rs
@@ -12,6 +12,7 @@ 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) }
     }
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/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 510cf36b1..d4866bbc3 100644
--- a/library/std/src/sys/wasi/fs.rs
+++ b/library/std/src/sys/wasi/fs.rs
@@ -1,7 +1,7 @@
 #![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, BorrowedCursor, IoSlice, IoSliceMut, SeekFrom};
 use crate::iter;
@@ -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);
     }
 }
 
@@ -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/stdio.rs b/library/std/src/sys/wasi/stdio.rs
index d2081771b..4cc0e4ed5 100644
--- a/library/std/src/sys/wasi/stdio.rs
+++ b/library/std/src/sys/wasi/stdio.rs
@@ -4,7 +4,7 @@ use super::fd::WasiFd;
 use crate::io::{self, IoSlice, IoSliceMut};
 use crate::mem::ManuallyDrop;
 use crate::os::raw;
-use crate::os::wasi::io::{AsFd, AsRawFd, BorrowedFd, FromRawFd};
+use crate::os::wasi::io::{AsRawFd, FromRawFd};
 
 pub struct Stdin;
 pub struct Stdout;
@@ -23,13 +23,6 @@ impl AsRawFd for Stdin {
     }
 }
 
-impl AsFd for Stdin {
-    #[inline]
-    fn as_fd(&self) -> BorrowedFd<'_> {
-        unsafe { BorrowedFd::borrow_raw(0) }
-    }
-}
-
 impl io::Read for Stdin {
     fn read(&mut self, data: &mut [u8]) -> io::Result<usize> {
         self.read_vectored(&mut [IoSliceMut::new(data)])
@@ -58,13 +51,6 @@ impl AsRawFd for Stdout {
     }
 }
 
-impl AsFd for Stdout {
-    #[inline]
-    fn as_fd(&self) -> BorrowedFd<'_> {
-        unsafe { BorrowedFd::borrow_raw(1) }
-    }
-}
-
 impl io::Write for Stdout {
     fn write(&mut self, data: &[u8]) -> io::Result<usize> {
         self.write_vectored(&[IoSlice::new(data)])
@@ -96,13 +82,6 @@ impl AsRawFd for Stderr {
     }
 }
 
-impl AsFd for Stderr {
-    #[inline]
-    fn as_fd(&self) -> BorrowedFd<'_> {
-        unsafe { BorrowedFd::borrow_raw(2) }
-    }
-}
-
 impl io::Write for Stderr {
     fn write(&mut self, data: &[u8]) -> io::Result<usize> {
         self.write_vectored(&[IoSlice::new(data)])
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/c.rs b/library/std/src/sys/windows/c.rs
index 89d0ab59b..be6fc2ebb 100644
--- a/library/std/src/sys/windows/c.rs
+++ b/library/std/src/sys/windows/c.rs
@@ -71,6 +71,7 @@ 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;
@@ -126,6 +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 {
@@ -194,6 +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;
@@ -279,7 +287,6 @@ pub const STATUS_INVALID_PARAMETER: NTSTATUS = 0xc000000d_u32 as _;
 pub const STATUS_PENDING: NTSTATUS = 0x103 as _;
 pub const STATUS_END_OF_FILE: NTSTATUS = 0xC0000011_u32 as _;
 pub const STATUS_NOT_IMPLEMENTED: NTSTATUS = 0xC0000002_u32 as _;
-pub const STATUS_NOT_SUPPORTED: NTSTATUS = 0xC00000BB_u32 as _;
 
 // Equivalent to the `NT_SUCCESS` C preprocessor macro.
 // See: https://docs.microsoft.com/en-us/windows-hardware/drivers/kernel/using-ntstatus-values
@@ -289,6 +296,7 @@ pub fn nt_success(status: NTSTATUS) -> bool {
 
 // "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)]
 pub struct UNICODE_STRING {
@@ -535,6 +543,12 @@ pub struct SYMBOLIC_LINK_REPARSE_BUFFER {
 /// 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,
     pub SubstituteNameLength: c_ushort,
@@ -565,6 +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 {
@@ -817,10 +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,
@@ -959,6 +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;
@@ -1101,6 +1116,7 @@ extern "system" {
         lpFileInformation: LPVOID,
         dwBufferSize: DWORD,
     ) -> BOOL;
+    pub fn GetFileType(hfile: HANDLE) -> DWORD;
     pub fn SleepConditionVariableSRW(
         ConditionVariable: PCONDITION_VARIABLE,
         SRWLock: PSRWLOCK,
@@ -1118,6 +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,
diff --git a/library/std/src/sys/windows/fs.rs b/library/std/src/sys/windows/fs.rs
index 155d0297e..378098038 100644
--- a/library/std/src/sys/windows/fs.rs
+++ b/library/std/src/sys/windows/fs.rs
@@ -1,5 +1,6 @@
 use crate::os::windows::prelude::*;
 
+use crate::borrow::Cow;
 use crate::ffi::OsString;
 use crate::fmt;
 use crate::io::{self, BorrowedCursor, Error, IoSlice, IoSliceMut, SeekFrom};
@@ -572,6 +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())
         })?;
@@ -711,7 +720,7 @@ 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..];
@@ -726,15 +735,19 @@ impl<'a> Iterator for DirBuffIter<'a> {
         //   `FileNameLength` bytes)
         let (name, is_directory, next_entry) = unsafe {
             let info = buffer.as_ptr().cast::<c::FILE_ID_BOTH_DIR_INFO>();
-            // Guaranteed to be aligned in documentation for
+            // 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
-            assert!(info.is_aligned());
-            let next_entry = (*info).NextEntryOffset as usize;
-            let name = crate::slice::from_raw_parts(
+            // 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>(),
-                (*info).FileNameLength as usize / size_of::<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)
         };
@@ -747,13 +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
@@ -1109,13 +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.
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/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 d6cd8f802..b5a49489d 100644
--- a/library/std/src/sys/windows/rand.rs
+++ b/library/std/src/sys/windows/rand.rs
@@ -13,15 +13,12 @@
 //! but significant number of users to experience panics caused by a failure of
 //! this function. See [#94098].
 //!
-//! The current version changes this to use the `BCRYPT_RNG_ALG_HANDLE`
-//! [Pseudo-handle], which gets the default RNG algorithm without querying the
-//! system preference thus hopefully avoiding the previous issue.
-//! This is only supported on Windows 10+ so a fallback is used for older versions.
+//! 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
-//! [Pseudo-handle]: https://docs.microsoft.com/en-us/windows/win32/seccng/cng-algorithm-pseudo-handles
 use crate::mem;
 use crate::ptr;
 use crate::sys::c;
@@ -33,37 +30,35 @@ use crate::sys::c;
 /// [`HashMap`]: crate::collections::HashMap
 /// [`RandomState`]: crate::collections::hash_map::RandomState
 pub fn hashmap_random_keys() -> (u64, u64) {
-    Rng::open().and_then(|rng| rng.gen_random_keys()).unwrap_or_else(fallback_rng)
+    Rng::SYSTEM.gen_random_keys().unwrap_or_else(fallback_rng)
 }
 
-struct Rng(c::BCRYPT_ALG_HANDLE);
+struct Rng {
+    algorithm: c::BCRYPT_ALG_HANDLE,
+    flags: u32,
+}
 impl Rng {
-    #[cfg(miri)]
-    fn open() -> Result<Self, c::NTSTATUS> {
-        const BCRYPT_RNG_ALG_HANDLE: c::BCRYPT_ALG_HANDLE = ptr::invalid_mut(0x81);
-        let _ = (
-            c::BCryptOpenAlgorithmProvider,
-            c::BCryptCloseAlgorithmProvider,
-            c::BCRYPT_RNG_ALGORITHM,
-            c::STATUS_NOT_SUPPORTED,
-        );
-        Ok(Self(BCRYPT_RNG_ALG_HANDLE))
+    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 }
     }
-    #[cfg(not(miri))]
-    // Open a handle to the RNG algorithm.
+
+    /// 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};
-        const ERROR_VALUE: c::LPVOID = ptr::invalid_mut(usize::MAX);
 
         // 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);
-        // We use a sentinel value to designate an error occurred last time.
-        if handle == ERROR_VALUE {
-            Err(c::STATUS_NOT_SUPPORTED)
-        } else if handle.is_null() {
+        if handle.is_null() {
             let status = unsafe {
                 c::BCryptOpenAlgorithmProvider(
                     &mut handle,
@@ -80,13 +75,12 @@ impl Rng {
                     unsafe { c::BCryptCloseAlgorithmProvider(handle, 0) };
                     handle = previous_handle;
                 }
-                Ok(Self(handle))
+                Ok(unsafe { Self::new(handle, 0) })
             } else {
-                HANDLE.store(ERROR_VALUE, Release);
                 Err(status)
             }
         } else {
-            Ok(Self(handle))
+            Ok(unsafe { Self::new(handle, 0) })
         }
     }
 
@@ -94,33 +88,19 @@ impl Rng {
         let mut v = (0, 0);
         let status = unsafe {
             let size = mem::size_of_val(&v).try_into().unwrap();
-            c::BCryptGenRandom(self.0, ptr::addr_of_mut!(v).cast(), size, 0)
+            c::BCryptGenRandom(self.algorithm, ptr::addr_of_mut!(v).cast(), size, self.flags)
         };
         if c::nt_success(status) { Ok(v) } else { Err(status) }
     }
 }
 
-/// Generate random numbers using the fallback RNG function (RtlGenRandom)
-#[cfg(not(target_vendor = "uwp"))]
+/// Generate random numbers using the fallback RNG function
 #[inline(never)]
 fn fallback_rng(rng_status: c::NTSTATUS) -> (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) };
-
-    if ret != 0 {
-        v
-    } else {
-        panic!(
-            "RNG broken: {rng_status:#x}, fallback RNG broken: {}",
-            crate::io::Error::last_os_error()
-        )
+    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}")
+        }
     }
 }
-
-/// We can't use RtlGenRandom with UWP, so there is no fallback
-#[cfg(target_vendor = "uwp")]
-#[inline(never)]
-fn fallback_rng(rng_status: c::NTSTATUS) -> (u64, u64) {
-    panic!("RNG broken: {rng_status:#x} fallback RNG broken: RtlGenRandom() not supported on UWP");
-}
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_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 3ad802afa..fad4a6333 100644
--- a/library/std/src/sys_common/net.rs
+++ b/library/std/src/sys_common/net.rs
@@ -2,12 +2,13 @@
 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::{AsInner, FromInner, IntoInner};
@@ -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 })
+            }
+        })
     }
 }
 
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/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 032bf604d..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.
 ///
@@ -151,25 +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
@@ -232,8 +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/thread/local.rs b/library/std/src/thread/local.rs
index 8aedfc4a6..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;
@@ -1036,7 +1037,10 @@ pub mod fast {
 }
 
 #[doc(hidden)]
-#[cfg(not(target_thread_local))]
+#[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;
diff --git a/library/std/src/thread/mod.rs b/library/std/src/thread/mod.rs
index ceea6986e..05023df1b 100644
--- a/library/std/src/thread/mod.rs
+++ b/library/std/src/thread/mod.rs
@@ -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;
@@ -192,32 +194,40 @@ pub use scoped::{scope, Scope, ScopedJoinHandle};
 #[stable(feature = "rust1", since = "1.0.0")]
 pub use self::local::{AccessError, LocalKey};
 
