Auto merge of #145785 - samueltardieu:rollup-81jw080, r=<try>

Rollup of 7 pull requests

try-job: test-various

Author: rust-bors[bot]

compiler/rustc_middle/src/mir/interpret/allocation/provenance_map.rs

@@ -120,6 +120,17 @@ impl<Prov: Provenance> ProvenanceMap<Prov> {
         }
     }
 
+    /// Gets the provenances of all bytes (including from pointers) in a range.
+    pub fn get_range(
+        &self,
+        cx: &impl HasDataLayout,
+        range: AllocRange,
+    ) -> impl Iterator<Item = Prov> {
+        let ptr_provs = self.range_ptrs_get(range, cx).iter().map(|(_, p)| *p);
+        let byte_provs = self.range_bytes_get(range).iter().map(|(_, (p, _))| *p);
+        ptr_provs.chain(byte_provs)
+    }
+
     /// Attempt to merge per-byte provenance back into ptr chunks, if the right fragments
     /// sit next to each other. Return `false` is that is not possible due to partial pointers.
     pub fn merge_bytes(&mut self, cx: &impl HasDataLayout) -> bool {

library/core/src/char/methods.rs

@@ -1872,28 +1872,33 @@ pub const unsafe fn encode_utf8_raw_unchecked(code: u32, dst: *mut u8) {
     // SAFETY: The caller must guarantee that the buffer pointed to by `dst`
     // is at least `len` bytes long.
     unsafe {
-        match len {
-            1 => {
-                *dst = code as u8;
-            }
-            2 => {
-                *dst = (code >> 6 & 0x1F) as u8 | TAG_TWO_B;
-                *dst.add(1) = (code & 0x3F) as u8 | TAG_CONT;
-            }
-            3 => {
-                *dst = (code >> 12 & 0x0F) as u8 | TAG_THREE_B;
-                *dst.add(1) = (code >> 6 & 0x3F) as u8 | TAG_CONT;
-                *dst.add(2) = (code & 0x3F) as u8 | TAG_CONT;
-            }
-            4 => {
-                *dst = (code >> 18 & 0x07) as u8 | TAG_FOUR_B;
-                *dst.add(1) = (code >> 12 & 0x3F) as u8 | TAG_CONT;
-                *dst.add(2) = (code >> 6 & 0x3F) as u8 | TAG_CONT;
-                *dst.add(3) = (code & 0x3F) as u8 | TAG_CONT;
-            }
-            // SAFETY: `char` always takes between 1 and 4 bytes to encode in UTF-8.
-            _ => crate::hint::unreachable_unchecked(),
+        if len == 1 {
+            *dst = code as u8;
+            return;
+        }
+
+        let last1 = (code >> 0 & 0x3F) as u8 | TAG_CONT;
+        let last2 = (code >> 6 & 0x3F) as u8 | TAG_CONT;
+        let last3 = (code >> 12 & 0x3F) as u8 | TAG_CONT;
+        let last4 = (code >> 18 & 0x3F) as u8 | TAG_FOUR_B;
+
+        if len == 2 {
+            *dst = last2 | TAG_TWO_B;
+            *dst.add(1) = last1;
+            return;
         }
+
+        if len == 3 {
+            *dst = last3 | TAG_THREE_B;
+            *dst.add(1) = last2;
+            *dst.add(2) = last1;
+            return;
+        }
+
+        *dst = last4;
+        *dst.add(1) = last3;
+        *dst.add(2) = last2;
+        *dst.add(3) = last1;
     }
 }
 

library/std/src/sync/lazy_lock.rs

@@ -244,7 +244,11 @@ impl<T, F: FnOnce() -> T> LazyLock<T, F> {
     #[inline]
     #[stable(feature = "lazy_cell", since = "1.80.0")]
     pub fn force(this: &LazyLock<T, F>) -> &T {
-        this.once.call_once(|| {
+        this.once.call_once_force(|state| {
+            if state.is_poisoned() {
+                panic_poisoned();
+            }
+
             // SAFETY: `call_once` only runs this closure once, ever.
             let data = unsafe { &mut *this.data.get() };
             let f = unsafe { ManuallyDrop::take(&mut data.f) };
@@ -257,8 +261,7 @@ impl<T, F: FnOnce() -> T> LazyLock<T, F> {
         // * the closure was called and initialized `value`.
         // * the closure was called and panicked, so this point is never reached.
         // * the closure was not called, but a previous call initialized `value`.
-        // * the closure was not called because the Once is poisoned, so this point
-        //   is never reached.
+        // * the closure was not called because the Once is poisoned, which we handled above.
         // So `value` has definitely been initialized and will not be modified again.
         unsafe { &*(*this.data.get()).value }
     }

