Rollup merge of rust-lang#154861 - mehdiakiki:fix/rlib-digest, r=petrochenkov

Add rlib digest to identify Rust object files

This adds a metadata entry to `rlib` archives that lists which members are Rust object files instead of relying on the filename heuristic in `looks_like_rust_object.file`. I also added a fallback to the old behavior for `rlibs` built by older compilers.

Part of rust-lang#138243.

Author: jhpratt

compiler/rustc_codegen_gcc/src/back/lto.rs

@@ -24,9 +24,10 @@ use std::path::{Path, PathBuf};
 use gccjit::OutputKind;
 use object::read::archive::ArchiveFile;
 use rustc_codegen_ssa::back::lto::SerializedModule;
+use rustc_codegen_ssa::back::rmeta_link;
 use rustc_codegen_ssa::back::write::{CodegenContext, FatLtoInput, SharedEmitter};
 use rustc_codegen_ssa::traits::*;
-use rustc_codegen_ssa::{CompiledModule, ModuleCodegen, ModuleKind, looks_like_rust_object_file};
+use rustc_codegen_ssa::{CompiledModule, ModuleCodegen, ModuleKind};
 use rustc_data_structures::memmap::Mmap;
 use rustc_data_structures::profiling::SelfProfilerRef;
 use rustc_errors::{DiagCtxt, DiagCtxtHandle};
@@ -63,6 +64,7 @@ fn prepare_lto(each_linked_rlib_for_lto: &[PathBuf], dcx: DiagCtxtHandle<'_>) ->
         let archive_data = unsafe {
             Mmap::map(File::open(path).expect("couldn't open rlib")).expect("couldn't map rlib")
         };
+        let metadata_link = rmeta_link::read_from_data(&archive_data, path).unwrap();
         let archive = ArchiveFile::parse(&*archive_data).expect("wanted an rlib");
         let obj_files = archive
             .members()
@@ -71,7 +73,7 @@ fn prepare_lto(each_linked_rlib_for_lto: &[PathBuf], dcx: DiagCtxtHandle<'_>) ->
                     .ok()
                     .and_then(|c| std::str::from_utf8(c.name()).ok().map(|name| (name.trim(), c)))
             })
-            .filter(|&(name, _)| looks_like_rust_object_file(name));
+            .filter(|&(name, _)| metadata_link.rust_object_files.iter().any(|f| f == name));
         for (name, child) in obj_files {
             info!("adding bitcode from {}", name);
             let path = tmp_path.path().join(name);

compiler/rustc_codegen_llvm/src/back/lto.rs

@@ -8,11 +8,12 @@ use std::{io, iter, slice};
 use object::read::archive::ArchiveFile;
 use object::{Object, ObjectSection};
 use rustc_codegen_ssa::back::lto::{SerializedModule, ThinModule, ThinShared};
+use rustc_codegen_ssa::back::rmeta_link;
 use rustc_codegen_ssa::back::write::{
     CodegenContext, FatLtoInput, SharedEmitter, TargetMachineFactoryFn, ThinLtoInput,
 };
 use rustc_codegen_ssa::traits::*;
-use rustc_codegen_ssa::{CompiledModule, ModuleCodegen, ModuleKind, looks_like_rust_object_file};
+use rustc_codegen_ssa::{CompiledModule, ModuleCodegen, ModuleKind};
 use rustc_data_structures::fx::FxHashMap;
 use rustc_data_structures::memmap::Mmap;
 use rustc_data_structures::profiling::SelfProfilerRef;
@@ -96,14 +97,15 @@ fn prepare_lto(
                 .expect("couldn't map rlib")
         };
         let archive = ArchiveFile::parse(&*archive_data).expect("wanted an rlib");
+        let metadata_link = rmeta_link::read(&archive, &archive_data, &path).unwrap();
         let obj_files = archive
             .members()
             .filter_map(|child| {
                 child
                     .ok()
                     .and_then(|c| std::str::from_utf8(c.name()).ok().map(|name| (name.trim(), c)))
             })
-            .filter(|&(name, _)| looks_like_rust_object_file(name));
+            .filter(|&(name, _)| metadata_link.rust_object_files.iter().any(|f| f == name));
         for (name, child) in obj_files {
             info!("adding bitcode from {}", name);
             match get_bitcode_slice_from_object_data(

compiler/rustc_codegen_ssa/src/back/archive.rs

@@ -21,6 +21,7 @@ use rustc_target::spec::Arch;
 use tracing::trace;
 