-// Select the type used by the thread_local! macro to access TLS keys. There
-// are three types: "static", "fast", "OS". The "OS" thread local key
+// 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. "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;
-#[unstable(feature = "libstd_thread_internals", issue = "none")]
-#[cfg(target_thread_local)]
-#[cfg(test)] // when building for tests, use real std's key
-pub use realstd::thread::__FastLocalKeyInner;
 
+// when building for tests, use real std's type
 #[unstable(feature = "libstd_thread_internals", issue = "none")]
-#[cfg(target_thread_local)]
 #[cfg(test)]
-pub use self::local::fast::Key as __FastLocalKeyInnerUnused; // we import this anyway to silence 'unused' warnings
+#[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)]
-#[cfg(not(target_thread_local))]
 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)]
@@ -499,6 +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);
@@ -506,6 +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.
@@ -518,6 +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 {
@@ -779,6 +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
@@ -849,10 +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
 ///
@@ -937,10 +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`].
@@ -1001,10 +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);
 }
 
 ////////////////////////////////////////////////////////////////////////////////
@@ -1065,24 +1128,21 @@ impl ThreadId {
                     }
                 }
             } else {
-                use crate::sys_common::mutex::StaticMutex;
+                use crate::sync::{Mutex, PoisonError};
 
-                // It is UB to attempt to acquire this mutex reentrantly!
-                static GUARD: StaticMutex = StaticMutex::new();
-                static mut COUNTER: u64 = 0;
+                static COUNTER: Mutex<u64> = Mutex::new(0);
 
-                unsafe {
-                    let guard = GUARD.lock();
+                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();
+                };
 
-                    let Some(id) = COUNTER.checked_add(1) else {
-                        drop(guard); // in case the panic handler ends up calling `ThreadId::new()`, avoid reentrant lock acquire.
-                        exhausted();
-                    };
-
-                    COUNTER = id;
-                    drop(guard);
-                    ThreadId(NonZeroU64::new(id).unwrap())
-                }
+                *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 777964f04..6c9ce6fa0 100644
--- a/library/std/src/thread/tests.rs
+++ b/library/std/src/thread/tests.rs
@@ -276,6 +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();
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/tests/run-time-detect.rs b/library/std/tests/run-time-detect.rs
index a57a52d9b..02c076f1b 100644
--- a/library/std/tests/run-time-detect.rs
+++ b/library/std/tests/run-time-detect.rs
@@ -14,77 +14,85 @@
 #[cfg(all(target_arch = "arm", any(target_os = "linux", target_os = "android")))]
 fn arm_linux() {
     use std::arch::is_arm_feature_detected;
+    // tidy-alphabetical-start
+    println!("aes: {}", is_arm_feature_detected!("aes"));
+    println!("crc: {}", is_arm_feature_detected!("crc"));
+    println!("crypto: {}", is_arm_feature_detected!("crypto"));
     println!("neon: {}", is_arm_feature_detected!("neon"));
     println!("pmull: {}", is_arm_feature_detected!("pmull"));
-    println!("crypto: {}", is_arm_feature_detected!("crypto"));
-    println!("crc: {}", is_arm_feature_detected!("crc"));
-    println!("aes: {}", is_arm_feature_detected!("aes"));
     println!("sha2: {}", is_arm_feature_detected!("sha2"));
+    // tidy-alphabetical-end
 }
 
 #[test]
 #[cfg(all(target_arch = "aarch64", any(target_os = "linux", target_os = "android")))]
 fn aarch64_linux() {
     use std::arch::is_aarch64_feature_detected;
-    println!("neon: {}", is_aarch64_feature_detected!("neon"));
+    // tidy-alphabetical-start
+    println!("aes: {}", is_aarch64_feature_detected!("aes"));
     println!("asimd: {}", is_aarch64_feature_detected!("asimd"));
-    println!("pmull: {}", is_aarch64_feature_detected!("pmull"));
-    println!("fp16: {}", is_aarch64_feature_detected!("fp16"));
-    println!("sve: {}", is_aarch64_feature_detected!("sve"));
+    println!("bf16: {}", is_aarch64_feature_detected!("bf16"));
+    println!("bti: {}", is_aarch64_feature_detected!("bti"));
     println!("crc: {}", is_aarch64_feature_detected!("crc"));
-    println!("lse: {}", is_aarch64_feature_detected!("lse"));
-    println!("lse2: {}", is_aarch64_feature_detected!("lse2"));
-    println!("rdm: {}", is_aarch64_feature_detected!("rdm"));
-    println!("rcpc: {}", is_aarch64_feature_detected!("rcpc"));
-    println!("rcpc2: {}", is_aarch64_feature_detected!("rcpc2"));
+    println!("dit: {}", is_aarch64_feature_detected!("dit"));
     println!("dotprod: {}", is_aarch64_feature_detected!("dotprod"));
-    println!("tme: {}", is_aarch64_feature_detected!("tme"));
+    println!("dpb2: {}", is_aarch64_feature_detected!("dpb2"));
+    println!("dpb: {}", is_aarch64_feature_detected!("dpb"));
+    println!("f32mm: {}", is_aarch64_feature_detected!("f32mm"));
+    println!("f64mm: {}", is_aarch64_feature_detected!("f64mm"));
+    println!("fcma: {}", is_aarch64_feature_detected!("fcma"));
     println!("fhm: {}", is_aarch64_feature_detected!("fhm"));
-    println!("dit: {}", is_aarch64_feature_detected!("dit"));
     println!("flagm: {}", is_aarch64_feature_detected!("flagm"));
-    println!("ssbs: {}", is_aarch64_feature_detected!("ssbs"));
-    println!("sb: {}", is_aarch64_feature_detected!("sb"));
-    println!("paca: {}", is_aarch64_feature_detected!("paca"));
-    println!("pacg: {}", is_aarch64_feature_detected!("pacg"));
-    println!("dpb: {}", is_aarch64_feature_detected!("dpb"));
-    println!("dpb2: {}", is_aarch64_feature_detected!("dpb2"));
-    println!("sve2: {}", is_aarch64_feature_detected!("sve2"));
-    println!("sve2-aes: {}", is_aarch64_feature_detected!("sve2-aes"));
-    println!("sve2-sm4: {}", is_aarch64_feature_detected!("sve2-sm4"));
-    println!("sve2-sha3: {}", is_aarch64_feature_detected!("sve2-sha3"));
-    println!("sve2-bitperm: {}", is_aarch64_feature_detected!("sve2-bitperm"));
+    println!("fp16: {}", is_aarch64_feature_detected!("fp16"));
     println!("frintts: {}", is_aarch64_feature_detected!("frintts"));
     println!("i8mm: {}", is_aarch64_feature_detected!("i8mm"));
-    println!("f32mm: {}", is_aarch64_feature_detected!("f32mm"));
-    println!("f64mm: {}", is_aarch64_feature_detected!("f64mm"));
-    println!("bf16: {}", is_aarch64_feature_detected!("bf16"));
-    println!("rand: {}", is_aarch64_feature_detected!("rand"));
-    println!("bti: {}", is_aarch64_feature_detected!("bti"));
-    println!("mte: {}", is_aarch64_feature_detected!("mte"));
     println!("jsconv: {}", is_aarch64_feature_detected!("jsconv"));
-    println!("fcma: {}", is_aarch64_feature_detected!("fcma"));
-    println!("aes: {}", is_aarch64_feature_detected!("aes"));
+    println!("lse2: {}", is_aarch64_feature_detected!("lse2"));
+    println!("lse: {}", is_aarch64_feature_detected!("lse"));
+    println!("mte: {}", is_aarch64_feature_detected!("mte"));
+    println!("neon: {}", is_aarch64_feature_detected!("neon"));
+    println!("paca: {}", is_aarch64_feature_detected!("paca"));
+    println!("pacg: {}", is_aarch64_feature_detected!("pacg"));
+    println!("pmull: {}", is_aarch64_feature_detected!("pmull"));
+    println!("rand: {}", is_aarch64_feature_detected!("rand"));
+    println!("rcpc2: {}", is_aarch64_feature_detected!("rcpc2"));
+    println!("rcpc: {}", is_aarch64_feature_detected!("rcpc"));
+    println!("rdm: {}", is_aarch64_feature_detected!("rdm"));
+    println!("sb: {}", is_aarch64_feature_detected!("sb"));
     println!("sha2: {}", is_aarch64_feature_detected!("sha2"));
     println!("sha3: {}", is_aarch64_feature_detected!("sha3"));
     println!("sm4: {}", is_aarch64_feature_detected!("sm4"));
+    println!("ssbs: {}", is_aarch64_feature_detected!("ssbs"));
+    println!("sve2-aes: {}", is_aarch64_feature_detected!("sve2-aes"));
+    println!("sve2-bitperm: {}", is_aarch64_feature_detected!("sve2-bitperm"));
+    println!("sve2-sha3: {}", is_aarch64_feature_detected!("sve2-sha3"));
+    println!("sve2-sm4: {}", is_aarch64_feature_detected!("sve2-sm4"));
+    println!("sve2: {}", is_aarch64_feature_detected!("sve2"));
+    println!("sve: {}", is_aarch64_feature_detected!("sve"));
+    println!("tme: {}", is_aarch64_feature_detected!("tme"));
+    // tidy-alphabetical-end
 }
 
 #[test]
 #[cfg(all(target_arch = "powerpc", target_os = "linux"))]
 fn powerpc_linux() {
     use std::arch::is_powerpc_feature_detected;
+    // tidy-alphabetical-start
     println!("altivec: {}", is_powerpc_feature_detected!("altivec"));
-    println!("vsx: {}", is_powerpc_feature_detected!("vsx"));
     println!("power8: {}", is_powerpc_feature_detected!("power8"));
+    println!("vsx: {}", is_powerpc_feature_detected!("vsx"));
+    // tidy-alphabetical-end
 }
 
 #[test]
 #[cfg(all(target_arch = "powerpc64", target_os = "linux"))]
 fn powerpc64_linux() {
     use std::arch::is_powerpc64_feature_detected;
+    // tidy-alphabetical-start
     println!("altivec: {}", is_powerpc64_feature_detected!("altivec"));
-    println!("vsx: {}", is_powerpc64_feature_detected!("vsx"));
     println!("power8: {}", is_powerpc64_feature_detected!("power8"));
+    println!("vsx: {}", is_powerpc64_feature_detected!("vsx"));
+    // tidy-alphabetical-end
 }
 
 #[test]
@@ -102,9 +110,9 @@ fn x86_all() {
     // the below is in alphabetical order and matches
     // the order of X86_ALLOWED_FEATURES in rustc_codegen_ssa's target_features.rs
 
