[BOLT][DWARF] Skip processing DWARF CUs with a DWO ID but no DWO name

bolt/include/bolt/Core/BinaryContext.h

@@ -326,6 +326,9 @@ class BinaryContext {
   /// Returns true if DWARF4 or lower is used.
   bool isDWARFLegacyUsed() const { return ContainsDwarfLegacy; }
 
+  /// Returns true if DWARFUnit is valid.
+  bool isValidDwarfUnit(DWARFUnit &DU) const;
+
   std::map<unsigned, DwarfLineTable> &getDwarfLineTables() {
     return DwarfLineTablesCUMap;
   }

bolt/include/bolt/Core/DIEBuilder.h

@@ -217,7 +217,8 @@ class DIEBuilder {
                                std::optional<BOLTDWARF5AccelTableData *> Parent,
                                uint32_t NumberParentsInChain);
 
-  void registerUnit(DWARFUnit &DU, bool NeedSort);
+  /// Returns true if DWARFUnit is registered successfully.
+  bool registerUnit(DWARFUnit &DU, bool NeedSort);
 
   /// \return the unique ID of \p U if it exists.
   std::optional<uint32_t> getUnitId(const DWARFUnit &DU);

bolt/lib/Core/BinaryContext.cpp

@@ -1624,10 +1624,26 @@ DWARFContext *BinaryContext::getDWOContext() const {
   return &DWOCUs.begin()->second->getContext();
 }
 
+bool BinaryContext::isValidDwarfUnit(DWARFUnit &DU) const {
+  // Invalid DWARF unit with a DWOId but lacking a dwo_name.
+  if (DU.getDWOId() && !DU.isDWOUnit() &&
+      !DU.getUnitDIE().find(
+          {dwarf::DW_AT_dwo_name, dwarf::DW_AT_GNU_dwo_name})) {
+    this->outs() << "BOLT-ERROR: broken DWARF found in CU at offset 0x"
+                 << Twine::utohexstr(DU.getOffset()) << " (DWOId=0x"
+                 << Twine::utohexstr(*(DU.getDWOId()))
+                 << ", missing DW_AT_dwo_name / DW_AT_GNU_dwo_name)\n";
+    return false;
+  }
+  return true;
+}
+
 /// Handles DWO sections that can either be in .o, .dwo or .dwp files.
 void BinaryContext::preprocessDWODebugInfo() {
   for (const std::unique_ptr<DWARFUnit> &CU : DwCtx->compile_units()) {
     DWARFUnit *const DwarfUnit = CU.get();
+    if (!isValidDwarfUnit(*DwarfUnit))
+      continue;
     if (std::optional<uint64_t> DWOId = DwarfUnit->getDWOId()) {
       std::string DWOName = dwarf::toString(
           DwarfUnit->getUnitDIE().find(

bolt/lib/Core/DIEBuilder.cpp

@@ -584,7 +584,8 @@ DWARFDie DIEBuilder::resolveDIEReference(
   if ((RefCU =
            getUnitForOffset(*this, *DwarfContext, TmpRefOffset, AttrSpec))) {
     /// Trying to add to current working set in case it's cross CU reference.
-    registerUnit(*RefCU, true);
+    if (!registerUnit(*RefCU, true))
+      return DWARFDie();
     DWARFDataExtractor DebugInfoData = RefCU->getDebugInfoExtractor();
     if (DwarfDebugInfoEntry.extractFast(*RefCU, &TmpRefOffset, DebugInfoData,
                                         RefCU->getNextUnitOffset(), 0)) {
@@ -1008,12 +1009,14 @@ static uint64_t getHash(const DWARFUnit &DU) {
   return DU.getOffset();
 }
 
-void DIEBuilder::registerUnit(DWARFUnit &DU, bool NeedSort) {
+bool DIEBuilder::registerUnit(DWARFUnit &DU, bool NeedSort) {
+  if (!BC.isValidDwarfUnit(DU))
+    return false;
   auto IterGlobal = AllProcessed.insert(getHash(DU));
   // If DU is already in a current working set or was already processed we can
   // skip it.
   if (!IterGlobal.second)
-    return;
+    return true;
   if (getState().Type == ProcessingType::DWARF4TUs) {
     getState().DWARF4TUVector.push_back(&DU);
   } else if (getState().Type == ProcessingType::DWARF5TUs) {
@@ -1034,6 +1037,7 @@ void DIEBuilder::registerUnit(DWARFUnit &DU, bool NeedSort) {
   if (getState().DUList.size() == getState().CloneUnitCtxMap.size())
     getState().CloneUnitCtxMap.emplace_back();
   getState().DUList.push_back(&DU);
+  return true;
 }
 
 std::optional<uint32_t> DIEBuilder::getUnitId(const DWARFUnit &DU) {