Revert "[lld] Merge equivalent symbols found during ICF (#134342)"

The change would also merge *non-equivalent* symbols under some circumstances,
see comment with a reproducer on the PR.

> Fixes a correctness issue for AArch64 when ADRP and LDR instructions are
> outlined in separate sections and sections are fed to ICF for
> deduplication.
>
> See test case (based on
> #129122) for details. All
> rodata.* sections are folded into a single section with ICF. This leads
> to all f2_* function sections getting folded into one (as their
> relocation target symbols g* belong to .rodata.g* sections that have
> already been folded into one). Since relocations still refer original g*
> symbols, we end up creating duplicate GOT entry for all such symbols.
> This PR addresses that by tracking such folded symbols and create one
> GOT entry for all such symbols.
>
> Fixes #129122
>
> Co-authored by: @jyknight

This reverts commit 8389d6f.

Author: zmodem

lld/ELF/ICF.cpp

@@ -80,7 +80,6 @@
 #include "SymbolTable.h"
 #include "Symbols.h"
 #include "SyntheticSections.h"
-#include "llvm/ADT/EquivalenceClasses.h"
 #include "llvm/BinaryFormat/ELF.h"
 #include "llvm/Object/ELF.h"
 #include "llvm/Support/Parallel.h"
@@ -334,28 +333,6 @@ bool ICF<ELFT>::equalsConstant(const InputSection *a, const InputSection *b) {
              : constantEq(a, ra.relas, b, rb.relas);
 }
 
-template <class RelTy>
-static SmallVector<Symbol *> getReloc(const InputSection *sec,
-                                      Relocs<RelTy> relocs) {
-  SmallVector<Symbol *> syms;
-  for (auto ri = relocs.begin(), re = relocs.end(); ri != re; ++ri) {
-    Symbol &sym = sec->file->getRelocTargetSym(*ri);
-    syms.push_back(&sym);
-  }
-  return syms;
-}
-
-template <class ELFT>
-static SmallVector<Symbol *> getRelocTargetSyms(const InputSection *sec) {
-  const RelsOrRelas<ELFT> rel = sec->template relsOrRelas<ELFT>();
-  if (rel.areRelocsCrel())
-    return getReloc(sec, rel.crels);
-  if (rel.areRelocsRel())
-    return getReloc(sec, rel.rels);
-
-  return getReloc(sec, rel.relas);
-}
-
 // Compare two lists of relocations. Returns true if all pairs of
 // relocations point to the same section in terms of ICF.
 template <class ELFT>
@@ -560,28 +537,14 @@ template <class ELFT> void ICF<ELFT>::run() {
   auto print = [&ctx = ctx]() -> ELFSyncStream {
     return {ctx, ctx.arg.printIcfSections ? DiagLevel::Msg : DiagLevel::None};
   };
-
-  EquivalenceClasses<Symbol *> symbolEquivalence;
   // Merge sections by the equivalence class.
-  // Merge symbols identified as equivalent during ICF.
   forEachClassRange(0, sections.size(), [&](size_t begin, size_t end) {
     if (end - begin == 1)
       return;
     print() << "selected section " << sections[begin];
-    SmallVector<Symbol *> syms = getRelocTargetSyms<ELFT>(sections[begin]);
     for (size_t i = begin + 1; i < end; ++i) {
       print() << "  removing identical section " << sections[i];
       sections[begin]->replace(sections[i]);
-      SmallVector<Symbol *> replacedSyms =
-          getRelocTargetSyms<ELFT>(sections[i]);
-      assert(syms.size() == replacedSyms.size() &&
-             "Should have same number of syms!");
-      for (size_t i = 0; i < syms.size(); i++) {
-        if (syms[i] == replacedSyms[i] || !syms[i]->isGlobal() ||
-            !replacedSyms[i]->isGlobal())
-          continue;
-        symbolEquivalence.unionSets(syms[i], replacedSyms[i]);
-      }
 
       // At this point we know sections merged are fully identical and hence
       // we want to remove duplicate implicit dependencies such as link order
@@ -600,26 +563,11 @@ template <class ELFT> void ICF<ELFT>::run() {
           d->folded = true;
         }
   };
-  for (Symbol *sym : ctx.symtab->getSymbols()) {
+  for (Symbol *sym : ctx.symtab->getSymbols())
     fold(sym);
-    auto it = symbolEquivalence.findLeader(sym);
-    if (it != symbolEquivalence.member_end() && *it != sym) {
-      print() << "redirecting '" << sym->getName() << "' in symtab to '"
-              << (*it)->getName() << "'";
-      ctx.symtab->redirect(sym, *it);
-    }
-  }
   parallelForEach(ctx.objectFiles, [&](ELFFileBase *file) {
     for (Symbol *sym : file->getLocalSymbols())
       fold(sym);
-    for (Symbol *&sym : file->getMutableGlobalSymbols()) {
-      auto it = symbolEquivalence.findLeader(sym);
-      if (it != symbolEquivalence.member_end() && *it != sym) {
-        print() << "redirecting '" << sym->getName() << "' to '"
-                << (*it)->getName() << "'";
-        sym = *it;
-      }
-    }
   });
 
   // InputSectionDescription::sections is populated by processSectionCommands().

lld/ELF/SymbolTable.cpp

@@ -29,12 +29,6 @@ using namespace llvm::ELF;
 using namespace lld;
 using namespace lld::elf;
 
-void SymbolTable::redirect(Symbol *from, Symbol *to) {
-  int &fromIdx = symMap[CachedHashStringRef(from->getName())];
-  const int toIdx = symMap[CachedHashStringRef(to->getName())];
-  fromIdx = toIdx;
-}
-
 void SymbolTable::wrap(Symbol *sym, Symbol *real, Symbol *wrap) {
   // Redirect __real_foo to the original foo and foo to the original __wrap_foo.
   int &idx1 = symMap[CachedHashStringRef(sym->getName())];

lld/ELF/SymbolTable.h

@@ -41,7 +41,6 @@ class SymbolTable {
   SymbolTable(Ctx &ctx) : ctx(ctx) {}
   ArrayRef<Symbol *> getSymbols() const { return symVector; }
 
-  void redirect(Symbol *from, Symbol *to);
   void wrap(Symbol *sym, Symbol *real, Symbol *wrap);
 
   Symbol *insert(StringRef name);

lld/test/ELF/aarch64-got-merging-icf.s

@@ -17,12 +17,6 @@
 # EXE-NEXT: selected section {{.*}}:(.text.h1)
 # EXE-NEXT:   removing identical section {{.*}}:(.text.h2)
 # EXE-NEXT:   removing identical section {{.*}}:(.text.h3)
-# EXE-NEXT: redirecting 'f2' in symtab to 'f1'
-# EXE-NEXT: redirecting 'g2' in symtab to 'g1'
-# EXE-NEXT: redirecting 'g3' in symtab to 'g1'
-# EXE-NEXT: redirecting 'f2' to 'f1'
-# EXE-NEXT: redirecting 'g2' to 'g1'
-# EXE-NEXT: redirecting 'g3' to 'g1'
 # EXE-NOT:  {{.}}
 
 ## Definitions are preemptible in a DSO. Only leaf functions can be folded.