+    // tidy-alphabetical-start
     println!("adx: {:?}", is_x86_feature_detected!("adx"));
     println!("aes: {:?}", is_x86_feature_detected!("aes"));
-    println!("avx: {:?}", is_x86_feature_detected!("avx"));
     println!("avx2: {:?}", is_x86_feature_detected!("avx2"));
     println!("avx512bf16: {:?}", is_x86_feature_detected!("avx512bf16"));
     println!("avx512bitalg: {:?}", is_x86_feature_detected!("avx512bitalg"));
@@ -117,13 +125,14 @@ fn x86_all() {
     println!("avx512ifma: {:?}", is_x86_feature_detected!("avx512ifma"));
     println!("avx512pf: {:?}", is_x86_feature_detected!("avx512pf"));
     println!("avx512vaes: {:?}", is_x86_feature_detected!("avx512vaes"));
-    println!("avx512vbmi: {:?}", is_x86_feature_detected!("avx512vbmi"));
     println!("avx512vbmi2: {:?}", is_x86_feature_detected!("avx512vbmi2"));
+    println!("avx512vbmi: {:?}", is_x86_feature_detected!("avx512vbmi"));
     println!("avx512vl: {:?}", is_x86_feature_detected!("avx512vl"));
     println!("avx512vnni: {:?}", is_x86_feature_detected!("avx512vnni"));
     println!("avx512vp2intersect: {:?}", is_x86_feature_detected!("avx512vp2intersect"));
     println!("avx512vpclmulqdq: {:?}", is_x86_feature_detected!("avx512vpclmulqdq"));
     println!("avx512vpopcntdq: {:?}", is_x86_feature_detected!("avx512vpopcntdq"));
+    println!("avx: {:?}", is_x86_feature_detected!("avx"));
     println!("bmi1: {:?}", is_x86_feature_detected!("bmi1"));
     println!("bmi2: {:?}", is_x86_feature_detected!("bmi2"));
     println!("cmpxchg16b: {:?}", is_x86_feature_detected!("cmpxchg16b"));
@@ -138,16 +147,17 @@ fn x86_all() {
     println!("rdseed: {:?}", is_x86_feature_detected!("rdseed"));
     println!("rtm: {:?}", is_x86_feature_detected!("rtm"));
     println!("sha: {:?}", is_x86_feature_detected!("sha"));
-    println!("sse: {:?}", is_x86_feature_detected!("sse"));
     println!("sse2: {:?}", is_x86_feature_detected!("sse2"));
     println!("sse3: {:?}", is_x86_feature_detected!("sse3"));
     println!("sse4.1: {:?}", is_x86_feature_detected!("sse4.1"));
     println!("sse4.2: {:?}", is_x86_feature_detected!("sse4.2"));
     println!("sse4a: {:?}", is_x86_feature_detected!("sse4a"));
+    println!("sse: {:?}", is_x86_feature_detected!("sse"));
     println!("ssse3: {:?}", is_x86_feature_detected!("ssse3"));
     println!("tbm: {:?}", is_x86_feature_detected!("tbm"));
     println!("xsave: {:?}", is_x86_feature_detected!("xsave"));
     println!("xsavec: {:?}", is_x86_feature_detected!("xsavec"));
     println!("xsaveopt: {:?}", is_x86_feature_detected!("xsaveopt"));
     println!("xsaves: {:?}", is_x86_feature_detected!("xsaves"));
+    // tidy-alphabetical-end
 }
diff --git a/library/stdarch/ci/android-install-sdk.sh b/library/stdarch/ci/android-install-sdk.sh
index 1beeb312a..3383dcb7f 100644
--- a/library/stdarch/ci/android-install-sdk.sh
+++ b/library/stdarch/ci/android-install-sdk.sh
@@ -19,8 +19,8 @@ set -ex
 # which apparently magically accepts the licenses.
 
 mkdir sdk
-curl --retry 5 https://dl.google.com/android/repository/sdk-tools-linux-3859397.zip -O
-unzip -d sdk sdk-tools-linux-3859397.zip
+curl --retry 5 https://dl.google.com/android/repository/sdk-tools-linux-4333796.zip -O
+unzip -d sdk sdk-tools-linux-4333796.zip
 
 case "$1" in
   arm | armv7)
diff --git a/library/stdarch/ci/docker/aarch64-linux-android/Dockerfile b/library/stdarch/ci/docker/aarch64-linux-android/Dockerfile
index 27bde89c5..6cf9b5061 100644
--- a/library/stdarch/ci/docker/aarch64-linux-android/Dockerfile
+++ b/library/stdarch/ci/docker/aarch64-linux-android/Dockerfile
@@ -1,17 +1,16 @@
-FROM ubuntu:16.04
+FROM ubuntu:22.04
 
-RUN dpkg --add-architecture i386 && \
-    apt-get update && \
+RUN apt-get update && \
     apt-get install -y --no-install-recommends \
   file \
   make \
   curl \
   ca-certificates \
-  python \
+  python-is-python3 \
   unzip \
   expect \
-  openjdk-9-jre \
-  libstdc++6:i386 \
+  openjdk-8-jre \
+  libstdc++6-i386-cross \
   libpulse0 \
   gcc \
   libc6-dev
diff --git a/library/stdarch/ci/docker/arm-linux-androideabi/Dockerfile b/library/stdarch/ci/docker/arm-linux-androideabi/Dockerfile
index 995a9e30e..fb1a0cecf 100644
--- a/library/stdarch/ci/docker/arm-linux-androideabi/Dockerfile
+++ b/library/stdarch/ci/docker/arm-linux-androideabi/Dockerfile
@@ -1,17 +1,16 @@
-FROM ubuntu:16.04
+FROM ubuntu:22.04
 
-RUN dpkg --add-architecture i386 && \
-    apt-get update && \
+RUN apt-get update && \
     apt-get install -y --no-install-recommends \
   file \
   make \
   curl \
   ca-certificates \
-  python \
+  python-is-python3 \
   unzip \
   expect \
-  openjdk-9-jre \
-  libstdc++6:i386 \
+  openjdk-8-jre \
+  libstdc++6-i386-cross \
   libpulse0 \
   gcc \
   libc6-dev
diff --git a/library/stdarch/ci/docker/x86_64-linux-android/Dockerfile b/library/stdarch/ci/docker/x86_64-linux-android/Dockerfile
index c2830b15f..82119be74 100644
--- a/library/stdarch/ci/docker/x86_64-linux-android/Dockerfile
+++ b/library/stdarch/ci/docker/x86_64-linux-android/Dockerfile
@@ -1,4 +1,4 @@
-FROM ubuntu:16.04
+FROM ubuntu:22.04
 
 RUN apt-get update && \
     apt-get install -y --no-install-recommends \
@@ -6,7 +6,7 @@ RUN apt-get update && \
   curl \
   gcc \
   libc-dev \
-  python \
+  python-is-python3 \
   unzip \
   file \
   make
diff --git a/library/stdarch/crates/core_arch/src/arm_shared/neon/mod.rs b/library/stdarch/crates/core_arch/src/arm_shared/neon/mod.rs
index 043f7ed51..0559aea83 100644
--- a/library/stdarch/crates/core_arch/src/arm_shared/neon/mod.rs
+++ b/library/stdarch/crates/core_arch/src/arm_shared/neon/mod.rs
@@ -12461,30 +12461,30 @@ mod tests {
     }
     #[simd_test(enable = "neon,i8mm")]
     unsafe fn test_vmmlaq_s32() {
-        let a: i32x4 = i32x4::new(1, 3, 4, 9);
-        let b: i8x16 = i8x16::new(1, 21, 31, 14, 5, 6, 17, 8, 9, 13, 15, 12, 13, 19, 20, 16);
-        let c: i8x16 = i8x16::new(12, 22, 3, 4, 5, 56, 7, 8, 91, 10, 11, 15, 13, 14, 17, 16);
-        let e: i32x4 = i32x4::new(1, 2, 3, 4);
+        let a = i32x4::new(1, 3, 4, -0x10000);
+        let b = i8x16::new(1, 21, 31, 14, 5, 6, -128, 8, 9, 13, 15, 12, 13, -1, 20, 16);
+        let c = i8x16::new(12, 22, 3, 4, -1, 56, 7, 8, 91, 10, -128, 15, 13, 14, 17, 16);
+        let e = i32x4::new(123, -5353, 690, -65576);
         let r: i32x4 = transmute(vmmlaq_s32(transmute(a), transmute(b), transmute(c)));
         assert_eq!(r, e);
     }
 
     #[simd_test(enable = "neon,i8mm")]
     unsafe fn test_vmmlaq_u32() {
-        let a: u32x4 = u32x4::new(1, 3, 4, 9);
-        let b: i8x16 = i8x16::new(1, 21, 31, 14, 5, 6, 17, 8, 9, 13, 15, 12, 13, 19, 20, 16);
-        let c: i8x16 = i8x16::new(12, 22, 3, 4, 5, 56, 7, 8, 91, 10, 11, 15, 13, 14, 17, 16);
-        let e: u32x4 = u32x4::new(1, 2, 3, 4);
+        let a = u32x4::new(1, 3, 4, 0xffff0000);
+        let b = u8x16::new(1, 21, 31, 14, 5, 6, 128, 8, 9, 13, 15, 12, 13, 255, 20, 16);
+        let c = u8x16::new(12, 22, 3, 4, 255, 56, 7, 8, 91, 10, 128, 15, 13, 14, 17, 16);
+        let e = u32x4::new(3195, 6935, 18354, 4294909144);
         let r: u32x4 = transmute(vmmlaq_u32(transmute(a), transmute(b), transmute(c)));
         assert_eq!(r, e);
     }
 
     #[simd_test(enable = "neon,i8mm")]
     unsafe fn test_vusmmlaq_s32() {
-        let a: i32x4 = i32x4::new(1, 3, 4, 9);
-        let b: i8x16 = i8x16::new(1, 21, 31, 14, 5, 6, 17, 8, 9, 13, 15, 12, 13, 19, 20, 16);
-        let c: i8x16 = i8x16::new(12, 22, 3, 4, 5, 56, 7, 8, 91, 10, 11, 15, 13, 14, 17, 16);
-        let e: i32x4 = i32x4::new(1, 2, 3, 4);
+        let a = i32x4::new(1, 3, 4, -0x10000);
+        let b = u8x16::new(1, 21, 31, 14, 5, 6, 128, 8, 9, 13, 15, 12, 13, 255, 20, 16);
+        let c = i8x16::new(12, 22, 3, 4, -1, 56, 7, 8, 91, 10, -128, 15, 13, 14, 17, 16);
+        let e = i32x4::new(1915, -1001, 15026, -61992);
         let r: i32x4 = transmute(vusmmlaq_s32(transmute(a), transmute(b), transmute(c)));
         assert_eq!(r, e);
     }
diff --git a/library/stdarch/crates/core_arch/src/lib.rs b/library/stdarch/crates/core_arch/src/lib.rs
index 9240d0e84..5a9727a0a 100644
--- a/library/stdarch/crates/core_arch/src/lib.rs
+++ b/library/stdarch/crates/core_arch/src/lib.rs
@@ -19,6 +19,7 @@
     doc_cfg,
     tbm_target_feature,
     sse4a_target_feature,
+    riscv_target_feature,
     arm_target_feature,
     cmpxchg16b_target_feature,
     avx512_target_feature,
@@ -30,8 +31,8 @@
     f16c_target_feature,
     allow_internal_unstable,
     decl_macro,
-    bench_black_box,
-    asm_const
+    asm_const,
+    target_feature_11
 )]
 #![cfg_attr(test, feature(test, abi_vectorcall))]
 #![deny(clippy::missing_inline_in_public_items)]
diff --git a/library/stdarch/crates/core_arch/src/macros.rs b/library/stdarch/crates/core_arch/src/macros.rs
index 1e6a3f405..1c917c52b 100644
--- a/library/stdarch/crates/core_arch/src/macros.rs
+++ b/library/stdarch/crates/core_arch/src/macros.rs
@@ -101,11 +101,11 @@ macro_rules! simd_shuffle2 {
             const IDX: [u32; 2] = $idx;
         }
 