 use super::metadata::{create_compressed_metadata_file, search_for_section};
+use super::rmeta_link::{self, RmetaLink};
 use crate::common;
 // Public for ArchiveBuilderBuilder::extract_bundled_libs
 pub use crate::errors::ExtractBundledLibsError;
@@ -313,7 +314,7 @@ pub trait ArchiveBuilder {
     fn add_archive(
         &mut self,
         archive: &Path,
-        skip: Box<dyn FnMut(&str) -> bool + 'static>,
+        skip: Option<Box<dyn FnMut(&str, Option<&RmetaLink>) -> bool + 'static>>,
     ) -> io::Result<()>;
 
     fn build(self: Box<Self>, output: &Path) -> bool;
@@ -445,7 +446,7 @@ impl<'a> ArchiveBuilder for ArArchiveBuilder<'a> {
     fn add_archive(
         &mut self,
         archive_path: &Path,
-        mut skip: Box<dyn FnMut(&str) -> bool + 'static>,
+        mut skip: Option<Box<dyn FnMut(&str, Option<&RmetaLink>) -> bool + 'static>>,
     ) -> io::Result<()> {
         let mut archive_path = archive_path.to_path_buf();
         if self.sess.target.llvm_target.contains("-apple-macosx")
@@ -461,6 +462,8 @@ impl<'a> ArchiveBuilder for ArArchiveBuilder<'a> {
         let archive_map = unsafe { Mmap::map(File::open(&archive_path)?)? };
         let archive = ArchiveFile::parse(&*archive_map)
             .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?;
+        let metadata_link =
+            skip.as_ref().and_then(|_| rmeta_link::read(&archive, &archive_map, &archive_path));
         let archive_index = self.src_archives.len();
 
         if let Some(expected_kind) =
@@ -480,7 +483,8 @@ impl<'a> ArchiveBuilder for ArArchiveBuilder<'a> {
             let entry = entry.map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?;
             let file_name = String::from_utf8(entry.name().to_vec())
                 .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?;
-            if !skip(&file_name) {
+            let drop = skip.as_mut().is_some_and(|f| f(&file_name, metadata_link.as_ref()));
+            if !drop {
                 if entry.is_thin() {
                     let member_path = archive_path.parent().unwrap().join(Path::new(&file_name));
                     self.entries.push((file_name.into_bytes(), ArchiveEntry::File(member_path)));

compiler/rustc_codegen_ssa/src/back/link.rs

@@ -58,12 +58,9 @@ use super::command::Command;
 use super::linker::{self, Linker};
 use super::metadata::{MetadataPosition, create_wrapper_file};
 use super::rpath::{self, RPathConfig};
-use super::{apple, versioned_llvm_target};
+use super::{apple, rmeta_link, versioned_llvm_target};
 use crate::base::needs_allocator_shim_for_linking;
-use crate::{
-    CodegenLintLevels, CompiledModule, CompiledModules, CrateInfo, NativeLib, errors,
-    looks_like_rust_object_file,
-};
+use crate::{CodegenLintLevels, CompiledModule, CompiledModules, CrateInfo, NativeLib, errors};
 
 pub fn ensure_removed(dcx: DiagCtxtHandle<'_>, path: &Path) {
     if let Err(e) = fs::remove_file(path) {
@@ -307,6 +304,27 @@ fn link_rlib<'a>(
 ) -> Box<dyn ArchiveBuilder + 'a> {
     let mut ab = archive_builder_builder.new_archive_builder(sess);
 
