[llvm] [BOLT] Use symbol table info in registerFragment (PR #89648)

[Maksim Panchenko via llvm-commits]

================
@@ -1417,50 +1418,125 @@ void RewriteInstance::registerFragments() {
   if (!BC->HasSplitFunctions)
     return;
 
+  // Process fragments with ambiguous parents separately as they are typically a
+  // vanishing minority of cases and require expensive symbol table lookups.
+  std::vector<std::pair<StringRef, BinaryFunction *>> AmbiguousFragments;
   for (auto &BFI : BC->getBinaryFunctions()) {
     BinaryFunction &Function = BFI.second;
     if (!Function.isFragment())
       continue;
-    unsigned ParentsFound = 0;
     for (StringRef Name : Function.getNames()) {
-      StringRef BaseName, Suffix;
-      std::tie(BaseName, Suffix) = Name.split('/');
+      StringRef BaseName = NR.restore(Name);
+      const bool IsGlobal = BaseName == Name;
       const size_t ColdSuffixPos = BaseName.find(".cold");
       if (ColdSuffixPos == StringRef::npos)
         continue;
-      // For cold function with local (foo.cold/1) symbol, prefer a parent with
-      // local symbol as well (foo/1) over global symbol (foo).
-      std::string ParentName = BaseName.substr(0, ColdSuffixPos).str();
+      StringRef ParentName = BaseName.substr(0, ColdSuffixPos);
       const BinaryData *BD = BC->getBinaryDataByName(ParentName);
-      if (Suffix != "") {
-        ParentName.append(Twine("/", Suffix).str());
-        const BinaryData *BDLocal = BC->getBinaryDataByName(ParentName);
-        if (BDLocal || !BD)
-          BD = BDLocal;
-      }
-      if (!BD) {
-        if (opts::Verbosity >= 1)
-          BC->outs() << "BOLT-INFO: parent function not found for " << Name
-                     << "\n";
+      const uint64_t NumPossibleLocalParents =
+          NR.getUniquifiedNameCount(ParentName);
+      // The most common case: single local parent fragment.
+      if (!BD && NumPossibleLocalParents == 1) {
+        BD = BC->getBinaryDataByName(NR.getUniqueName(ParentName, 1));
+      } else if (BD && (!NumPossibleLocalParents || IsGlobal)) {
+        // Global parent and either no local candidates (second most common), or
+        // the fragment is global as well (uncommon).
+      } else {
+        // Any other case: need to disambiguate using FILE symbols.
+        AmbiguousFragments.emplace_back(ParentName, &Function);
         continue;
       }
-      const uint64_t Address = BD->getAddress();
-      BinaryFunction *BF = BC->getBinaryFunctionAtAddress(Address);
-      if (!BF) {
-        if (opts::Verbosity >= 1)
-          BC->outs() << formatv(
-              "BOLT-INFO: parent function not found at {0:x}\n", Address);
-        continue;
+      if (BD) {
+        BinaryFunction *BF = BC->getFunctionForSymbol(BD->getSymbol());
+        if (BF) {
+          BC->registerFragment(Function, *BF);
+          continue;
+        }
       }
-      BC->registerFragment(Function, *BF);
-      ++ParentsFound;
-    }
-    if (!ParentsFound) {
       BC->errs() << "BOLT-ERROR: parent function not found for " << Function
                  << '\n';
       exit(1);
     }
   }
+
+  if (AmbiguousFragments.empty())
+    return;
+
+  if (!BC->hasSymbolsWithFileName()) {
+    BC->errs() << "BOLT-ERROR: input file has split functions but does not "
+                  "have FILE symbols. If the binary was stripped, preserve "
+                  "FILE symbols with --keep-file-symbols strip option";
+    exit(1);
+  }
+
+  // The first global symbol is identified by the symbol table sh_info value.
+  // Used as local symbol search stopping point.
+  uint32_t FirstGlobal{0};
+  auto *ELF64LEFile = cast<ELF64LEObjectFile>(InputFile);
+  const ELFFile<ELF64LE> &Obj = ELF64LEFile->getELFFile();
+  for (const auto &Sec : cantFail(Obj.sections())) {
+    if (Sec.sh_type == ELF::SHT_SYMTAB) {
+      FirstGlobal = Sec.sh_info;
+      break;
+    }
+  }
+  if (FirstGlobal == 0) {
----------------
maksfb wrote:

Is it possible to have a symbol table without globals? Even if it’s unlikely but still a valid file, we shouldn’t fail to process it. 

https://github.com/llvm/llvm-project/pull/89648