-        simd_shuffle2($x, $y, ConstParam::<$($imm),+>::IDX)
+        simd_shuffle($x, $y, ConstParam::<$($imm),+>::IDX)
     }};
     ($x:expr, $y:expr, $idx:expr $(,)?) => {{
         const IDX: [u32; 2] = $idx;
-        simd_shuffle2($x, $y, IDX)
+        simd_shuffle($x, $y, IDX)
     }};
 }
 
@@ -117,11 +117,11 @@ macro_rules! simd_shuffle4 {
             const IDX: [u32; 4] = $idx;
         }
 
-        simd_shuffle4($x, $y, ConstParam::<$($imm),+>::IDX)
+        simd_shuffle($x, $y, ConstParam::<$($imm),+>::IDX)
     }};
     ($x:expr, $y:expr, $idx:expr $(,)?) => {{
         const IDX: [u32; 4] = $idx;
-        simd_shuffle4($x, $y, IDX)
+        simd_shuffle($x, $y, IDX)
     }};
 }
 
@@ -133,11 +133,11 @@ macro_rules! simd_shuffle8 {
             const IDX: [u32; 8] = $idx;
         }
 
-        simd_shuffle8($x, $y, ConstParam::<$($imm),+>::IDX)
+        simd_shuffle($x, $y, ConstParam::<$($imm),+>::IDX)
     }};
     ($x:expr, $y:expr, $idx:expr $(,)?) => {{
         const IDX: [u32; 8] = $idx;
-        simd_shuffle8($x, $y, IDX)
+        simd_shuffle($x, $y, IDX)
     }};
 }
 
@@ -149,11 +149,11 @@ macro_rules! simd_shuffle16 {
             const IDX: [u32; 16] = $idx;
         }
 
-        simd_shuffle16($x, $y, ConstParam::<$($imm),+>::IDX)
+        simd_shuffle($x, $y, ConstParam::<$($imm),+>::IDX)
     }};
     ($x:expr, $y:expr, $idx:expr $(,)?) => {{
         const IDX: [u32; 16] = $idx;
-        simd_shuffle16($x, $y, IDX)
+        simd_shuffle($x, $y, IDX)
     }};
 }
 
@@ -165,11 +165,11 @@ macro_rules! simd_shuffle32 {
             const IDX: [u32; 32] = $idx;
         }
 
-        simd_shuffle32($x, $y, ConstParam::<$($imm),+>::IDX)
+        simd_shuffle($x, $y, ConstParam::<$($imm),+>::IDX)
     }};
     ($x:expr, $y:expr, $idx:expr $(,)?) => {{
         const IDX: [u32; 32] = $idx;
-        simd_shuffle32($x, $y, IDX)
+        simd_shuffle($x, $y, IDX)
     }};
 }
 
@@ -181,10 +181,10 @@ macro_rules! simd_shuffle64 {
             const IDX: [u32; 64] = $idx;
         }
 
-        simd_shuffle64($x, $y, ConstParam::<$($imm),+>::IDX)
+        simd_shuffle($x, $y, ConstParam::<$($imm),+>::IDX)
     }};
     ($x:expr, $y:expr, $idx:expr $(,)?) => {{
         const IDX: [u32; 64] = $idx;
-        simd_shuffle64($x, $y, IDX)
+        simd_shuffle($x, $y, IDX)
     }};
 }
diff --git a/library/stdarch/crates/core_arch/src/mod.rs b/library/stdarch/crates/core_arch/src/mod.rs
index 20751eeec..2f7af22cb 100644
--- a/library/stdarch/crates/core_arch/src/mod.rs
+++ b/library/stdarch/crates/core_arch/src/mod.rs
@@ -3,6 +3,9 @@
 #[macro_use]
 mod macros;
 
+#[cfg(any(target_arch = "riscv32", target_arch = "riscv64", doc))]
+mod riscv_shared;
+
 #[cfg(any(target_arch = "arm", target_arch = "aarch64", doc))]
 mod arm_shared;
 
@@ -276,10 +279,6 @@ mod aarch64;
 #[doc(cfg(any(target_arch = "arm")))]
 mod arm;
 
-#[cfg(any(target_arch = "riscv32", target_arch = "riscv64", doc))]
-#[doc(cfg(any(target_arch = "riscv32", target_arch = "riscv64")))]
-mod riscv_shared;
-
 #[cfg(any(target_arch = "riscv64", doc))]
 #[doc(cfg(any(target_arch = "riscv64")))]
 mod riscv64;
diff --git a/library/stdarch/crates/core_arch/src/riscv_shared/mod.rs b/library/stdarch/crates/core_arch/src/riscv_shared/mod.rs
index 347735df1..0e35fe1f1 100644
--- a/library/stdarch/crates/core_arch/src/riscv_shared/mod.rs
+++ b/library/stdarch/crates/core_arch/src/riscv_shared/mod.rs
@@ -1,4 +1,7 @@
 //! Shared RISC-V intrinsics
+mod p;
+
+pub use p::*;
 
 use crate::arch::asm;
 
@@ -469,6 +472,17 @@ pub unsafe fn hinval_gvma_vmid(vmid: usize) {
     asm!(".insn r 0x73, 0, 0x33, x0, x0, {}", in(reg) vmid, options(nostack))
 }
 
+/// Invalidate hypervisor translation cache for all virtual machines and guest physical addresses
+///
+/// This instruction invalidates any address-translation cache entries that an
+/// `HFENCE.GVMA` instruction with the same values of `gaddr` and `vmid` would invalidate.
+///
+/// This fence specifies all guest physical addresses and all virtual machines.
+#[inline]
+pub unsafe fn hinval_gvma_all() {
+    asm!(".insn r 0x73, 0, 0x33, x0, x0, x0", options(nostack))
+}
+
 /// Reads the floating-point control and status register `fcsr`
 ///
 /// Register `fcsr` is a 32-bit read/write register that selects the dynamic rounding mode
@@ -574,17 +588,6 @@ pub fn fsflags(value: u32) -> u32 {
     original
 }
 
-/// Invalidate hypervisor translation cache for all virtual machines and guest physical addresses
-///
-/// This instruction invalidates any address-translation cache entries that an
-/// `HFENCE.GVMA` instruction with the same values of `gaddr` and `vmid` would invalidate.
-///
-/// This fence specifies all guest physical addresses and all virtual machines.
-#[inline]
-pub unsafe fn hinval_gvma_all() {
-    asm!(".insn r 0x73, 0, 0x33, x0, x0, x0", options(nostack))
-}
-
 /// `P0` transformation function as is used in the SM3 hash algorithm
 ///
 /// This function is included in `Zksh` extension. It's defined as:
@@ -602,12 +605,10 @@ pub unsafe fn hinval_gvma_all() {
 /// According to RISC-V Cryptography Extensions, Volume I, the execution latency of
 /// this instruction must always be independent from the data it operates on.
 #[inline]
+#[target_feature(enable = "zksh")]
 pub fn sm3p0(x: u32) -> u32 {
     let ans: u32;
-    unsafe {
-        // asm!("sm3p0 {}, {}", out(reg) ans, in(reg) x, options(nomem, nostack))
-        asm!(".insn i 0x13, 0x1, {}, {}, 0x108", out(reg) ans, in(reg) x, options(nomem, nostack))
-    };
+    unsafe { asm!("sm3p0 {}, {}", lateout(reg) ans, in(reg) x, options(pure, nomem, nostack)) };
     ans
 }
 
@@ -634,12 +635,10 @@ pub fn sm3p0(x: u32) -> u32 {
 /// According to RISC-V Cryptography Extensions, Volume I, the execution latency of
 /// this instruction must always be independent from the data it operates on.
 #[inline]
+#[target_feature(enable = "zksh")]
 pub fn sm3p1(x: u32) -> u32 {
     let ans: u32;
-    unsafe {
-        // asm!("sm3p1 {}, {}", out(reg) ans, in(reg) x, options(nomem, nostack))
-        asm!(".insn i 0x13, 0x1, {}, {}, 0x109", out(reg) ans, in(reg) x, options(nomem, nostack))
-    };
+    unsafe { asm!("sm3p1 {}, {}", lateout(reg) ans, in(reg) x, options(pure, nomem, nostack)) };
     ans
 }
 
