Auto merge of rust-lang#111001 - matthiaskrgr:rollup-u590scu, r=matthiaskrgr

Rollup of 7 pull requests

Successful merges:

 - rust-lang#110586 (Fix Unreadable non-UTF-8 output on localized MSVC)
 - rust-lang#110652 (Add test for warning-free builds of `core` under `no_global_oom_handling`)
 - rust-lang#110973 (improve error notes for packed struct reference diagnostic)
 - rust-lang#110981 (Move most rustdoc-ui tests into subdirectories)
 - rust-lang#110983 (rustdoc: Get `repr` information through `AdtDef` for foreign items)
 - rust-lang#110984 (Do not resolve anonymous lifetimes in consts to be static.)
 - rust-lang#110997 (Improve internal field comments on `slice::Iter(Mut)`)

Failed merges:

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

Author: bors

Cargo.lock

@@ -3250,6 +3250,7 @@ dependencies = [
  "tempfile",
  "thorin-dwp",
  "tracing",
+ "windows 0.46.0",
 ]
 
 [[package]]

compiler/rustc_codegen_ssa/Cargo.toml

@@ -49,3 +49,7 @@ libc = "0.2.50"
 version = "0.30.1"
 default-features = false
 features = ["read_core", "elf", "macho", "pe", "unaligned", "archive", "write"]
+
+[target.'cfg(windows)'.dependencies.windows]
+version = "0.46.0"
+features = ["Win32_Globalization"]

compiler/rustc_codegen_ssa/src/back/link.rs

@@ -860,7 +860,7 @@ fn link_natively<'a>(
             if !prog.status.success() {
                 let mut output = prog.stderr.clone();
                 output.extend_from_slice(&prog.stdout);
-                let escaped_output = escape_string(&output);
+                let escaped_output = escape_linker_output(&output, flavor);
                 // FIXME: Add UI tests for this error.
                 let err = errors::LinkingFailed {
                     linker_path: &linker_path,
@@ -1052,6 +1052,83 @@ fn escape_string(s: &[u8]) -> String {
     }
 }
 
+#[cfg(not(windows))]
+fn escape_linker_output(s: &[u8], _flavour: LinkerFlavor) -> String {
+    escape_string(s)
+}
+
+/// If the output of the msvc linker is not UTF-8 and the host is Windows,
+/// then try to convert the string from the OEM encoding.
+#[cfg(windows)]
+fn escape_linker_output(s: &[u8], flavour: LinkerFlavor) -> String {
+    // This only applies to the actual MSVC linker.
+    if flavour != LinkerFlavor::Msvc(Lld::No) {
+        return escape_string(s);
+    }
+    match str::from_utf8(s) {
+        Ok(s) => return s.to_owned(),
+        Err(_) => match win::locale_byte_str_to_string(s, win::oem_code_page()) {
+            Some(s) => s,
+            // The string is not UTF-8 and isn't valid for the OEM code page
+            None => format!("Non-UTF-8 output: {}", s.escape_ascii()),
+        },
+    }
+}
+
+/// Wrappers around the Windows API.
+#[cfg(windows)]
+mod win {
+    use windows::Win32::Globalization::{
+        GetLocaleInfoEx, MultiByteToWideChar, CP_OEMCP, LOCALE_IUSEUTF8LEGACYOEMCP,
+        LOCALE_NAME_SYSTEM_DEFAULT, LOCALE_RETURN_NUMBER, MB_ERR_INVALID_CHARS,
+    };
+
+    /// Get the Windows system OEM code page. This is most notably the code page
+    /// used for link.exe's output.
+    pub fn oem_code_page() -> u32 {
+        unsafe {
+            let mut cp: u32 = 0;
+            // We're using the `LOCALE_RETURN_NUMBER` flag to return a u32.
+            // But the API requires us to pass the data as though it's a [u16] string.
+            let len = std::mem::size_of::<u32>() / std::mem::size_of::<u16>();
+            let data = std::slice::from_raw_parts_mut(&mut cp as *mut u32 as *mut u16, len);
+            let len_written = GetLocaleInfoEx(
+                LOCALE_NAME_SYSTEM_DEFAULT,
+                LOCALE_IUSEUTF8LEGACYOEMCP | LOCALE_RETURN_NUMBER,
+                Some(data),
+            );
+            if len_written as usize == len { cp } else { CP_OEMCP }
+        }
+    }
+    /// Try to convert a multi-byte string to a UTF-8 string using the given code page
+    /// The string does not need to be null terminated.
+    ///
+    /// This is implemented as a wrapper around `MultiByteToWideChar`.
+    /// See <https://learn.microsoft.com/en-us/windows/win32/api/stringapiset/nf-stringapiset-multibytetowidechar>
+    ///
+    /// It will fail if the multi-byte string is longer than `i32::MAX` or if it contains
+    /// any invalid bytes for the expected encoding.
+    pub fn locale_byte_str_to_string(s: &[u8], code_page: u32) -> Option<String> {
+        // `MultiByteToWideChar` requires a length to be a "positive integer".
+        if s.len() > isize::MAX as usize {
+            return None;
+        }
+        // Error if the string is not valid for the expected code page.
+        let flags = MB_ERR_INVALID_CHARS;
+        // Call MultiByteToWideChar twice.
+        // First to calculate the length then to convert the string.
+        let mut len = unsafe { MultiByteToWideChar(code_page, flags, s, None) };
+        if len > 0 {
+            let mut utf16 = vec![0; len as usize];
+            len = unsafe { MultiByteToWideChar(code_page, flags, s, Some(&mut utf16)) };
+            if len > 0 {
+                return utf16.get(..len as usize).map(String::from_utf16_lossy);
+            }
+        }
+        None
+    }
+}
+
 fn add_sanitizer_libraries(sess: &Session, crate_type: CrateType, linker: &mut dyn Linker) {
     // On macOS the runtimes are distributed as dylibs which should be linked to
     // both executables and dynamic shared objects. Everywhere else the runtimes

