Auto merge of #113911 - matthiaskrgr:rollup-wk6cr7v, r=matthiaskrgr

Rollup of 5 pull requests

Successful merges:

 - #113380 (style-guide: clean up "must"/"should"/"may")
 - #113723 (Resurrect: rustc_llvm: Add a -Z `print-codegen-stats` option to expose LLVM statistics.)
 - #113780 (Support `--print KIND=PATH` command line syntax)
 - #113810 (Make {Rc,Arc}::allocator associated functions)
 - #113907 (Minor improvements to Windows TLS dtors)

Failed merges:

 - #113392 (style-guide: Some cleanups from the fmt-rfcs repo history)

r? `@ghost`
`@rustbot` modify labels: rollup

library/alloc/src/rc.rs

@@ -661,10 +661,14 @@ impl<T> Rc<T> {
 
 impl<T, A: Allocator> Rc<T, A> {
     /// Returns a reference to the underlying allocator.
+    ///
+    /// Note: this is an associated function, which means that you have
+    /// to call it as `Rc::allocator(&r)` instead of `r.allocator()`. This
+    /// is so that there is no conflict with a method on the inner type.
     #[inline]
     #[unstable(feature = "allocator_api", issue = "32838")]
-    pub fn allocator(&self) -> &A {
-        &self.alloc
+    pub fn allocator(this: &Self) -> &A {
+        &this.alloc
     }
     /// Constructs a new `Rc` in the provided allocator.
     ///

library/alloc/src/sync.rs

@@ -678,10 +678,14 @@ impl<T> Arc<T> {
 
 impl<T, A: Allocator> Arc<T, A> {
     /// Returns a reference to the underlying allocator.
+    ///
+    /// Note: this is an associated function, which means that you have
+    /// to call it as `Arc::allocator(&a)` instead of `a.allocator()`. This
+    /// is so that there is no conflict with a method on the inner type.
     #[inline]
     #[unstable(feature = "allocator_api", issue = "32838")]
-    pub fn allocator(&self) -> &A {
-        &self.alloc
+    pub fn allocator(this: &Self) -> &A {
+        &this.alloc
     }
     /// Constructs a new `Arc<T>` in the provided allocator.
     ///

library/std/src/sys/windows/thread_local_dtor.rs

@@ -4,29 +4,4 @@
 #![unstable(feature = "thread_local_internals", issue = "none")]
 #![cfg(target_thread_local)]
 
-// 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();
-
-// Ensure this can never be inlined because otherwise this may break in dylibs.
-// See #44391.
-#[inline(never)]
-pub unsafe fn register_dtor(t: *mut u8, dtor: unsafe extern "C" fn(*mut u8)) {
-    DESTRUCTORS.push((t, dtor));
-}
-
-#[inline(never)] // See comment above
-/// Runs destructors. This should not be called until thread exit.
-pub unsafe fn run_keyless_dtors() {
-    // Drop all the destructors.
-    //
-    // 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();
-}
+pub use super::thread_local_key::register_keyless_dtor as register_dtor;

library/std/src/sys/windows/thread_local_key.rs

@@ -1,7 +1,7 @@
 use crate::cell::UnsafeCell;
 use crate::ptr;
 use crate::sync::atomic::{
-    AtomicPtr, AtomicU32,
+    AtomicBool, AtomicPtr, AtomicU32,
     Ordering::{AcqRel, Acquire, Relaxed, Release},
 };
 use crate::sys::c;
@@ -9,6 +9,41 @@ use crate::sys::c;
 #[cfg(test)]
 mod tests;
 
+/// An optimization hint. The compiler is often smart enough to know if an atomic
+/// is never set and can remove dead code based on that fact.
+static HAS_DTORS: AtomicBool = AtomicBool::new(false);
+
+// Using a per-thread list avoids the problems in synchronizing global state.
+#[thread_local]
+#[cfg(target_thread_local)]
+static mut DESTRUCTORS: Vec<(*mut u8, unsafe extern "C" fn(*mut u8))> = Vec::new();
+
+// Ensure this can never be inlined because otherwise this may break in dylibs.
+// See #44391.
+#[inline(never)]
+#[cfg(target_thread_local)]
+pub unsafe fn register_keyless_dtor(t: *mut u8, dtor: unsafe extern "C" fn(*mut u8)) {
+    DESTRUCTORS.push((t, dtor));
+    HAS_DTORS.store(true, Relaxed);
+}
+
+#[inline(never)] // See comment above
+#[cfg(target_thread_local)]
+/// Runs destructors. This should not be called until thread exit.
+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();
+}
+
 type Key = c::DWORD;
 type Dtor = unsafe extern "C" fn(*mut u8);
 
@@ -156,6 +191,8 @@ static DTORS: AtomicPtr<StaticKey> = AtomicPtr::new(ptr::null_mut());
 /// Should only be called once per key, otherwise loops or breaks may occur in
 /// the linked list.
 unsafe fn register_dtor(key: &'static StaticKey) {
+    // Ensure this is never run when native thread locals are available.
+    assert_eq!(false, cfg!(target_thread_local));
     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.
@@ -167,6 +204,7 @@ unsafe fn register_dtor(key: &'static StaticKey) {
             Err(new) => head = new,
         }
     }
+    HAS_DTORS.store(true, Release);
 }
 
 // -------------------------------------------------------------------------
@@ -240,10 +278,14 @@ pub static p_thread_callback: unsafe extern "system" fn(c::LPVOID, c::DWORD, c::
 
 #[allow(dead_code, unused_variables)]
 unsafe extern "system" fn on_tls_callback(h: c::LPVOID, dwReason: c::DWORD, pv: c::LPVOID) {
+    if !HAS_DTORS.load(Acquire) {
+        return;
+    }
     if dwReason == c::DLL_THREAD_DETACH || dwReason == c::DLL_PROCESS_DETACH {
+        #[cfg(not(target_thread_local))]
         run_dtors();
         #[cfg(target_thread_local)]
-        super::thread_local_dtor::run_keyless_dtors();
+        run_keyless_dtors();
     }
 
     // See comments above for what this is doing. Note that we don't need this

library/std/src/sys/windows/thread_local_key/tests.rs

@@ -1,3 +1,7 @@
+// This file only tests the thread local key fallback.
+// Windows targets with native thread local support do not use this.
+#![cfg(not(target_thread_local))]
+
 use super::StaticKey;
 use crate::ptr;