@@ -674,33 +673,28 @@ pub fn sm3p1(x: u32) -> u32 {
 /// It can be implemented by `sm4ed` instruction like:
 ///
 /// ```no_run
+/// # #[cfg(any(target_arch = "riscv32", target_arch = "riscv64"))]
+/// # fn round_function(x0: u32, x1: u32, x2: u32, x3: u32, rk: u32) -> u32 {
+/// # #[cfg(target_arch = "riscv32")] use core::arch::riscv32::sm4ed;
+/// # #[cfg(target_arch = "riscv64")] use core::arch::riscv64::sm4ed;
 /// let a = x1 ^ x2 ^ x3 ^ rk;
 /// let c0 = sm4ed::<0>(x0, a);
 /// let c1 = sm4ed::<1>(c0, a); // c1 represents c[0..=1], etc.
 /// let c2 = sm4ed::<2>(c1, a);
 /// let c3 = sm4ed::<3>(c2, a);
 /// return c3; // c3 represents c[0..=3]
+/// # }
 /// ```
 ///
 /// According to RISC-V Cryptography Extensions, Volume I, the execution latency of
 /// this instruction must always be independent from the data it operates on.
+#[inline]
+#[target_feature(enable = "zksed")]
 pub fn sm4ed<const BS: u8>(x: u32, a: u32) -> u32 {
     static_assert!(BS: u8 where BS <= 3);
     let ans: u32;
-    match BS {
-        0 => unsafe {
-            asm!(".insn r 0x33, 0, 0x18, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) a, options(nomem, nostack))
-        },
-        1 => unsafe {
-            asm!(".insn r 0x33, 0, 0x38, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) a, options(nomem, nostack))
-        },
-        2 => unsafe {
-            asm!(".insn r 0x33, 0, 0x58, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) a, options(nomem, nostack))
-        },
-        3 => unsafe {
-            asm!(".insn r 0x33, 0, 0x78, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) a, options(nomem, nostack))
-        },
-        _ => unreachable!(),
+    unsafe {
+        asm!("sm4ed {}, {}, {}, {}", lateout(reg) ans, in(reg) x, in(reg) a, const BS, options(pure, nomem, nostack))
     };
     ans
 }
@@ -739,33 +733,28 @@ pub fn sm4ed<const BS: u8>(x: u32, a: u32) -> u32 {
 /// Hence, the key schedule operation can be implemented by `sm4ks` instruction like:
 ///
 /// ```no_run
+/// # #[cfg(any(target_arch = "riscv32", target_arch = "riscv64"))]
+/// # fn key_schedule(k0: u32, k1: u32, k2: u32, k3: u32, ck_i: u32) -> u32 {
+/// # #[cfg(target_arch = "riscv32")] use core::arch::riscv32::sm4ks;
+/// # #[cfg(target_arch = "riscv64")] use core::arch::riscv64::sm4ks;
 /// let k = k1 ^ k2 ^ k3 ^ ck_i;
 /// let c0 = sm4ks::<0>(k0, k);
 /// let c1 = sm4ks::<1>(c0, k); // c1 represents c[0..=1], etc.
 /// let c2 = sm4ks::<2>(c1, k);
 /// let c3 = sm4ks::<3>(c2, k);
 /// return c3; // c3 represents c[0..=3]
+/// # }
 /// ```
 ///
 /// According to RISC-V Cryptography Extensions, Volume I, the execution latency of
 /// this instruction must always be independent from the data it operates on.
+#[inline]
+#[target_feature(enable = "zksed")]
 pub fn sm4ks<const BS: u8>(x: u32, k: u32) -> u32 {
     static_assert!(BS: u8 where BS <= 3);
     let ans: u32;
-    match BS {
-        0 => unsafe {
-            asm!(".insn r 0x33, 0, 0x1A, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) k, options(nomem, nostack))
-        },
-        1 => unsafe {
-            asm!(".insn r 0x33, 0, 0x3A, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) k, options(nomem, nostack))
-        },
-        2 => unsafe {
-            asm!(".insn r 0x33, 0, 0x5A, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) k, options(nomem, nostack))
-        },
-        3 => unsafe {
-            asm!(".insn r 0x33, 0, 0x7A, {}, {}, {}", out(reg) ans, in(reg) x, in(reg) k, options(nomem, nostack))
-        },
-        _ => unreachable!(),
+    unsafe {
+        asm!("sm4ks {}, {}, {}, {}", lateout(reg) ans, in(reg) x, in(reg) k, const BS, options(pure, nomem, nostack))
     };
     ans
 }
diff --git a/library/stdarch/crates/core_arch/src/riscv_shared/p.rs b/library/stdarch/crates/core_arch/src/riscv_shared/p.rs
new file mode 100644
index 000000000..a26044aee
--- /dev/null
+++ b/library/stdarch/crates/core_arch/src/riscv_shared/p.rs
@@ -0,0 +1,1061 @@
+//! RISC-V Packed SIMD intrinsics; shared part.
+//!
+//! RV64 only part is placed in riscv64 folder.
+use crate::arch::asm;
+
+/// Adds packed 16-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn add16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x20, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the sum of packed 16-bit signed numbers, dropping least bits
+#[inline]
+pub fn radd16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x00, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the sum of packed 16-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn uradd16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x10, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds packed 16-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn kadd16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x08, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds packed 16-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn ukadd16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x18, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts packed 16-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn sub16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x21, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the subtraction result of packed 16-bit signed numbers, dropping least bits
+#[inline]
+pub fn rsub16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x01, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the subtraction result of packed 16-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn ursub16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x11, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts packed 16-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn ksub16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x09, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts packed 16-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn uksub16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x19, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross adds and subtracts packed 16-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn cras16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x22, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross halves of adds and subtracts packed 16-bit signed numbers, dropping least bits
+#[inline]
+pub fn rcras16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x02, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross halves of adds and subtracts packed 16-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn urcras16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x12, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross adds and subtracts packed 16-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn kcras16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross adds and subtracts packed 16-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn ukcras16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x1A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross subtracts and adds packed 16-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn crsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x23, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross halves of subtracts and adds packed 16-bit signed numbers, dropping least bits
+#[inline]
+pub fn rcrsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x03, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross halves of subtracts and adds packed 16-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn urcrsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x13, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross subtracts and adds packed 16-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn kcrsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Cross subtracts and adds packed 16-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn ukcrsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x1B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight adds and subtracts packed 16-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn stas16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x7A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight halves of adds and subtracts packed 16-bit signed numbers, dropping least bits
+#[inline]
+pub fn rstas16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x5A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight halves of adds and subtracts packed 16-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn urstas16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x6A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight adds and subtracts packed 16-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn kstas16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x62, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight adds and subtracts packed 16-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn ukstas16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x72, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight subtracts and adds packed 16-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn stsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x7B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight halves of subtracts and adds packed 16-bit signed numbers, dropping least bits
+#[inline]
+pub fn rstsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x5B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight halves of subtracts and adds packed 16-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn urstsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x6B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight subtracts and adds packed 16-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn kstsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x63, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Straight subtracts and adds packed 16-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn ukstsa16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x73, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds packed 8-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn add8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x24, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the sum of packed 8-bit signed numbers, dropping least bits
+#[inline]
+pub fn radd8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x04, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the sum of packed 8-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn uradd8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x14, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds packed 8-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn kadd8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0C, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds packed 8-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn ukadd8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x1C, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts packed 8-bit signed numbers, discarding overflow bits
+#[inline]
+pub fn sub8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x25, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the subtraction result of packed 8-bit signed numbers, dropping least bits
+#[inline]
+pub fn rsub8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x05, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Halves the subtraction result of packed 8-bit unsigned numbers, dropping least bits
+#[inline]
+pub fn ursub8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x15, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts packed 8-bit signed numbers, saturating at the numeric bounds
+#[inline]
+pub fn ksub8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0D, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts packed 8-bit unsigned numbers, saturating at the numeric bounds
+#[inline]
+pub fn uksub8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x1D, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Arithmetic right shift packed 16-bit elements without rounding up
+#[inline]
+pub fn sra16(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x28, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Arithmetic right shift packed 16-bit elements with rounding up
+#[inline]
+pub fn sra16u(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x30, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical right shift packed 16-bit elements without rounding up
+#[inline]
+pub fn srl16(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x29, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical right shift packed 16-bit elements with rounding up
+#[inline]
+pub fn srl16u(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x31, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical left shift packed 16-bit elements, discarding overflow bits
+#[inline]
+pub fn sll16(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x2A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical left shift packed 16-bit elements, saturating at the numeric bounds
+#[inline]
+pub fn ksll16(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x32, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical saturating left then arithmetic right shift packed 16-bit elements
+#[inline]
+pub fn kslra16(a: usize, b: i32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x2B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical saturating left then arithmetic right shift packed 16-bit elements
+#[inline]
+pub fn kslra16u(a: usize, b: i32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x33, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Arithmetic right shift packed 8-bit elements without rounding up
+#[inline]
+pub fn sra8(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x2C, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Arithmetic right shift packed 8-bit elements with rounding up
+#[inline]
+pub fn sra8u(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x34, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical right shift packed 8-bit elements without rounding up
+#[inline]
+pub fn srl8(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x2D, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical right shift packed 8-bit elements with rounding up
+#[inline]
+pub fn srl8u(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x35, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical left shift packed 8-bit elements, discarding overflow bits
+#[inline]
+pub fn sll8(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x2E, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical left shift packed 8-bit elements, saturating at the numeric bounds
+#[inline]
+pub fn ksll8(a: usize, b: u32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x36, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical saturating left then arithmetic right shift packed 8-bit elements
+#[inline]
+pub fn kslra8(a: usize, b: i32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x2F, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Logical saturating left then arithmetic right shift packed 8-bit elements
+#[inline]
+pub fn kslra8u(a: usize, b: i32) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x37, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare equality for packed 16-bit elements
+#[inline]
+pub fn cmpeq16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x26, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 16-bit packed signed integers are less than the others
+#[inline]
+pub fn scmplt16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x06, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 16-bit packed signed integers are less than or equal to the others
+#[inline]
+pub fn scmple16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0E, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 16-bit packed unsigned integers are less than the others
+#[inline]
+pub fn ucmplt16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x16, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 16-bit packed unsigned integers are less than or equal to the others
+#[inline]
+pub fn ucmple16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x1E, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare equality for packed 8-bit elements
+#[inline]
+pub fn cmpeq8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x27, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 8-bit packed signed integers are less than the others
+#[inline]
+pub fn scmplt8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x07, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 8-bit packed signed integers are less than or equal to the others
+#[inline]
+pub fn scmple8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0F, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 8-bit packed unsigned integers are less than the others
+#[inline]
+pub fn ucmplt8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x17, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Compare whether 8-bit packed unsigned integers are less than or equal to the others
+#[inline]
+pub fn ucmple8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x1F, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get minimum values from 16-bit packed signed integers
+#[inline]
+pub fn smin16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x40, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get minimum values from 16-bit packed unsigned integers
+#[inline]
+pub fn umin16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x48, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get maximum values from 16-bit packed signed integers
+#[inline]
+pub fn smax16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x41, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get maximum values from 16-bit packed unsigned integers
+#[inline]
+pub fn umax16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x49, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/* todo: sclip16, uclip16 */
+
+/// Compute the absolute value of packed 16-bit signed integers
+#[inline]
+pub fn kabs16(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAD1", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Count the number of redundant sign bits of the packed 16-bit elements
+#[inline]
+pub fn clrs16(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAE8", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Count the number of leading zero bits of the packed 16-bit elements
+#[inline]
+pub fn clz16(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAE9", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Swap the 16-bit halfwords within each 32-bit word of a register
+#[inline]
+pub fn swap16(a: usize) -> usize {
+    let value: usize;
+    // this instruction is an alias for `pkbt rd, rs1, rs1`.
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x0F, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get minimum values from 8-bit packed signed integers
+#[inline]
+pub fn smin8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x44, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get minimum values from 8-bit packed unsigned integers
+#[inline]
+pub fn umin8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x4C, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get maximum values from 8-bit packed signed integers
+#[inline]
+pub fn smax8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x45, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Get maximum values from 8-bit packed unsigned integers
+#[inline]
+pub fn umax8(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x4D, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/* todo: sclip8, uclip8 */
+
+/// Compute the absolute value of packed 8-bit signed integers
+#[inline]
+pub fn kabs8(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAD0", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Count the number of redundant sign bits of the packed 8-bit elements
+#[inline]
+pub fn clrs8(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAE0", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Count the number of leading zero bits of the packed 8-bit elements
+#[inline]
+pub fn clz8(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAE1", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Swap the 8-bit bytes within each 16-bit halfword of a register.
+#[inline]
+pub fn swap8(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAD8", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack first and zeroth into two 16-bit signed halfwords in each 32-bit chunk
+#[inline]
+pub fn sunpkd810(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAC8", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack second and zeroth into two 16-bit signed halfwords in each 32-bit chunk
+#[inline]
+pub fn sunpkd820(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAC9", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack third and zeroth into two 16-bit signed halfwords in each 32-bit chunk
+#[inline]
+pub fn sunpkd830(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xACA", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack third and first into two 16-bit signed halfwords in each 32-bit chunk
+#[inline]
+pub fn sunpkd831(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xACB", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack third and second into two 16-bit signed halfwords in each 32-bit chunk
+#[inline]
+pub fn sunpkd832(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAD3", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack first and zeroth into two 16-bit unsigned halfwords in each 32-bit chunk
+#[inline]
+pub fn zunpkd810(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xACC", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack second and zeroth into two 16-bit unsigned halfwords in each 32-bit chunk
+#[inline]
+pub fn zunpkd820(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xACD", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack third and zeroth into two 16-bit unsigned halfwords in each 32-bit chunk
+#[inline]
+pub fn zunpkd830(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xACE", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack third and first into two 16-bit unsigned halfwords in each 32-bit chunk
+#[inline]
+pub fn zunpkd831(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xACF", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Unpack third and second into two 16-bit unsigned halfwords in each 32-bit chunk
+#[inline]
+pub fn zunpkd832(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAD7", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+// todo: pkbb16, pktt16
+
+/// Pack two 16-bit data from bottom and top half from 32-bit chunks
+#[inline]
+pub fn pkbt16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x1, 0x0F, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Pack two 16-bit data from top and bottom half from 32-bit chunks
+#[inline]
+pub fn pktb16(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x1, 0x1F, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Count the number of redundant sign bits of the packed 32-bit elements
+#[inline]
+pub fn clrs32(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAF8", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Count the number of leading zero bits of the packed 32-bit elements
+#[inline]
+pub fn clz32(a: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn i 0x77, 0x0, {}, {}, 0xAF9", lateout(reg) value, in(reg) a, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Calculate the sum of absolute difference of unsigned 8-bit data elements
+#[inline]
+pub fn pbsad(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x7E, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Calculate and accumulate the sum of absolute difference of unsigned 8-bit data elements
+#[inline]
+pub fn pbsada(t: usize, a: usize, b: usize) -> usize {
+    let mut value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x7F, {}, {}, {}", inlateout(reg) t => value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Multiply signed 8-bit elements and add 16-bit elements on results for packed 32-bit chunks
+#[inline]
+pub fn smaqa(t: usize, a: usize, b: usize) -> usize {
+    let mut value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x64, {}, {}, {}", inlateout(reg) t => value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Multiply unsigned 8-bit elements and add 16-bit elements on results for packed 32-bit chunks
+#[inline]
+pub fn umaqa(t: usize, a: usize, b: usize) -> usize {
+    let mut value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x66, {}, {}, {}", inlateout(reg) t => value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Multiply signed to unsigned 8-bit and add 16-bit elements on results for packed 32-bit chunks
+#[inline]
+pub fn smaqasu(t: usize, a: usize, b: usize) -> usize {
+    let mut value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x0, 0x65, {}, {}, {}", inlateout(reg) t => value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds signed lower 16-bit content of two registers with Q15 saturation
+#[inline]
+pub fn kaddh(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x1, 0x02, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts signed lower 16-bit content of two registers with Q15 saturation
+#[inline]
+pub fn ksubh(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x1, 0x03, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Adds signed lower 16-bit content of two registers with U16 saturation
+#[inline]
+pub fn ukaddh(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x1, 0x0A, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
+
+/// Subtracts signed lower 16-bit content of two registers with U16 saturation
+#[inline]
+pub fn uksubh(a: usize, b: usize) -> usize {
+    let value: usize;
+    unsafe {
+        asm!(".insn r 0x77, 0x1, 0x0B, {}, {}, {}", lateout(reg) value, in(reg) a, in(reg) b, options(pure, nomem, nostack))
+    }
+    value
+}
diff --git a/library/stdarch/crates/core_arch/src/simd_llvm.rs b/library/stdarch/crates/core_arch/src/simd_llvm.rs
index 1970e5c69..decdecaaf 100644
--- a/library/stdarch/crates/core_arch/src/simd_llvm.rs
+++ b/library/stdarch/crates/core_arch/src/simd_llvm.rs
@@ -9,13 +9,7 @@ extern "platform-intrinsic" {
     pub fn simd_gt<T, U>(x: T, y: T) -> U;
     pub fn simd_ge<T, U>(x: T, y: T) -> U;
 