compiler/rustc_feature/src/builtin_attrs.rs

@@ -344,7 +344,7 @@ pub const BUILTIN_ATTRIBUTES: &[BuiltinAttribute] = &[
     ),
     ungated!(link_name, Normal, template!(NameValueStr: "name"), FutureWarnPreceding),
     ungated!(no_link, Normal, template!(Word), WarnFollowing),
-    ungated!(repr, Normal, template!(List: "C"), DuplicatesOk),
+    ungated!(repr, Normal, template!(List: "C"), DuplicatesOk, @only_local: true),
     ungated!(export_name, Normal, template!(NameValueStr: "name"), FutureWarnPreceding),
     ungated!(link_section, Normal, template!(NameValueStr: "name"), FutureWarnPreceding),
     ungated!(no_mangle, Normal, template!(Word), WarnFollowing, @only_local: true),

compiler/rustc_mir_transform/src/check_packed_ref.rs

@@ -56,8 +56,11 @@ impl<'tcx> Visitor<'tcx> for PackedRefChecker<'_, 'tcx> {
                         "reference to packed field is unaligned"
                     )
                     .note(
-                        "fields of packed structs are not properly aligned, and creating \
-                        a misaligned reference is undefined behavior (even if that \
+                        "packed structs are only aligned by one byte, and many modern architectures \
+                        penalize unaligned field accesses"
+                    )
+                    .note(
+                        "creating a misaligned reference is undefined behavior (even if that \
                         reference is never dereferenced)",
                     ).help(
                         "copy the field contents to a local variable, or replace the \

compiler/rustc_resolve/src/late.rs

@@ -656,7 +656,7 @@ impl<'a: 'ast, 'ast, 'tcx> Visitor<'ast> for LateResolutionVisitor<'a, '_, 'ast,
     fn visit_anon_const(&mut self, constant: &'ast AnonConst) {
         // We deal with repeat expressions explicitly in `resolve_expr`.
         self.with_lifetime_rib(LifetimeRibKind::AnonConst, |this| {
-            this.with_lifetime_rib(LifetimeRibKind::Elided(LifetimeRes::Static), |this| {
+            this.with_lifetime_rib(LifetimeRibKind::Elided(LifetimeRes::Infer), |this| {
                 this.resolve_anon_const(constant, IsRepeatExpr::No);
             })
         })
@@ -4126,7 +4126,7 @@ impl<'a: 'ast, 'b, 'ast, 'tcx> LateResolutionVisitor<'a, 'b, 'ast, 'tcx> {
             ExprKind::Repeat(ref elem, ref ct) => {
                 self.visit_expr(elem);
                 self.with_lifetime_rib(LifetimeRibKind::AnonConst, |this| {
-                    this.with_lifetime_rib(LifetimeRibKind::Elided(LifetimeRes::Static), |this| {
+                    this.with_lifetime_rib(LifetimeRibKind::Elided(LifetimeRes::Infer), |this| {
                         this.resolve_anon_const(ct, IsRepeatExpr::Yes)
                     })
                 });

