[ELF] - Fix for "Bug 34238 - LTO is optimizing away symbols referenced from linker scripts"

[grimar]

Summary

Code suggested by Rui Ueyama in PR34238 comments.

Previously LTO optimized away symbols referenced from linker script
because did not see that them are used from regular objects.

Patch adds such symbols as undefined earlier, before running LTO,
what sets IsUsedInRegularObj for them and fixes the issue.

[ruiu]

ELF/Driver.cpp

1017–1018

You don't need to mention that because that is also applicable to all code that adds undefined symbols. If you add this comment, you need to add the same comment to line 1026, for example. Essentially, all undefined symbols must be added line 1025 as the comment for that line explains.

[grimar]

• Restored original comment.

ELF/Driver.cpp

1017–1018

Ok.

[ruiu]

LGTM

ELF/Driver.cpp

template <class ELFT> void LinkerDriver::link(opt::InputArgList &Args) {
   for (auto *Arg : Args.filtered(OPT_trace_symbol))
     Symtab->trace(Arg->getValue());
 
   // Add all files to the symbol table. This will add almost all
   // symbols that we need to the symbol table.
   for (InputFile *F : Files)
     Symtab->addFile<ELFT>(F);
 
+  // Some symbols (such as __ehdr_start) are defined lazily only when there
+  // are undefined symbols for them, so we add these to trigger that logic.
+  for (StringRef Sym : Script->Opt.ReferencedSymbols)
+    Symtab->addUndefined<ELFT>(Sym);
+
   // If an entry symbol is in a static archive, pull out that file now
   // to complete the symbol table. After this, no new names except a
   // few linker-synthesized ones will be added to the symbol table.
   if (Symtab->find(Config->Entry))
     Symtab->addUndefined<ELFT>(Config->Entry);
 
   // Return if there were name resolution errors.
   if (ErrorCount)
template <class ELFT> void LinkerDriver::link(opt::InputArgList &Args) {
   // Create alias symbols for -defsym option.
   for (std::pair<StringRef, StringRef> &Def : getDefsym(Args))
     Symtab->addSymbolAlias<ELFT>(Def.first, Def.second);
 
   Symtab->addCombinedLTOObject<ELFT>();
   if (ErrorCount)
     return;
 
-  // Some symbols (such as __ehdr_start) are defined lazily only when there
-  // are undefined symbols for them, so we add these to trigger that logic.
-  for (StringRef Sym : Script->Opt.ReferencedSymbols)
-    Symtab->addUndefined<ELFT>(Sym);
-
   // Apply symbol renames for -wrap and -defsym
   Symtab->applySymbolRenames();
 
   // Now that we have a complete list of input files.
   // Beyond this point, no new files are added.
   // Aggregate all input sections into one place.
   for (ObjFile<ELFT> *F : ObjFile<ELFT>::Instances)
     for (InputSectionBase *S : F->getSections())

test/ELF/lto/linker-script-symbols.ll

+; REQUIRES: x86
+; RUN: llvm-as %s -o %t.o
+; RUN: echo "foo = bar;" > %t.script
+
+; RUN: ld.lld -m elf_x86_64 %t.o -o %t2 --script %t.script -save-temps
+; RUN: llvm-readobj -symbols %t2.lto.o | FileCheck %s
+
+; CHECK-NOT:  zed
+; CHECK:      Symbol {
+; CHECK:        Name: bar
+; CHECK-NEXT:   Value:
+; CHECK-NEXT:   Size:
+; CHECK-NEXT:   Binding: Global
+; CHECK-NEXT:   Type: Function
+; CHECK-NEXT:   Other:
+; CHECK-NEXT:   Section:
+; CHECK-NEXT: }
+; CHECK-NOT:  zed
+
+target triple = "x86_64-unknown-linux-gnu"
+target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
+
+define void @bar() {
+  ret void
+}
+
+define void @zed() {
+  ret void
+}