diff --git a/compiler/rustc_target/src/spec/aarch64_apple_darwin.rs b/compiler/rustc_target/src/spec/aarch64_apple_darwin.rs
index 2c71fb8afeede..ca3550e9278d1 100644
--- a/compiler/rustc_target/src/spec/aarch64_apple_darwin.rs
+++ b/compiler/rustc_target/src/spec/aarch64_apple_darwin.rs
@@ -2,7 +2,7 @@ use crate::spec::{FramePointer, LinkerFlavor, SanitizerSet, Target, TargetOption
 
 pub fn target() -> Target {
     let mut base = super::apple_base::opts("macos");
-    base.cpu = "apple-a12".to_string();
+    base.cpu = "apple-a14".to_string();
     base.max_atomic_width = Some(128);
 
     // FIXME: The leak sanitizer currently fails the tests, see #88132.
diff --git a/library/std/src/sys/windows/thread_local_dtor.rs b/library/std/src/sys/windows/thread_local_dtor.rs
index 7be13bc4b2bc7..25d1c6e8e87c2 100644
--- a/library/std/src/sys/windows/thread_local_dtor.rs
+++ b/library/std/src/sys/windows/thread_local_dtor.rs
@@ -1,4 +1,28 @@
+//! Implements thread-local destructors that are not associated with any
+//! particular data.
+
 #![unstable(feature = "thread_local_internals", issue = "none")]
 #![cfg(target_thread_local)]
 
-pub use crate::sys_common::thread_local_dtor::register_dtor_fallback as register_dtor;
+// Using a per-thread list avoids the problems in synchronizing global state.
+#[thread_local]
+static mut DESTRUCTORS: Vec<(*mut u8, unsafe extern "C" fn(*mut u8))> = Vec::new();
+
+pub unsafe fn register_dtor(t: *mut u8, dtor: unsafe extern "C" fn(*mut u8)) {
+    DESTRUCTORS.push((t, dtor));
+}
+
+/// Runs destructors. This should not be called until thread exit.
+pub unsafe fn run_keyless_dtors() {
+    // Drop all the destructors.
+    //
+    // Note: While this is potentially an infinite loop, it *should* be
+    // the case that this loop always terminates because we provide the
+    // guarantee that a TLS key cannot be set after it is flagged for
+    // destruction.
+    while let Some((ptr, dtor)) = DESTRUCTORS.pop() {
+        (dtor)(ptr);
+    }
+    // We're done so free the memory.
+    DESTRUCTORS = Vec::new();
+}
diff --git a/library/std/src/sys/windows/thread_local_key.rs b/library/std/src/sys/windows/thread_local_key.rs
index 0bc511146654b..ec670238e6f0e 100644
--- a/library/std/src/sys/windows/thread_local_key.rs
+++ b/library/std/src/sys/windows/thread_local_key.rs
@@ -196,6 +196,8 @@ pub static p_thread_callback: unsafe extern "system" fn(c::LPVOID, c::DWORD, c::
 unsafe extern "system" fn on_tls_callback(h: c::LPVOID, dwReason: c::DWORD, pv: c::LPVOID) {
     if dwReason == c::DLL_THREAD_DETACH || dwReason == c::DLL_PROCESS_DETACH {
         run_dtors();
+        #[cfg(target_thread_local)]
+        super::thread_local_dtor::run_keyless_dtors();
     }
 
     // See comments above for what this is doing. Note that we don't need this
diff --git a/src/bootstrap/builder.rs b/src/bootstrap/builder.rs
index d5656f0f37e03..6ba1b1b6036ea 100644
--- a/src/bootstrap/builder.rs
+++ b/src/bootstrap/builder.rs
@@ -482,6 +482,7 @@ impl<'a> Builder<'a> {
                 doc::RustByExample,
                 doc::RustcBook,
                 doc::CargoBook,
+                doc::Clippy,
                 doc::EmbeddedBook,
                 doc::EditionGuide,
             ),
diff --git a/src/bootstrap/doc.rs b/src/bootstrap/doc.rs
index 6f2470b706a64..2804e7119fbc1 100644
--- a/src/bootstrap/doc.rs
+++ b/src/bootstrap/doc.rs
@@ -755,6 +755,7 @@ tool_doc!(
     "src/tools/rustfmt",
     ["rustfmt-nightly", "rustfmt-config_proc_macro"],
 );
+tool_doc!(Clippy, "clippy", "src/tools/clippy", ["clippy_utils"]);
 
 #[derive(Ord, PartialOrd, Debug, Copy, Clone, Hash, PartialEq, Eq)]
 pub struct ErrorIndex {
diff --git a/src/doc/unstable-book/src/library-features/asm.md b/src/doc/unstable-book/src/library-features/asm.md
index bd7234522e1fe..6bf21e8f58b25 100644
--- a/src/doc/unstable-book/src/library-features/asm.md
+++ b/src/doc/unstable-book/src/library-features/asm.md
@@ -257,7 +257,7 @@ unsafe {
 }
 
 println!(
-    "L1 Cache: {}",
+    "L0 Cache: {}",
     ((ebx >> 22) + 1) * (((ebx >> 12) & 0x3ff) + 1) * ((ebx & 0xfff) + 1) * (ecx + 1)
 );
 ```