error on invalid macho section specifier

Author: folkertdev

compiler/rustc_attr_parsing/src/attributes/link_attrs.rs

@@ -16,8 +16,9 @@ use crate::attributes::cfg::parse_cfg_entry;
 use crate::session_diagnostics::{
     AsNeededCompatibility, BundleNeedsStatic, EmptyLinkName, ExportSymbolsNeedsStatic,
     ImportNameTypeRaw, ImportNameTypeX86, IncompatibleWasmLink, InvalidLinkModifier,
-    LinkFrameworkApple, LinkOrdinalOutOfRange, LinkRequiresName, MultipleModifiers,
-    NullOnLinkSection, RawDylibNoNul, RawDylibOnlyWindows, WholeArchiveNeedsStatic,
+    InvalidMachoSection, InvalidMachoSectionReason, LinkFrameworkApple, LinkOrdinalOutOfRange,
+    LinkRequiresName, MultipleModifiers, NullOnLinkSection, RawDylibNoNul, RawDylibOnlyWindows,
+    WholeArchiveNeedsStatic,
 };
 
 pub(crate) struct LinkNameParser;
@@ -462,6 +463,29 @@ impl LinkParser {
 
 pub(crate) struct LinkSectionParser;
 
+fn check_link_section_macho(name: Symbol) -> Result<(), InvalidMachoSectionReason> {
+    let mut parts = name.as_str().split(',').map(|s| s.trim());
+
+    // The segment can be empty.
+    let _segment = parts.next();
+
+    // But the section is required.
+    let section = match parts.next() {
+        None | Some("") => return Err(InvalidMachoSectionReason::MissingSection),
+        Some(section) => section,
+    };
+
+    if section.len() > 16 {
+        return Err(InvalidMachoSectionReason::SectionTooLong { section: section.to_string() });
+    }
+
+    // LLVM also checks the other components of the section specifier, but that logic is hard to
+    // keep in sync. We skip it here for now, assuming that if you got that far you'll be able
+    // to interpret the LLVM errors.
+
+    Ok(())
+}
+
 impl<S: Stage> SingleAttributeParser<S> for LinkSectionParser {
     const PATH: &[Symbol] = &[sym::link_section];
     const ON_DUPLICATE: OnDuplicate<S> = OnDuplicate::WarnButFutureError;
@@ -495,6 +519,18 @@ impl<S: Stage> SingleAttributeParser<S> for LinkSectionParser {
             return None;
         }
 
+        // We (currently) only validate macho section specifiers.
+        match cx.sess.target.binary_format {
+            BinaryFormat::MachO => match check_link_section_macho(name) {
+                Ok(()) => {}
+                Err(reason) => {
+                    cx.emit_err(InvalidMachoSection { name_span: nv.value_span, reason });
+                    return None;
+                }
+            },
+            BinaryFormat::Coff | BinaryFormat::Elf | BinaryFormat::Wasm | BinaryFormat::Xcoff => {}
+        }
+
         Some(LinkSection { name, span: cx.attr_span })
     }
 }

compiler/rustc_attr_parsing/src/session_diagnostics.rs

@@ -1137,3 +1137,22 @@ pub(crate) struct UnstableAttrForAlreadyStableFeature {
     #[label("the stability attribute annotates this item")]
     pub item_span: Span,
 }
+
+#[derive(Diagnostic)]
+#[diag("invalid Mach-O section specifier")]
+pub(crate) struct InvalidMachoSection {
+    #[primary_span]
+    #[label("not a valid Mach-O section specifier")]
+    pub name_span: Span,
+    #[subdiagnostic]
+    pub reason: InvalidMachoSectionReason,
+}
+
+#[derive(Subdiagnostic)]
+pub(crate) enum InvalidMachoSectionReason {
+    #[note("a Mach-O section specifier requires a segment and a section, separated by a comma")]
+    #[help("an example of a valid Mach-O section specifier is `__TEXT,__cstring`")]
+    MissingSection,
+    #[note("section name `{$section}` is longer than 16 bytes")]
+    SectionTooLong { section: String },
+}

compiler/rustc_hir/src/attrs/data_structures.rs

@@ -1098,8 +1098,8 @@ pub enum AttributeKind {
 
     /// Represents [`#[link_section]`](https://doc.rust-lang.org/reference/abi.html#the-link_section-attribute)
     LinkSection {
-        name: Symbol,
         span: Span,
+        name: Symbol,
     },
 
     /// Represents `#[linkage]`.

tests/codegen-llvm/link_section.rs

@@ -3,14 +3,14 @@
 
 #![crate_type = "lib"]
 
-// CHECK: @VAR1 = {{(dso_local )?}}constant [4 x i8] c"\01\00\00\00", section ".test_one"
+// CHECK: @VAR1 = {{(dso_local )?}}constant [4 x i8] c"\01\00\00\00", section "__TEST,one"
 #[no_mangle]
-#[link_section = ".test_one"]
+#[link_section = "__TEST,one"]
 #[cfg(target_endian = "little")]
 pub static VAR1: u32 = 1;
 