+    // Pre-compute the list of Rust object filenames and materialize the rmeta-link
+    // wrapper file before any `add_file` calls. This lets the rmeta-link member be
+    // placed immediately after metadata in the archive, so consumers can find
+    // it without iterating every archive member.
+    let rust_object_files: Vec<String> = compiled_modules
+        .modules
+        .iter()
+        .filter_map(|m| m.object.as_ref())
+        .map(|obj| obj.file_name().unwrap().to_str().unwrap().to_string())
+        .collect();
+
+    let metadata_link_file = if matches!(flavor, RlibFlavor::Normal) {
+        let metadata_link = rmeta_link::RmetaLink { rust_object_files };
+        let metadata_link_data = metadata_link.encode();
+        let (wrapper, _) =
+            create_wrapper_file(sess, rmeta_link::SECTION.to_string(), &metadata_link_data);
+        Some(emit_wrapper_file(sess, &wrapper, tmpdir.as_ref(), rmeta_link::FILENAME))
+    } else {
+        None
+    };
+
     let trailing_metadata = match flavor {
         RlibFlavor::Normal => {
             let (metadata, metadata_position) =
@@ -320,6 +338,11 @@ fn link_rlib<'a>(
                     // If it is possible however, placing the metadata object first improves
                     // performance of getting metadata from rlibs.
                     ab.add_file(&metadata);
+                    // Place the rmeta-link member immediately after metadata so consumers
+                    // can find it without iterating the whole archive.
+                    if let Some(file) = &metadata_link_file {
+                        ab.add_file(file);
+                    }
                     None
                 }
                 MetadataPosition::Last => Some(metadata),
@@ -383,7 +406,7 @@ fn link_rlib<'a>(
             packed_bundled_libs.push(wrapper_file);
         } else {
             let path = find_native_static_library(lib.name.as_str(), lib.verbatim, sess);
-            ab.add_archive(&path, Box::new(|_| false)).unwrap_or_else(|error| {
+            ab.add_archive(&path, None).unwrap_or_else(|error| {
                 sess.dcx().emit_fatal(errors::AddNativeLibrary { library_path: path, error })
             });
         }
@@ -400,7 +423,7 @@ fn link_rlib<'a>(
             tmpdir.as_ref(),
             true,
         ) {
-            ab.add_archive(&output_path, Box::new(|_| false)).unwrap_or_else(|error| {
+            ab.add_archive(&output_path, None).unwrap_or_else(|error| {
                 sess.dcx()
                     .emit_fatal(errors::AddNativeLibrary { library_path: output_path, error });
             });
@@ -434,6 +457,11 @@ fn link_rlib<'a>(
         // Basically, all this means is that this code should not move above the
         // code above.
         ab.add_file(&trailing_metadata);
+        // Place the rmeta-link member immediately after metadata so consumers can
+        // find it without iterating the whole archive.
+        if let Some(file) = &metadata_link_file {
+            ab.add_file(file);
+        }
     }
 
     // Add all bundled static native library dependencies.
@@ -488,14 +516,16 @@ fn link_staticlib(
         let bundled_libs: FxIndexSet<_> = native_libs.filter_map(|lib| lib.filename).collect();
         ab.add_archive(
             path,
-            Box::new(move |fname: &str| {
-                // Ignore metadata files, no matter the name.
-                if fname == METADATA_FILENAME {
+            Some(Box::new(move |fname: &str, metadata_link| {
+                // Ignore metadata and rmeta-link files.
+                if fname == METADATA_FILENAME || fname == rmeta_link::FILENAME {
                     return true;
                 }
 
-                // Don't include Rust objects if LTO is enabled
-                if lto && looks_like_rust_object_file(fname) {
+                // Don't include Rust objects if LTO is enabled.
+                if lto
+                    && metadata_link.is_some_and(|m| m.rust_object_files.iter().any(|f| f == fname))
+                {
                     return true;
                 }
 
@@ -505,7 +535,7 @@ fn link_staticlib(
                 }
 
                 false
-            }),
+            })),
         )
         .unwrap();
 
@@ -516,7 +546,7 @@ fn link_staticlib(
         for filename in relevant_libs.iter() {
             let joined = tempdir.as_ref().join(filename.as_str());
             let path = joined.as_path();
-            ab.add_archive(path, Box::new(|_| false)).unwrap();
+            ab.add_archive(path, None).unwrap();
         }
 
         all_native_libs.extend(crate_info.native_libraries[&cnum].iter().cloned());
@@ -3167,23 +3197,20 @@ fn add_static_crate(
     let bundled_lib_file_names = bundled_lib_file_names.clone();
 
     sess.prof.generic_activity_with_arg("link_altering_rlib", name).run(|| {
-        let canonical_name = name.replace('-', "_");
         let upstream_rust_objects_already_included =
             are_upstream_rust_objects_already_included(sess);
         let is_builtins = sess.target.no_builtins || !crate_info.is_no_builtins.contains(&cnum);
 
         let mut archive = archive_builder_builder.new_archive_builder(sess);
         if let Err(error) = archive.add_archive(
             cratepath,
-            Box::new(move |f| {
-                if f == METADATA_FILENAME {
+            Some(Box::new(move |f, metadata_link| {
+                if f == METADATA_FILENAME || f == rmeta_link::FILENAME {
                     return true;
                 }
 
-                let canonical = f.replace('-', "_");
-
                 let is_rust_object =
-                    canonical.starts_with(&canonical_name) && looks_like_rust_object_file(f);
+                    metadata_link.is_some_and(|m| m.rust_object_files.iter().any(|rf| rf == f));
 