library/std/src/sys/fd/unix.rs

@@ -37,10 +37,10 @@ pub struct FileDesc(OwnedFd);
 //
 // On Apple targets however, apparently the 64-bit libc is either buggy or
 // intentionally showing odd behavior by rejecting any read with a size
-// larger than or equal to INT_MAX. To handle both of these the read
-// size is capped on both platforms.
+// larger than INT_MAX. To handle both of these the read size is capped on
+// both platforms.
 const READ_LIMIT: usize = if cfg!(target_vendor = "apple") {
-    libc::c_int::MAX as usize - 1
+    libc::c_int::MAX as usize
 } else {
     libc::ssize_t::MAX as usize
 };

library/std/tests/sync/lazy_lock.rs

@@ -33,16 +33,6 @@ fn lazy_default() {
     assert_eq!(CALLED.load(SeqCst), 1);
 }
 
-#[test]
-#[cfg_attr(not(panic = "unwind"), ignore = "test requires unwinding support")]
-fn lazy_poisoning() {
-    let x: LazyCell<String> = LazyCell::new(|| panic!("kaboom"));
-    for _ in 0..2 {
-        let res = panic::catch_unwind(panic::AssertUnwindSafe(|| x.len()));
-        assert!(res.is_err());
-    }
-}
-
 #[test]
 #[cfg_attr(any(target_os = "emscripten", target_os = "wasi"), ignore)] // no threads
 fn sync_lazy_new() {
@@ -123,16 +113,6 @@ fn static_sync_lazy_via_fn() {
     assert_eq!(xs(), &vec![1, 2, 3]);
 }
 
-#[test]
-#[cfg_attr(not(panic = "unwind"), ignore = "test requires unwinding support")]
-fn sync_lazy_poisoning() {
-    let x: LazyLock<String> = LazyLock::new(|| panic!("kaboom"));
-    for _ in 0..2 {
-        let res = panic::catch_unwind(|| x.len());
-        assert!(res.is_err());
-    }
-}
-
 // Check that we can infer `T` from closure's type.
 #[test]
 fn lazy_type_inference() {
@@ -145,17 +125,6 @@ fn is_sync_send() {
     assert_traits::<LazyLock<String>>();
 }
 
-#[test]
-#[should_panic = "has previously been poisoned"]
-fn lazy_force_mut_panic() {
-    let mut lazy = LazyLock::<String>::new(|| panic!());
-    panic::catch_unwind(panic::AssertUnwindSafe(|| {
-        let _ = LazyLock::force_mut(&mut lazy);
-    }))
-    .unwrap_err();
-    let _ = &*lazy;
-}
-
 #[test]
 fn lazy_force_mut() {
     let s = "abc".to_owned();
@@ -165,3 +134,56 @@ fn lazy_force_mut() {
     p.clear();
     LazyLock::force_mut(&mut lazy);
 }
+
+#[test]
+#[cfg_attr(not(panic = "unwind"), ignore = "test requires unwinding support")]
+fn lazy_poisoning() {
+    let x: LazyCell<String> = LazyCell::new(|| panic!("kaboom"));
+    for _ in 0..2 {
+        let res = panic::catch_unwind(panic::AssertUnwindSafe(|| x.len()));
+        assert!(res.is_err());
+    }
+}
+
+/// Verifies that when a `LazyLock` is poisoned, it panics with the correct error message ("LazyLock
+/// instance has previously been poisoned") instead of the underlying `Once` error message.
+#[test]
+#[cfg_attr(not(panic = "unwind"), ignore = "test requires unwinding support")]
+#[should_panic(expected = "LazyLock instance has previously been poisoned")]
+fn lazy_lock_deref_panic() {
+    let lazy: LazyLock<String> = LazyLock::new(|| panic!("initialization failed"));
+
+    // First access will panic during initialization.
+    let _ = panic::catch_unwind(panic::AssertUnwindSafe(|| {
+        let _ = &*lazy;
+    }));
+
+    // Second access should panic with the poisoned message.
+    let _ = &*lazy;
+}
+
+#[test]
+#[should_panic(expected = "LazyLock instance has previously been poisoned")]
+fn lazy_lock_deref_mut_panic() {
+    let mut lazy: LazyLock<String> = LazyLock::new(|| panic!("initialization failed"));
+
+    // First access will panic during initialization.
+    let _ = panic::catch_unwind(panic::AssertUnwindSafe(|| {
+        let _ = LazyLock::force_mut(&mut lazy);
+    }));
+
+    // Second access should panic with the poisoned message.
+    let _ = &*lazy;
+}
+
+/// Verifies that when the initialization closure panics with a custom message, that message is
+/// preserved and not overridden by `LazyLock`.
+#[test]
+#[cfg_attr(not(panic = "unwind"), ignore = "test requires unwinding support")]
+#[should_panic(expected = "custom panic message from closure")]
+fn lazy_lock_preserves_closure_panic_message() {
+    let lazy: LazyLock<String> = LazyLock::new(|| panic!("custom panic message from closure"));
+
+    // This should panic with the original message from the closure.
+    let _ = &*lazy;
+}