-    pub fn simd_shuffle2<T, U>(x: T, y: T, idx: [u32; 2]) -> U;
-    pub fn simd_shuffle4<T, U>(x: T, y: T, idx: [u32; 4]) -> U;
-    pub fn simd_shuffle8<T, U>(x: T, y: T, idx: [u32; 8]) -> U;
-    pub fn simd_shuffle16<T, U>(x: T, y: T, idx: [u32; 16]) -> U;
-    pub fn simd_shuffle32<T, U>(x: T, y: T, idx: [u32; 32]) -> U;
-    pub fn simd_shuffle64<T, U>(x: T, y: T, idx: [u32; 64]) -> U;
-    pub fn simd_shuffle128<T, U>(x: T, y: T, idx: [u32; 128]) -> U;
+    pub fn simd_shuffle<T, U, V>(x: T, y: T, idx: U) -> V;
 
     #[rustc_const_unstable(feature = "const_simd_insert", issue = "none")]
     pub fn simd_insert<T, U>(x: T, idx: u32, val: U) -> T;
diff --git a/library/stdarch/crates/core_arch/src/x86/avx2.rs b/library/stdarch/crates/core_arch/src/x86/avx2.rs
index 24f9c0301..16add3dbb 100644
--- a/library/stdarch/crates/core_arch/src/x86/avx2.rs
+++ b/library/stdarch/crates/core_arch/src/x86/avx2.rs
@@ -2001,7 +2001,7 @@ pub unsafe fn _mm256_min_epu8(a: __m256i, b: __m256i) -> __m256i {
 #[cfg_attr(test, assert_instr(vpmovmskb))]
 #[stable(feature = "simd_x86", since = "1.27.0")]
 pub unsafe fn _mm256_movemask_epi8(a: __m256i) -> i32 {
-    pmovmskb(a.as_i8x32())
+    simd_bitmask::<_, u32>(a.as_i8x32()) as i32
 }
 
 /// Computes the sum of absolute differences (SADs) of quadruplets of unsigned
@@ -3642,8 +3642,6 @@ extern "C" {
     fn pminud(a: u32x8, b: u32x8) -> u32x8;
     #[link_name = "llvm.x86.avx2.pminu.b"]
     fn pminub(a: u8x32, b: u8x32) -> u8x32;
-    #[link_name = "llvm.x86.avx2.pmovmskb"]
-    fn pmovmskb(a: i8x32) -> i32;
     #[link_name = "llvm.x86.avx2.mpsadbw"]
     fn mpsadbw(a: u8x32, b: u8x32, imm8: i32) -> u16x16;
     #[link_name = "llvm.x86.avx2.pmulhu.w"]
diff --git a/library/stdarch/crates/core_arch/src/x86/cpuid.rs b/library/stdarch/crates/core_arch/src/x86/cpuid.rs
index 6b90295ef..2624e8bdf 100644
--- a/library/stdarch/crates/core_arch/src/x86/cpuid.rs
+++ b/library/stdarch/crates/core_arch/src/x86/cpuid.rs
@@ -62,27 +62,27 @@ pub unsafe fn __cpuid_count(leaf: u32, sub_leaf: u32) -> CpuidResult {
     #[cfg(target_arch = "x86")]
     {
         asm!(
-            "movl %ebx, {0}",
+            "mov {0}, ebx",
             "cpuid",
-            "xchgl %ebx, {0}",
-            lateout(reg) ebx,
-            inlateout("eax") leaf => eax,
-            inlateout("ecx") sub_leaf => ecx,
-            lateout("edx") edx,
-            options(nostack, preserves_flags, att_syntax),
+            "xchg {0}, ebx",
+            out(reg) ebx,
+            inout("eax") leaf => eax,
+            inout("ecx") sub_leaf => ecx,
+            out("edx") edx,
+            options(nostack, preserves_flags),
         );
     }
     #[cfg(target_arch = "x86_64")]
     {
         asm!(
-            "movq %rbx, {0:r}",
+            "mov {0:r}, rbx",
             "cpuid",
-            "xchgq %rbx, {0:r}",
-            lateout(reg) ebx,
-            inlateout("eax") leaf => eax,
-            inlateout("ecx") sub_leaf => ecx,
-            lateout("edx") edx,
-            options(nostack, preserves_flags, att_syntax),
+            "xchg {0:r}, rbx",
+            out(reg) ebx,
+            inout("eax") leaf => eax,
+            inout("ecx") sub_leaf => ecx,
+            out("edx") edx,
+            options(nostack, preserves_flags),
         );
     }
     CpuidResult { eax, ebx, ecx, edx }
diff --git a/library/stdarch/crates/core_arch/src/x86/mod.rs b/library/stdarch/crates/core_arch/src/x86/mod.rs
index 547bfe67d..6b50e95b2 100644
--- a/library/stdarch/crates/core_arch/src/x86/mod.rs
+++ b/library/stdarch/crates/core_arch/src/x86/mod.rs
@@ -306,7 +306,7 @@ types! {
 
     /// 256-bit wide set of 16 'u16' types, x86-specific
     ///
-    /// This type is the same as the `__m128bh` type defined by Intel,
+    /// This type is the same as the `__m256bh` type defined by Intel,
     /// representing a 256-bit SIMD register which internally is consisted of
     /// 16 packed `u16` instances. Its purpose is for bf16 related intrinsic
     /// implementations.
@@ -317,7 +317,7 @@ types! {
 
     /// 512-bit wide set of 32 'u16' types, x86-specific
     ///
-    /// This type is the same as the `__m128bh` type defined by Intel,
+    /// This type is the same as the `__m512bh` type defined by Intel,
     /// representing a 512-bit SIMD register which internally is consisted of
     /// 32 packed `u16` instances. Its purpose is for bf16 related intrinsic
     /// implementations.
diff --git a/library/stdarch/crates/core_arch/src/x86/sse2.rs b/library/stdarch/crates/core_arch/src/x86/sse2.rs
index d82b8641f..3e79b3539 100644
--- a/library/stdarch/crates/core_arch/src/x86/sse2.rs
+++ b/library/stdarch/crates/core_arch/src/x86/sse2.rs
@@ -1378,7 +1378,7 @@ pub unsafe fn _mm_insert_epi16<const IMM8: i32>(a: __m128i, i: i32) -> __m128i {
 #[cfg_attr(test, assert_instr(pmovmskb))]
 #[stable(feature = "simd_x86", since = "1.27.0")]
 pub unsafe fn _mm_movemask_epi8(a: __m128i) -> i32 {
-    pmovmskb(a.as_i8x16())
+    simd_bitmask::<_, u16>(a.as_i8x16()) as u32 as i32
 }
 
 /// Shuffles 32-bit integers in `a` using the control in `IMM8`.
@@ -2856,8 +2856,6 @@ extern "C" {
     fn packssdw(a: i32x4, b: i32x4) -> i16x8;
     #[link_name = "llvm.x86.sse2.packuswb.128"]
     fn packuswb(a: i16x8, b: i16x8) -> u8x16;
-    #[link_name = "llvm.x86.sse2.pmovmskb.128"]
-    fn pmovmskb(a: i8x16) -> i32;
     #[link_name = "llvm.x86.sse2.max.sd"]
     fn maxsd(a: __m128d, b: __m128d) -> __m128d;
     #[link_name = "llvm.x86.sse2.max.pd"]
diff --git a/library/stdarch/crates/core_arch/src/x86/sse3.rs b/library/stdarch/crates/core_arch/src/x86/sse3.rs
index ab0dd38fe..61f8a4e78 100644
--- a/library/stdarch/crates/core_arch/src/x86/sse3.rs
+++ b/library/stdarch/crates/core_arch/src/x86/sse3.rs
@@ -1,11 +1,7 @@
 //! Streaming SIMD Extensions 3 (SSE3)
 
 use crate::{
-    core_arch::{
-        simd::*,
-        simd_llvm::{simd_shuffle2, simd_shuffle4},
-        x86::*,
-    },
+    core_arch::{simd::*, simd_llvm::simd_shuffle, x86::*},
     mem::transmute,
 };
 
diff --git a/library/stdarch/crates/std_detect/Cargo.toml b/library/stdarch/crates/std_detect/Cargo.toml
index 1ca0d9c5d..3a482564e 100644
--- a/library/stdarch/crates/std_detect/Cargo.toml
+++ b/library/stdarch/crates/std_detect/Cargo.toml
@@ -22,7 +22,7 @@ maintenance = { status = "experimental" }
 
 [dependencies]
 libc = { version = "0.2", optional = true, default-features = false }
-cfg-if = "0.1.10"
+cfg-if = "1.0.0"
 