library/core/src/slice/iter.rs

@@ -60,10 +60,17 @@ impl<'a, T> IntoIterator for &'a mut [T] {
 #[stable(feature = "rust1", since = "1.0.0")]
 #[must_use = "iterators are lazy and do nothing unless consumed"]
 pub struct Iter<'a, T: 'a> {
+    /// The pointer to the next element to return, or the past-the-end location
+    /// if the iterator is empty.
+    ///
+    /// This address will be used for all ZST elements, never changed.
     ptr: NonNull<T>,
-    end: *const T, // If T is a ZST, this is actually ptr+len. This encoding is picked so that
-    // ptr == end is a quick test for the Iterator being empty, that works
-    // for both ZST and non-ZST.
+    /// For non-ZSTs, the non-null pointer to the past-the-end element.
+    ///
+    /// For ZSTs, this is `ptr.wrapping_byte_add(len)`.
+    ///
+    /// For all types, `ptr == end` tests whether the iterator is empty.
+    end: *const T,
     _marker: PhantomData<&'a T>,
 }
 
@@ -179,10 +186,17 @@ impl<T> AsRef<[T]> for Iter<'_, T> {
 #[stable(feature = "rust1", since = "1.0.0")]
 #[must_use = "iterators are lazy and do nothing unless consumed"]
 pub struct IterMut<'a, T: 'a> {
+    /// The pointer to the next element to return, or the past-the-end location
+    /// if the iterator is empty.
+    ///
+    /// This address will be used for all ZST elements, never changed.
     ptr: NonNull<T>,
-    end: *mut T, // If T is a ZST, this is actually ptr+len. This encoding is picked so that
-    // ptr == end is a quick test for the Iterator being empty, that works
-    // for both ZST and non-ZST.
+    /// For non-ZSTs, the non-null pointer to the past-the-end element.
+    ///
+    /// For ZSTs, this is `ptr.wrapping_byte_add(len)`.
+    ///
+    /// For all types, `ptr == end` tests whether the iterator is empty.
+    end: *mut T,
     _marker: PhantomData<&'a mut T>,
 }
 

src/librustdoc/clean/types.rs

@@ -11,6 +11,7 @@ use arrayvec::ArrayVec;
 use thin_vec::ThinVec;
 
 use rustc_ast as ast;
+use rustc_ast_pretty::pprust;
 use rustc_attr::{ConstStability, Deprecation, Stability, StabilityLevel};
 use rustc_const_eval::const_eval::is_unstable_const_fn;
 use rustc_data_structures::fx::{FxHashMap, FxHashSet};
@@ -711,6 +712,78 @@ impl Item {
         };
         Some(tcx.visibility(def_id))
     }
+
+    pub(crate) fn attributes(&self, tcx: TyCtxt<'_>, keep_as_is: bool) -> Vec<String> {
+        const ALLOWED_ATTRIBUTES: &[Symbol] =
+            &[sym::export_name, sym::link_section, sym::no_mangle, sym::repr, sym::non_exhaustive];
+
+        use rustc_abi::IntegerType;
+        use rustc_middle::ty::ReprFlags;
+
+        let mut attrs: Vec<String> = self
+            .attrs
+            .other_attrs
+            .iter()
+            .filter_map(|attr| {
+                if keep_as_is {
+                    Some(pprust::attribute_to_string(attr))
+                } else if ALLOWED_ATTRIBUTES.contains(&attr.name_or_empty()) {
+                    Some(
+                        pprust::attribute_to_string(attr)
+                            .replace("\\\n", "")
+                            .replace('\n', "")
+                            .replace("  ", " "),
+                    )
+                } else {
+                    None
+                }
+            })
+            .collect();
+        if let Some(def_id) = self.item_id.as_def_id() &&
+            !def_id.is_local() &&
+            // This check is needed because `adt_def` will panic if not a compatible type otherwise...
+            matches!(self.type_(), ItemType::Struct | ItemType::Enum | ItemType::Union)
+        {
+            let repr = tcx.adt_def(def_id).repr();
+            let mut out = Vec::new();
+            if repr.flags.contains(ReprFlags::IS_C) {
+                out.push("C");
+            }
+            if repr.flags.contains(ReprFlags::IS_TRANSPARENT) {
+                out.push("transparent");
+            }
+            if repr.flags.contains(ReprFlags::IS_SIMD) {
+                out.push("simd");
+            }
+            let pack_s;
+            if let Some(pack) = repr.pack {
+                pack_s = format!("packed({})", pack.bytes());
+                out.push(&pack_s);
+            }
+            let align_s;
+            if let Some(align) = repr.align {
+                align_s = format!("align({})", align.bytes());
+                out.push(&align_s);
+            }
+            let int_s;
+            if let Some(int) = repr.int {
+                int_s = match int {
+                    IntegerType::Pointer(is_signed) => {
+                        format!("{}size", if is_signed { 'i' } else { 'u' })
+                    }
+                    IntegerType::Fixed(size, is_signed) => {
+                        format!("{}{}", if is_signed { 'i' } else { 'u' }, size.size().bytes() * 8)
+                    }
+                };
+                out.push(&int_s);
+            }
+            if out.is_empty() {
+                return Vec::new();
+            }
+            attrs.push(format!("#[repr({})]", out.join(", ")));
+        }
+        attrs
+    }
 }
 
