Fixes LTO + build-std + Oz failed to resolve undefined symbols for the compiler_builtins.

Author: dianqk

compiler/rustc_codegen_llvm/src/back/lto.rs

@@ -53,8 +53,17 @@ fn prepare_lto(
         Lto::No => panic!("didn't request LTO but we're doing LTO"),
     };
 
+    let export_for_undefined_symbols =
+        &|name: &str| match &cgcx.undefined_symbols_from_ignored_for_lto {
+            Some(undefined_symbols) => undefined_symbols.contains(name),
+            None => false,
+        };
+
     let symbol_filter = &|&(ref name, info): &(String, SymbolExportInfo)| {
-        if info.level.is_below_threshold(export_threshold) || info.used {
+        if info.level.is_below_threshold(export_threshold)
+            || info.used
+            || export_for_undefined_symbols(name)
+        {
             Some(CString::new(name.as_str()).unwrap())
         } else {
             None

compiler/rustc_codegen_ssa/src/back/archive.rs

@@ -1,3 +1,4 @@
+use object::{Object, ObjectSymbol};
 use rustc_data_structures::fx::FxHashSet;
 use rustc_data_structures::memmap::Mmap;
 use rustc_session::cstore::DllImport;
@@ -308,3 +309,32 @@ impl<'a> ArArchiveBuilder<'a> {
 fn io_error_context(context: &str, err: io::Error) -> io::Error {
     io::Error::new(io::ErrorKind::Other, format!("{context}: {err}"))
 }
+
+pub fn read_archive_file_undefined_symbols(archive_path: &Path) -> io::Result<Vec<String>> {
+    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 mut undefined_symbols = Vec::new();
+    for entry in archive.members() {
+        let entry = entry.map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?;
+        if entry.name().ends_with(b".rmeta") {
+            continue;
+        }
+        let data = entry
+            .data(&*archive_map)
+            .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?;
+        let file = object::File::parse(&*data)
+            .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?;
+        for symbol in file.symbols() {
+            if symbol.is_undefined() {
+                undefined_symbols.push(
+                    symbol
+                        .name()
+                        .map_err(|err| io::Error::new(io::ErrorKind::InvalidData, err))?
+                        .to_string(),
+                );
+            }
+        }
+    }
+    Ok(undefined_symbols)
+}

compiler/rustc_codegen_ssa/src/back/write.rs

@@ -2,14 +2,15 @@ use super::link::{self, ensure_removed};
 use super::lto::{self, SerializedModule};
 use super::symbol_export::symbol_name_for_instance_in_crate;
 
+use crate::back::archive::read_archive_file_undefined_symbols;
 use crate::errors;
 use crate::traits::*;
 use crate::{
     CachedModuleCodegen, CodegenResults, CompiledModule, CrateInfo, ModuleCodegen, ModuleKind,
 };
 use jobserver::{Acquired, Client};
 use rustc_ast::attr;
-use rustc_data_structures::fx::{FxHashMap, FxIndexMap};
+use rustc_data_structures::fx::{FxHashMap, FxHashSet, FxIndexMap};
 use rustc_data_structures::memmap::Mmap;
 use rustc_data_structures::profiling::{SelfProfilerRef, VerboseTimingGuard};
 use rustc_data_structures::sync::Lrc;
@@ -327,6 +328,7 @@ pub struct CodegenContext<B: WriteBackendMethods> {
     pub fewer_names: bool,
     pub time_trace: bool,
     pub exported_symbols: Option<Arc<ExportedSymbols>>,
+    pub undefined_symbols_from_ignored_for_lto: Option<FxHashSet<String>>,
     pub opts: Arc<config::Options>,
     pub crate_types: Vec<CrateType>,
     pub each_linked_rlib_for_lto: Vec<(CrateNum, PathBuf)>,
@@ -989,13 +991,36 @@ fn start_executing_work<B: ExtraBackendMethods>(
     let sess = tcx.sess;
 
     let mut each_linked_rlib_for_lto = Vec::new();
+    let mut ignored_cnum_for_lto = Vec::new();
     drop(link::each_linked_rlib(crate_info, None, &mut |cnum, path| {
         if link::ignored_for_lto(sess, crate_info, cnum) {
+            ignored_cnum_for_lto.push(cnum);
             return;
         }
         each_linked_rlib_for_lto.push((cnum, path.to_path_buf()));
     }));
 
+    let undefined_symbols_from_ignored_for_lto = {
+        match sess.lto() {
+            Lto::Fat | Lto::Thin => {
+                let mut undefined_symbols_from_ignored_for_lto = FxHashSet::default();
+                if !sess.target.is_like_wasm {
+                    // FIXME: Add an undefined symbol lookup for wasm. https://github.com/WebAssembly/tool-conventions/blob/main/Linking.md
+                    for cnum in ignored_cnum_for_lto {
+                        let c_src = tcx.used_crate_source(cnum);
+                        if let Some((path, _)) = &c_src.rlib {
+                            let undefined_symbols = read_archive_file_undefined_symbols(path)
+                                .expect("failed to read undefined symbols");
+                            undefined_symbols_from_ignored_for_lto.extend(undefined_symbols);
+                        }
+                    }
+                }
+                Some(undefined_symbols_from_ignored_for_lto)
+            }
+            Lto::No | Lto::ThinLocal => None,
+        }
+    };
+
     // Compute the set of symbols we need to retain when doing LTO (if we need to)
     let exported_symbols = {
         let mut exported_symbols = FxHashMap::default();
@@ -1049,6 +1074,7 @@ fn start_executing_work<B: ExtraBackendMethods>(
         backend: backend.clone(),
         crate_types: sess.crate_types().to_vec(),
         each_linked_rlib_for_lto,
+        undefined_symbols_from_ignored_for_lto,
         lto: sess.lto(),
         fewer_names: sess.fewer_names(),
         save_temps: sess.opts.cg.save_temps,