 # When built as part of libstd
 core = { version = "1.0.0", optional = true, package = "rustc-std-workspace-core" }
diff --git a/library/stdarch/crates/std_detect/src/detect/os/linux/aarch64.rs b/library/stdarch/crates/std_detect/src/detect/os/linux/aarch64.rs
index b6a2e5218..6c79ba86d 100644
--- a/library/stdarch/crates/std_detect/src/detect/os/linux/aarch64.rs
+++ b/library/stdarch/crates/std_detect/src/detect/os/linux/aarch64.rs
@@ -23,58 +23,62 @@ pub(crate) fn detect_features() -> cache::Initializer {
 /// The names match those used for cpuinfo.
 ///
 /// [hwcap]: https://github.com/torvalds/linux/blob/master/arch/arm64/include/uapi/asm/hwcap.h
+#[derive(Debug, Default, PartialEq)]
 struct AtHwcap {
-    fp: bool,    // 0
-    asimd: bool, // 1
-    // evtstrm: bool, // 2 No LLVM support
-    aes: bool,     // 3
-    pmull: bool,   // 4
-    sha1: bool,    // 5
-    sha2: bool,    // 6
-    crc32: bool,   // 7
-    atomics: bool, // 8
-    fphp: bool,    // 9
-    asimdhp: bool, // 10
-    // cpuid: bool, // 11 No LLVM support
-    asimdrdm: bool, // 12
-    jscvt: bool,    // 13
-    fcma: bool,     // 14
-    lrcpc: bool,    // 15
-    dcpop: bool,    // 16
-    sha3: bool,     // 17
-    sm3: bool,      // 18
-    sm4: bool,      // 19
-    asimddp: bool,  // 20
-    sha512: bool,   // 21
-    sve: bool,      // 22
-    fhm: bool,      // 23
-    dit: bool,      // 24
-    uscat: bool,    // 25
-    ilrcpc: bool,   // 26
-    flagm: bool,    // 27
-    ssbs: bool,     // 28
-    sb: bool,       // 29
-    paca: bool,     // 30
-    pacg: bool,     // 31
-    dcpodp: bool,   // 32
-    sve2: bool,     // 33
-    sveaes: bool,   // 34
-    // svepmull: bool, // 35 No LLVM support
-    svebitperm: bool, // 36
-    svesha3: bool,    // 37
-    svesm4: bool,     // 38
-    // flagm2: bool, // 39 No LLVM support
-    frint: bool, // 40
-    // svei8mm: bool, // 41 See i8mm feature
-    svef32mm: bool, // 42
-    svef64mm: bool, // 43
-    // svebf16: bool, // 44 See bf16 feature
-    i8mm: bool, // 45
-    bf16: bool, // 46
-    // dgh: bool, // 47 No LLVM support
-    rng: bool, // 48
-    bti: bool, // 49
-    mte: bool, // 50
+    // AT_HWCAP
+    fp: bool,
+    asimd: bool,
+    // evtstrm: No LLVM support.
+    aes: bool,
+    pmull: bool,
+    sha1: bool,
+    sha2: bool,
+    crc32: bool,
+    atomics: bool,
+    fphp: bool,
+    asimdhp: bool,
+    // cpuid: No LLVM support.
+    asimdrdm: bool,
+    jscvt: bool,
+    fcma: bool,
+    lrcpc: bool,
+    dcpop: bool,
+    sha3: bool,
+    sm3: bool,
+    sm4: bool,
+    asimddp: bool,
+    sha512: bool,
+    sve: bool,
+    fhm: bool,
+    dit: bool,
+    uscat: bool,
+    ilrcpc: bool,
+    flagm: bool,
+    ssbs: bool,
+    sb: bool,
+    paca: bool,
+    pacg: bool,
+
+    // AT_HWCAP2
+    dcpodp: bool,
+    sve2: bool,
+    sveaes: bool,
+    // svepmull: No LLVM support.
+    svebitperm: bool,
+    svesha3: bool,
+    svesm4: bool,
+    // flagm2: No LLVM support.
+    frint: bool,
+    // svei8mm: See i8mm feature.
+    svef32mm: bool,
+    svef64mm: bool,
+    // svebf16: See bf16 feature.
+    i8mm: bool,
+    bf16: bool,
+    // dgh: No LLVM support.
+    rng: bool,
+    bti: bool,
+    mte: bool,
 }
 
 impl From<auxvec::AuxVec> for AtHwcap {
@@ -113,25 +117,25 @@ impl From<auxvec::AuxVec> for AtHwcap {
             sb: bit::test(auxv.hwcap, 29),
             paca: bit::test(auxv.hwcap, 30),
             pacg: bit::test(auxv.hwcap, 31),
-            dcpodp: bit::test(auxv.hwcap, 32),
-            sve2: bit::test(auxv.hwcap, 33),
-            sveaes: bit::test(auxv.hwcap, 34),
-            // svepmull: bit::test(auxv.hwcap, 35),
-            svebitperm: bit::test(auxv.hwcap, 36),
-            svesha3: bit::test(auxv.hwcap, 37),
-            svesm4: bit::test(auxv.hwcap, 38),
-            // flagm2: bit::test(auxv.hwcap, 39),
-            frint: bit::test(auxv.hwcap, 40),
-            // svei8mm: bit::test(auxv.hwcap, 41),
-            svef32mm: bit::test(auxv.hwcap, 42),
-            svef64mm: bit::test(auxv.hwcap, 43),
-            // svebf16: bit::test(auxv.hwcap, 44),
-            i8mm: bit::test(auxv.hwcap, 45),
-            bf16: bit::test(auxv.hwcap, 46),
-            // dgh: bit::test(auxv.hwcap, 47),
-            rng: bit::test(auxv.hwcap, 48),
-            bti: bit::test(auxv.hwcap, 49),
-            mte: bit::test(auxv.hwcap, 50),
+            dcpodp: bit::test(auxv.hwcap2, 0),
+            sve2: bit::test(auxv.hwcap2, 1),
+            sveaes: bit::test(auxv.hwcap2, 2),
+            // svepmull: bit::test(auxv.hwcap2, 3),
+            svebitperm: bit::test(auxv.hwcap2, 4),
+            svesha3: bit::test(auxv.hwcap2, 5),
+            svesm4: bit::test(auxv.hwcap2, 6),
+            // flagm2: bit::test(auxv.hwcap2, 7),
+            frint: bit::test(auxv.hwcap2, 8),
+            // svei8mm: bit::test(auxv.hwcap2, 9),
+            svef32mm: bit::test(auxv.hwcap2, 10),
+            svef64mm: bit::test(auxv.hwcap2, 11),
+            // svebf16: bit::test(auxv.hwcap2, 12),
+            i8mm: bit::test(auxv.hwcap2, 13),
+            bf16: bit::test(auxv.hwcap2, 14),
+            // dgh: bit::test(auxv.hwcap2, 15),
+            rng: bit::test(auxv.hwcap2, 16),
+            bti: bit::test(auxv.hwcap2, 17),
+            mte: bit::test(auxv.hwcap2, 18),
         }
     }
 }
@@ -288,3 +292,86 @@ impl AtHwcap {
         value
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[cfg(feature = "std_detect_file_io")]
+    mod auxv_from_file {
+        use super::auxvec::auxv_from_file;
+        use super::*;
+        // The baseline hwcaps used in the (artificial) auxv test files.
+        fn baseline_hwcaps() -> AtHwcap {
+            AtHwcap {
+                fp: true,
+                asimd: true,
+                aes: true,
+                pmull: true,
+                sha1: true,
+                sha2: true,
+                crc32: true,
+                atomics: true,
+                fphp: true,
+                asimdhp: true,
+                asimdrdm: true,
+                lrcpc: true,
+                dcpop: true,
+                asimddp: true,
+                ssbs: true,
+                ..AtHwcap::default()
+            }
+        }
+
+        #[test]
+        fn linux_empty_hwcap2_aarch64() {
+            let file = concat!(
+                env!("CARGO_MANIFEST_DIR"),
+                "/src/detect/test_data/linux-empty-hwcap2-aarch64.auxv"
+            );
+            println!("file: {}", file);
+            let v = auxv_from_file(file).unwrap();
+            println!("HWCAP : 0x{:0x}", v.hwcap);
+            println!("HWCAP2: 0x{:0x}", v.hwcap2);
+            assert_eq!(AtHwcap::from(v), baseline_hwcaps());
+        }
+        #[test]
+        fn linux_no_hwcap2_aarch64() {
+            let file = concat!(
+                env!("CARGO_MANIFEST_DIR"),
+                "/src/detect/test_data/linux-no-hwcap2-aarch64.auxv"
+            );
+            println!("file: {}", file);
+            let v = auxv_from_file(file).unwrap();
+            println!("HWCAP : 0x{:0x}", v.hwcap);
+            println!("HWCAP2: 0x{:0x}", v.hwcap2);
+            assert_eq!(AtHwcap::from(v), baseline_hwcaps());
+        }
+        #[test]
+        fn linux_hwcap2_aarch64() {
+            let file = concat!(
+                env!("CARGO_MANIFEST_DIR"),
+                "/src/detect/test_data/linux-hwcap2-aarch64.auxv"
+            );
+            println!("file: {}", file);
+            let v = auxv_from_file(file).unwrap();
+            println!("HWCAP : 0x{:0x}", v.hwcap);
+            println!("HWCAP2: 0x{:0x}", v.hwcap2);
+            assert_eq!(
+                AtHwcap::from(v),
+                AtHwcap {
+                    // Some other HWCAP bits.
+                    paca: true,
+                    pacg: true,
+                    // HWCAP2-only bits.
+                    dcpodp: true,
+                    frint: true,
+                    rng: true,
+                    bti: true,
+                    mte: true,
+                    ..baseline_hwcaps()
+                }
+            );
+        }
+    }
+}
diff --git a/library/stdarch/crates/std_detect/src/detect/os/linux/auxvec.rs b/library/stdarch/crates/std_detect/src/detect/os/linux/auxvec.rs
index e6447d0cd..c903903bd 100644
--- a/library/stdarch/crates/std_detect/src/detect/os/linux/auxvec.rs
+++ b/library/stdarch/crates/std_detect/src/detect/os/linux/auxvec.rs
@@ -7,6 +7,7 @@ pub(crate) const AT_NULL: usize = 0;
 pub(crate) const AT_HWCAP: usize = 16;
 /// Key to access the CPU Hardware capabilities 2 bitfield.
 #[cfg(any(
+    target_arch = "aarch64",
     target_arch = "arm",
     target_arch = "powerpc",
     target_arch = "powerpc64"
@@ -21,6 +22,7 @@ pub(crate) const AT_HWCAP2: usize = 26;
 pub(crate) struct AuxVec {
     pub hwcap: usize,
     #[cfg(any(
+        target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "powerpc",
         target_arch = "powerpc64"
@@ -64,13 +66,14 @@ pub(crate) fn auxv() -> Result<AuxVec, ()> {
         if let Ok(hwcap) = getauxval(AT_HWCAP) {
             // Targets with only AT_HWCAP:
             #[cfg(any(
-                target_arch = "aarch64",
                 target_arch = "riscv32",
                 target_arch = "riscv64",
                 target_arch = "mips",
                 target_arch = "mips64"
             ))]
             {
+                // Zero could indicate that no features were detected, but it's also used to
+                // indicate an error. In either case, try the fallback.
                 if hwcap != 0 {
                     return Ok(AuxVec { hwcap });
                 }
@@ -78,13 +81,18 @@ pub(crate) fn auxv() -> Result<AuxVec, ()> {
 