src/bootstrap/src/bin/main.rs

@@ -198,14 +198,16 @@ fn check_version(config: &Config) -> Option<String> {
         Some(ChangeId::Id(id)) if id == latest_change_id => return None,
         Some(ChangeId::Ignore) => return None,
         Some(ChangeId::Id(id)) => id,
-        None => {
+        // The warning is only useful for development, not release tarballs
+        None if !config.rust_info.is_from_tarball() => {
             msg.push_str("WARNING: The `change-id` is missing in the `bootstrap.toml`. This means that you will not be able to track the major changes made to the bootstrap configurations.\n");
             msg.push_str("NOTE: to silence this warning, ");
             msg.push_str(&format!(
                 "add `change-id = {latest_change_id}` or `change-id = \"ignore\"` at the top of `bootstrap.toml`"
             ));
             return Some(msg);
         }
+        None => return None,
     };
 
     // Always try to use `change-id` from .last-warned-change-id first. If it doesn't exist,

src/bootstrap/src/core/config/config.rs

@@ -327,59 +327,6 @@ pub struct Config {
 }
 
 impl Config {
-    #[cfg_attr(
-        feature = "tracing",
-        instrument(target = "CONFIG_HANDLING", level = "trace", name = "Config::default_opts")
-    )]
-    pub fn default_opts() -> Config {
-        #[cfg(feature = "tracing")]
-        span!(target: "CONFIG_HANDLING", tracing::Level::TRACE, "constructing default config");
-
-        Config {
-            bypass_bootstrap_lock: false,
-            llvm_optimize: true,
-            ninja_in_file: true,
-            llvm_static_stdcpp: false,
-            llvm_libzstd: false,
-            backtrace: true,
-            rust_optimize: RustOptimize::Bool(true),
-            rust_optimize_tests: true,
-            rust_randomize_layout: false,
-            submodules: None,
-            docs: true,
-            docs_minification: true,
-            rust_rpath: true,
-            rust_strip: false,
-            channel: "dev".to_string(),
-            codegen_tests: true,
-            rust_dist_src: true,
-            rust_codegen_backends: vec![CodegenBackendKind::Llvm],
-            deny_warnings: true,
-            bindir: "bin".into(),
-            dist_include_mingw_linker: true,
-            dist_compression_profile: "fast".into(),
-
-            stdout_is_tty: std::io::stdout().is_terminal(),
-            stderr_is_tty: std::io::stderr().is_terminal(),
-
-            // set by build.rs
-            host_target: get_host_target(),
-
-            src: {
-                let manifest_dir = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
-                // Undo `src/bootstrap`
-                manifest_dir.parent().unwrap().parent().unwrap().to_owned()
-            },
-            out: PathBuf::from("build"),
-
-            // This is needed by codegen_ssa on macOS to ship `llvm-objcopy` aliased to
-            // `rust-objcopy` to workaround bad `strip`s on macOS.
-            llvm_tools_enabled: true,
-
-            ..Default::default()
-        }
-    }
-
     pub fn set_dry_run(&mut self, dry_run: DryRun) {
         self.exec_ctx.set_dry_run(dry_run);
     }

src/tools/clippy/Cargo.toml

@@ -65,13 +65,6 @@ harness = false
 name = "dogfood"
 harness = false
 
-# quine-mc_cluskey makes up a significant part of the runtime in dogfood
-# due to the number of conditions in the clippy_lints crate
-# and enabling optimizations for that specific dependency helps a bit
-# without increasing total build times.
-[profile.dev.package.quine-mc_cluskey]
-opt-level = 3
-
 [lints.rust.unexpected_cfgs]
 level = "warn"
 check-cfg = ['cfg(bootstrap)']

src/tools/miri/src/shims/native_lib/mod.rs

@@ -242,14 +242,9 @@ trait EvalContextExtPriv<'tcx>: crate::MiriInterpCxExt<'tcx> {
 
                 match evt {
                     AccessEvent::Read(_) => {
-                        // FIXME: ProvenanceMap should have something like get_range().
-                        let p_map = alloc.provenance();
-                        for idx in overlap {
-                            // If a provenance was read by the foreign code, expose it.
-                            if let Some((prov, _idx)) = p_map.get_byte(Size::from_bytes(idx), this)
-                            {
-                                this.expose_provenance(prov)?;
-                            }
+                        // If a provenance was read by the foreign code, expose it.
+                        for prov in alloc.provenance().get_range(this, overlap.into()) {
+                            this.expose_provenance(prov)?;
                         }
                     }
                     AccessEvent::Write(_, certain) => {