 #[no_mangle]
-#[link_section = ".test_one"]
+#[link_section = "__TEST,one"]
 #[cfg(target_endian = "big")]
 pub static VAR1: u32 = 0x01000000;
 
@@ -19,17 +19,17 @@ pub enum E {
     B(f32),
 }
 
-// CHECK: @VAR2 = {{(dso_local )?}}constant {{.*}}, section ".test_two"
+// CHECK: @VAR2 = {{(dso_local )?}}constant {{.*}}, section "__TEST,two"
 #[no_mangle]
-#[link_section = ".test_two"]
+#[link_section = "__TEST,two"]
 pub static VAR2: E = E::A(666);
 
-// CHECK: @VAR3 = {{(dso_local )?}}constant {{.*}}, section ".test_three"
+// CHECK: @VAR3 = {{(dso_local )?}}constant {{.*}}, section "__TEST,three"
 #[no_mangle]
-#[link_section = ".test_three"]
+#[link_section = "__TEST,three"]
 pub static VAR3: E = E::B(1.);
 
-// CHECK: define {{(dso_local )?}}void @fn1() {{.*}} section ".test_four" {
+// CHECK: define {{(dso_local )?}}void @fn1() {{.*}} section "__TEST,four" {
 #[no_mangle]
-#[link_section = ".test_four"]
+#[link_section = "__TEST,four"]
 pub fn fn1() {}

tests/codegen-llvm/naked-fn/naked-functions.rs

@@ -22,7 +22,7 @@
 //@[thumb] needs-llvm-components: arm
 
 #![crate_type = "lib"]
-#![feature(no_core, lang_items, rustc_attrs)]
+#![feature(no_core, lang_items, rustc_attrs, cfg_target_object_format)]
 #![no_core]
 
 extern crate minicore;
@@ -170,14 +170,17 @@ pub extern "C" fn naked_with_args_and_return(a: isize, b: isize) -> isize {
 }
 
 // linux,linux_no_function_sections: .pushsection .text.some_different_name,\22ax\22, @progbits
-// macos:            .pushsection .text.some_different_name,regular,pure_instructions
+// macos: .pushsection __TEXT,different,regular,pure_instructions
 // win_x86_msvc,win_x86_gnu,win_i686_gnu: .section .text.some_different_name,\22xr\22
 // win_x86_gnu_function_sections: .section .text.some_different_name,\22xr\22
 // thumb:            .pushsection .text.some_different_name,\22ax\22, %progbits
 // CHECK-LABEL: test_link_section:
 #[no_mangle]
 #[unsafe(naked)]
-#[link_section = ".text.some_different_name"]
+#[link_section = cfg_select!(
+    target_object_format = "mach-o" =>  "__TEXT,different",
+    _ => ".text.some_different_name"
+)]
 pub extern "C" fn test_link_section() {
     cfg_select! {
         all(target_arch = "arm", target_feature = "thumb-mode") => {

tests/rustdoc-html/attributes-2021-edition.rs

@@ -9,6 +9,6 @@ pub extern "C" fn f() {}
 #[export_name = "bar"]
 pub extern "C" fn g() {}
 
-//@ has foo/fn.example.html '//pre[@class="rust item-decl"]' '#[unsafe(link_section = ".text")]'
-#[link_section = ".text"]
+//@ has foo/fn.example.html '//pre[@class="rust item-decl"]' '#[unsafe(link_section = "__TEXT,__text")]'
+#[link_section = "__TEXT,__text"]
 pub extern "C" fn example() {}

tests/rustdoc-html/attributes.rs

@@ -1,4 +1,5 @@
 //@ edition: 2024
+//@ only-linux
 #![crate_name = "foo"]
 
 //@ has foo/fn.f.html '//*[@class="code-attribute"]' '#[unsafe(no_mangle)]'

tests/rustdoc-html/inline_cross/attributes.rs

@@ -9,7 +9,7 @@
 pub use attributes::no_mangle;
 
 //@ has 'user/fn.link_section.html' '//pre[@class="rust item-decl"]' \
-//                                  '#[unsafe(link_section = ".here")]'
+//                                  '#[unsafe(link_section = "__TEXT,__here")]'
 pub use attributes::link_section;
 
 //@ has 'user/fn.export_name.html' '//pre[@class="rust item-decl"]' \

tests/rustdoc-html/inline_cross/auxiliary/attributes.rs

@@ -1,7 +1,7 @@
 #[unsafe(no_mangle)]
 pub fn no_mangle() {}
 
-#[unsafe(link_section = ".here")]
+#[unsafe(link_section = "__TEXT,__here")]
 pub fn link_section() {}
 
 #[unsafe(export_name = "exonym")]

tests/rustdoc-json/attrs/link_section_2021.rs

@@ -2,6 +2,6 @@
 #![no_std]
 
 //@ count "$.index[?(@.name=='example')].attrs[*]" 1
-//@ is "$.index[?(@.name=='example')].attrs[*].link_section" '".text"'
-#[link_section = ".text"]
+//@ is "$.index[?(@.name=='example')].attrs[*].link_section" '"__TEXT,__text"'
+#[link_section = "__TEXT,__text"]
 pub extern "C" fn example() {}