             // Targets with AT_HWCAP and AT_HWCAP2:
             #[cfg(any(
+                target_arch = "aarch64",
                 target_arch = "arm",
                 target_arch = "powerpc",
                 target_arch = "powerpc64"
             ))]
             {
                 if let Ok(hwcap2) = getauxval(AT_HWCAP2) {
-                    if hwcap != 0 && hwcap2 != 0 {
+                    // Zero could indicate that no features were detected, but it's also used to
+                    // indicate an error. In particular, on many platforms AT_HWCAP2 will be
+                    // legitimately zero, since it contains the most recent feature flags. Use the
+                    // fallback only if no features were detected at all.
+                    if hwcap != 0 || hwcap2 != 0 {
                         return Ok(AuxVec { hwcap, hwcap2 });
                     }
                 }
@@ -97,7 +105,6 @@ pub(crate) fn auxv() -> Result<AuxVec, ()> {
     {
         // Targets with only AT_HWCAP:
         #[cfg(any(
-            target_arch = "aarch64",
             target_arch = "riscv32",
             target_arch = "riscv64",
             target_arch = "mips",
@@ -105,6 +112,8 @@ pub(crate) fn auxv() -> Result<AuxVec, ()> {
         ))]
         {
             let hwcap = unsafe { libc::getauxval(AT_HWCAP as libc::c_ulong) as usize };
+            // Zero could indicate that no features were detected, but it's also used to indicate
+            // an error. In either case, try the fallback.
             if hwcap != 0 {
                 return Ok(AuxVec { hwcap });
             }
@@ -112,6 +121,7 @@ pub(crate) fn auxv() -> Result<AuxVec, ()> {
 
         // Targets with AT_HWCAP and AT_HWCAP2:
         #[cfg(any(
+            target_arch = "aarch64",
             target_arch = "arm",
             target_arch = "powerpc",
             target_arch = "powerpc64"
@@ -119,7 +129,11 @@ pub(crate) fn auxv() -> Result<AuxVec, ()> {
         {
             let hwcap = unsafe { libc::getauxval(AT_HWCAP as libc::c_ulong) as usize };
             let hwcap2 = unsafe { libc::getauxval(AT_HWCAP2 as libc::c_ulong) as usize };
-            if hwcap != 0 && hwcap2 != 0 {
+            // Zero could indicate that no features were detected, but it's also used to indicate
+            // an error. In particular, on many platforms AT_HWCAP2 will be legitimately zero,
+            // since it contains the most recent feature flags. Use the fallback only if no
+            // features were detected at all.
+            if hwcap != 0 || hwcap2 != 0 {
                 return Ok(AuxVec { hwcap, hwcap2 });
             }
         }
@@ -158,7 +172,7 @@ fn getauxval(key: usize) -> Result<usize, ()> {
 /// Tries to read the auxiliary vector from the `file`. If this fails, this
 /// function returns `Err`.
 #[cfg(feature = "std_detect_file_io")]
-fn auxv_from_file(file: &str) -> Result<AuxVec, ()> {
+pub(super) fn auxv_from_file(file: &str) -> Result<AuxVec, ()> {
     let file = super::read_file(file)?;
 
     // See <https://github.com/torvalds/linux/blob/v3.19/include/uapi/linux/auxvec.h>.
@@ -181,7 +195,6 @@ fn auxv_from_file(file: &str) -> Result<AuxVec, ()> {
 fn auxv_from_buf(buf: &[usize; 64]) -> Result<AuxVec, ()> {
     // Targets with only AT_HWCAP:
     #[cfg(any(
-        target_arch = "aarch64",
         target_arch = "riscv32",
         target_arch = "riscv64",
         target_arch = "mips",
@@ -198,23 +211,25 @@ fn auxv_from_buf(buf: &[usize; 64]) -> Result<AuxVec, ()> {
     }
     // Targets with AT_HWCAP and AT_HWCAP2:
     #[cfg(any(
+        target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "powerpc",
         target_arch = "powerpc64"
     ))]
     {
         let mut hwcap = None;
-        let mut hwcap2 = None;
+        // For some platforms, AT_HWCAP2 was added recently, so let it default to zero.
+        let mut hwcap2 = 0;
         for el in buf.chunks(2) {
             match el[0] {
                 AT_NULL => break,
                 AT_HWCAP => hwcap = Some(el[1]),
-                AT_HWCAP2 => hwcap2 = Some(el[1]),
+                AT_HWCAP2 => hwcap2 = el[1],
                 _ => (),
             }
         }
 
-        if let (Some(hwcap), Some(hwcap2)) = (hwcap, hwcap2) {
+        if let Some(hwcap) = hwcap {
             return Ok(AuxVec { hwcap, hwcap2 });
         }
     }
@@ -256,7 +271,6 @@ mod tests {
     // FIXME: on mips/mips64 getauxval returns 0, and /proc/self/auxv
     // does not always contain the AT_HWCAP key under qemu.
     #[cfg(any(
-        target_arch = "aarch64",
         target_arch = "arm",
         target_arch = "powerpc",
         target_arch = "powerpc64"
@@ -271,6 +285,7 @@ mod tests {
 
         // Targets with AT_HWCAP and AT_HWCAP2:
         #[cfg(any(
+            target_arch = "aarch64",
             target_arch = "arm",
             target_arch = "powerpc",
             target_arch = "powerpc64"
@@ -305,24 +320,31 @@ mod tests {
             }
 
             #[test]
-            #[should_panic]
             fn linux_macos_vb() {
                 let file = concat!(env!("CARGO_MANIFEST_DIR"), "/src/detect/test_data/macos-virtualbox-linux-x86-4850HQ.auxv");
                 println!("file: {}", file);
+                // The file contains HWCAP but not HWCAP2. In that case, we treat HWCAP2 as zero.
                 let v = auxv_from_file(file).unwrap();
-                // this file is incomplete (contains hwcap but not hwcap2), we
-                // want to fall back to /proc/cpuinfo in this case, so
-                // reading should fail. assert_eq!(v.hwcap, 126614527);
-                // assert_eq!(v.hwcap2, 0);
-                let _ = v;
+                assert_eq!(v.hwcap, 126614527);
+                assert_eq!(v.hwcap2, 0);
             }
         } else if #[cfg(target_arch = "aarch64")] {
             #[test]
-            fn linux_x64() {
-                let file = concat!(env!("CARGO_MANIFEST_DIR"), "/src/detect/test_data/linux-x64-i7-6850k.auxv");
+            fn linux_artificial_aarch64() {
+                let file = concat!(env!("CARGO_MANIFEST_DIR"), "/src/detect/test_data/linux-artificial-aarch64.auxv");
                 println!("file: {}", file);
                 let v = auxv_from_file(file).unwrap();
-                assert_eq!(v.hwcap, 3219913727);
+                assert_eq!(v.hwcap, 0x0123456789abcdef);
+                assert_eq!(v.hwcap2, 0x02468ace13579bdf);
+            }
+            #[test]
+            fn linux_no_hwcap2_aarch64() {
+                let file = concat!(env!("CARGO_MANIFEST_DIR"), "/src/detect/test_data/linux-no-hwcap2-aarch64.auxv");
+                println!("file: {}", file);
+                let v = auxv_from_file(file).unwrap();
+                // An absent HWCAP2 is treated as zero, and does not prevent acceptance of HWCAP.
+                assert_ne!(v.hwcap, 0);
+                assert_eq!(v.hwcap2, 0);
             }
         }
     }
@@ -353,6 +375,7 @@ mod tests {
 
         // Targets with AT_HWCAP and AT_HWCAP2:
         #[cfg(any(
+            target_arch = "aarch64",
             target_arch = "arm",
             target_arch = "powerpc",
             target_arch = "powerpc64"
diff --git a/library/stdarch/crates/std_detect/src/detect/test_data/linux-artificial-aarch64.auxv b/library/stdarch/crates/std_detect/src/detect/test_data/linux-artificial-aarch64.auxv
new file mode 100644
index 000000000..ec826afcf
--- /dev/null
+++ b/library/stdarch/crates/std_detect/src/detect/test_data/linux-artificial-aarch64.auxv
diff --git a/library/stdarch/crates/std_detect/src/detect/test_data/linux-empty-hwcap2-aarch64.auxv b/library/stdarch/crates/std_detect/src/detect/test_data/linux-empty-hwcap2-aarch64.auxv
new file mode 100644
index 000000000..95537b73f
--- /dev/null
+++ b/library/stdarch/crates/std_detect/src/detect/test_data/linux-empty-hwcap2-aarch64.auxv
diff --git a/library/stdarch/crates/std_detect/src/detect/test_data/linux-hwcap2-aarch64.auxv b/library/stdarch/crates/std_detect/src/detect/test_data/linux-hwcap2-aarch64.auxv
new file mode 100644
index 000000000..1d87264b2
--- /dev/null
+++ b/library/stdarch/crates/std_detect/src/detect/test_data/linux-hwcap2-aarch64.auxv
diff --git a/library/stdarch/crates/std_detect/src/detect/test_data/linux-no-hwcap2-aarch64.auxv b/library/stdarch/crates/std_detect/src/detect/test_data/linux-no-hwcap2-aarch64.auxv
new file mode 100644
index 000000000..35f01cc76
--- /dev/null
+++ b/library/stdarch/crates/std_detect/src/detect/test_data/linux-no-hwcap2-aarch64.auxv
diff --git a/library/stdarch/crates/std_detect/src/detect/test_data/linux-x64-i7-6850k.auxv b/library/stdarch/crates/std_detect/src/detect/test_data/linux-x64-i7-6850k.auxv
deleted file mode 100644
index 6afe1b3b4..000000000
--- a/library/stdarch/crates/std_detect/src/detect/test_data/linux-x64-i7-6850k.auxv
+++ /dev/null
diff --git a/library/stdarch/crates/stdarch-test/Cargo.toml b/library/stdarch/crates/stdarch-test/Cargo.toml
index 9ac1057be..012b4e959 100644
--- a/library/stdarch/crates/stdarch-test/Cargo.toml
+++ b/library/stdarch/crates/stdarch-test/Cargo.toml
@@ -10,7 +10,7 @@ simd-test-macro = { path = "../simd-test-macro" }
 cc = "1.0"
 lazy_static = "1.0"
 rustc-demangle = "0.1.8"
-cfg-if = "0.1"
+cfg-if = "1.0"
 
 # We use a crates.io dependency to disassemble wasm binaries to look for
 # instructions for `#[assert_instr]`. Note that we use an `=` dependency here
diff --git a/library/stdarch/crates/stdarch-test/src/lib.rs b/library/stdarch/crates/stdarch-test/src/lib.rs
index 078736c66..eba17771c 100644
--- a/library/stdarch/crates/stdarch-test/src/lib.rs
+++ b/library/stdarch/crates/stdarch-test/src/lib.rs
@@ -3,7 +3,6 @@
 //! This basically just disassembles the current executable and then parses the
 //! output once globally and then provides the `assert` function which makes
 //! assertions about the disassembly of a function.
-#![feature(bench_black_box)] // For black_box
 #![deny(rust_2018_idioms)]
 #![allow(clippy::missing_docs_in_private_items, clippy::print_stdout)]