 #[derive(Clone, Debug)]

src/librustdoc/html/render/mod.rs

@@ -48,7 +48,6 @@ use std::str;
 use std::string::ToString;
 
 use askama::Template;
-use rustc_ast_pretty::pprust;
 use rustc_attr::{ConstStability, Deprecation, StabilityLevel};
 use rustc_data_structures::captures::Captures;
 use rustc_data_structures::fx::{FxHashMap, FxHashSet};
@@ -849,10 +848,10 @@ fn assoc_method(
     let (indent, indent_str, end_newline) = if parent == ItemType::Trait {
         header_len += 4;
         let indent_str = "    ";
-        write!(w, "{}", render_attributes_in_pre(meth, indent_str));
+        write!(w, "{}", render_attributes_in_pre(meth, indent_str, tcx));
         (4, indent_str, Ending::NoNewline)
     } else {
-        render_attributes_in_code(w, meth);
+        render_attributes_in_code(w, meth, tcx);
         (0, "", Ending::Newline)
     };
     w.reserve(header_len + "<a href=\"\" class=\"fn\">{".len() + "</a>".len());
@@ -1021,36 +1020,15 @@ fn render_assoc_item(
     }
 }
 
-const ALLOWED_ATTRIBUTES: &[Symbol] =
-    &[sym::export_name, sym::link_section, sym::no_mangle, sym::repr, sym::non_exhaustive];
-
-fn attributes(it: &clean::Item) -> Vec<String> {
-    it.attrs
-        .other_attrs
-        .iter()
-        .filter_map(|attr| {
-            if ALLOWED_ATTRIBUTES.contains(&attr.name_or_empty()) {
-                Some(
-                    pprust::attribute_to_string(attr)
-                        .replace("\\\n", "")
-                        .replace('\n', "")
-                        .replace("  ", " "),
-                )
-            } else {
-                None
-            }
-        })
-        .collect()
-}
-
 // When an attribute is rendered inside a `<pre>` tag, it is formatted using
 // a whitespace prefix and newline.
-fn render_attributes_in_pre<'a>(
+fn render_attributes_in_pre<'a, 'b: 'a>(
     it: &'a clean::Item,
     prefix: &'a str,
-) -> impl fmt::Display + Captures<'a> {
+    tcx: TyCtxt<'b>,
+) -> impl fmt::Display + Captures<'a> + Captures<'b> {
     crate::html::format::display_fn(move |f| {
-        for a in attributes(it) {
+        for a in it.attributes(tcx, false) {
             writeln!(f, "{}{}", prefix, a)?;
         }
         Ok(())
@@ -1059,8 +1037,8 @@ fn render_attributes_in_pre<'a>(
 
 // When an attribute is rendered inside a <code> tag, it is formatted using
 // a div to produce a newline after it.
-fn render_attributes_in_code(w: &mut Buffer, it: &clean::Item) {
-    for a in attributes(it) {
+fn render_attributes_in_code(w: &mut Buffer, it: &clean::Item, tcx: TyCtxt<'_>) {
+    for a in it.attributes(tcx, false) {
         write!(w, "<div class=\"code-attribute\">{}</div>", a);
     }
 }