                 // If we're performing LTO and this is a rust-generated object
                 // file, then we don't need the object file as it's part of the
@@ -3203,7 +3230,7 @@ fn add_static_crate(
                 }
 
                 false
-            }),
+            })),
         ) {
             sess.dcx()
                 .emit_fatal(errors::RlibArchiveBuildFailure { path: cratepath.clone(), error });

compiler/rustc_codegen_ssa/src/back/mod.rs

@@ -9,6 +9,7 @@ pub mod link;
 pub(crate) mod linker;
 pub mod lto;
 pub mod metadata;
+pub mod rmeta_link;
 pub(crate) mod rpath;
 pub mod symbol_export;
 pub mod write;

compiler/rustc_codegen_ssa/src/back/rmeta_link.rs

@@ -0,0 +1,56 @@
+//! Late-metadata archive member that lists which rlib entries are Rust object files,
+//! and potentially other data collected and used when building or linking a rlib.
+//! See <https://github.com/rust-lang/rust/issues/138243>.
+
+use std::path::Path;
+
+use object::read::archive::ArchiveFile;
+use rustc_serialize::opaque::mem_encoder::MemEncoder;
+use rustc_serialize::opaque::{MAGIC_END_BYTES, MemDecoder};
+use rustc_serialize::{Decodable, Encodable};
+
+use super::metadata::search_for_section;
+
+pub(crate) const FILENAME: &str = "lib.rmeta-link";
+pub(crate) const SECTION: &str = ".rmeta-link";
+
+pub struct RmetaLink {
+    pub rust_object_files: Vec<String>,
+}
+
+impl RmetaLink {
+    pub(crate) fn encode(&self) -> Vec<u8> {
+        let mut encoder = MemEncoder::new();
+        self.rust_object_files.encode(&mut encoder);
+        let mut data = encoder.finish();
+        data.extend_from_slice(MAGIC_END_BYTES);
+        data
+    }
+
+    pub(crate) fn decode(data: &[u8]) -> Option<RmetaLink> {
+        let mut decoder = MemDecoder::new(data, 0).ok()?;
+        let rust_object_files = Vec::<String>::decode(&mut decoder);
+        Some(RmetaLink { rust_object_files })
+    }
+}
+
+/// Reads the link-time metadata from an already-parsed archive.
+pub fn read(archive: &ArchiveFile<'_>, archive_data: &[u8], rlib_path: &Path) -> Option<RmetaLink> {
+    for entry in archive.members() {
+        let entry = entry.ok()?;
+        if entry.name() == FILENAME.as_bytes() {
+            let data = entry.data(archive_data).ok()?;
+            let section_data = search_for_section(rlib_path, data, SECTION).ok()?;
+            return RmetaLink::decode(section_data);
+        }
+    }
+    None
+}
+
+/// Like [`read`], but parses the archive from raw bytes.
+///
+/// Use this when the caller's `ArchiveFile` comes from a different version of the `object` crate.
+pub fn read_from_data(archive_data: &[u8], rlib_path: &Path) -> Option<RmetaLink> {
+    let archive = ArchiveFile::parse(archive_data).ok()?;
+    read(&archive, archive_data, rlib_path)
+}

compiler/rustc_codegen_ssa/src/lib.rs

@@ -35,7 +35,7 @@ use rustc_middle::util::Providers;
 use rustc_serialize::opaque::{FileEncoder, MemDecoder};
 use rustc_serialize::{Decodable, Decoder, Encodable, Encoder};
 use rustc_session::Session;
-use rustc_session::config::{CrateType, OutputFilenames, OutputType, RUST_CGU_EXT};
+use rustc_session::config::{CrateType, OutputFilenames, OutputType};
 use rustc_session::cstore::{self, CrateSource};
 use rustc_session::lint::builtin::LINKER_MESSAGES;
 use rustc_span::Symbol;
@@ -272,23 +272,6 @@ pub fn provide(providers: &mut Providers) {
     providers.queries.global_backend_features = |_tcx: TyCtxt<'_>, ()| vec![];
 }
 
-/// Checks if the given filename ends with the `.rcgu.o` extension that `rustc`
-/// uses for the object files it generates.
-pub fn looks_like_rust_object_file(filename: &str) -> bool {
-    let path = Path::new(filename);
-    let ext = path.extension().and_then(|s| s.to_str());
-    if ext != Some(OutputType::Object.extension()) {
-        // The file name does not end with ".o", so it can't be an object file.
-        return false;
-    }
-
-    // Strip the ".o" at the end
-    let ext2 = path.file_stem().and_then(|s| Path::new(s).extension()).and_then(|s| s.to_str());
-
-    // Check if the "inner" extension
-    ext2 == Some(RUST_CGU_EXT)
-}
-
 const RLINK_VERSION: u32 = 1;
 const RLINK_MAGIC: &[u8] = b"rustlink";