Correctly update .debug_addr DWARF section.

This is a minimal patch to add support for also updating the addresses
stored in the .debug_addr DWARF section, which is used to perform
address resolution at debug time with debug fission. This is minimal in
the sence that it only supports the pre-DWARFv5 .debug_addr format that
is emitted by Emscripten currently, and which only consists of a simple
list of addresses.

Ideally the copy of DWARFYAML in the Binaryen tree should be updated
with LLVM ToT at some point, which will provide full support for DWARFv5
index tables, including the new .debug_addr format.

Ref: #3460
Ref: emscripten-core/emscripten#13099
Bug: https://crbug.com/1161422

Author: bmeurer

src/wasm/wasm-debug.cpp

@@ -1045,6 +1045,19 @@ static void updateLoc(llvm::DWARFYAML::Data& yaml,
   }
 }
 
+static void updateAddr(llvm::DWARFYAML::Data& yaml,
+                       const LocationUpdater& locationUpdater) {
+  for (auto& addrTable : yaml.DebugAddr) {
+    for (auto& addr : addrTable.Addrs) {
+      BinaryLocation start = addr, newStart = 0;
+      if (!isTombstone(start)) {
+        newStart = locationUpdater.getNewStart(start);
+      }
+      addr = newStart;
+    }
+  }
+}
+
 void writeDWARFSections(Module& wasm, const BinaryLocations& newLocations) {
   BinaryenDWARFInfo info(wasm);
 
@@ -1064,6 +1077,8 @@ void writeDWARFSections(Module& wasm, const BinaryLocations& newLocations) {
 
   updateLoc(data, locationUpdater);
 
+  updateAddr(data, locationUpdater);
+
   // Convert to binary sections.
   auto newSections =
     EmitDebugSections(data, false /* EmitFixups for debug_info */);

third_party/llvm-project/DWARFEmitter.cpp

@@ -148,6 +148,17 @@ void DWARFYAML::EmitDebugLoc(raw_ostream &OS, const DWARFYAML::Data &DI) {
   }
 }
 
+// XXX BINARYEN
+void DWARFYAML::EmitDebugAddr(raw_ostream &OS, const DWARFYAML::Data &DI) {
+  // This is the legacy DWARF extension .debug_addr format, which is
+  // only a simple list of addresses. This doesn't support DWARF5.
+  for (auto AddrTable : DI.DebugAddr) {
+    for (auto Addr : AddrTable.Addrs) {
+      writeInteger((uint32_t)Addr, OS, DI.IsLittleEndian);
+    }
+  }
+}
+
 void DWARFYAML::EmitPubSection(raw_ostream &OS,
                                const DWARFYAML::PubSection &Sect,
                                bool IsLittleEndian) {
@@ -485,6 +496,8 @@ EmitDebugSections(llvm::DWARFYAML::Data &DI, bool ApplyFixups) {
                        DebugSections); // XXX BINARYEN
   EmitDebugSectionImpl(DI, &DWARFYAML::EmitDebugLoc, "debug_loc",
                        DebugSections); // XXX BINARYEN
+  EmitDebugSectionImpl(DI, &DWARFYAML::EmitDebugAddr, "debug_addr",
+                       DebugSections); // XXX BINARYEN
   return std::move(DebugSections);
 }
 } // namespace DWARFYAML

third_party/llvm-project/dwarf2yaml.cpp

@@ -8,6 +8,7 @@
 
 #include "Error.h"
 #include "llvm/DebugInfo/DWARF/DWARFContext.h"
+#include "llvm/DebugInfo/DWARF/DWARFDebugAddr.h"
 #include "llvm/DebugInfo/DWARF/DWARFDebugArangeSet.h"
 #include "llvm/DebugInfo/DWARF/DWARFFormValue.h"
 #include "llvm/ObjectYAML/DWARFYAML.h"
@@ -89,6 +90,31 @@ void dumpDebugARanges(DWARFContext &DCtx, DWARFYAML::Data &Y) {
   }
 }
 
+void dumpDebugAddr(DWARFContext &DCtx, DWARFYAML::Data &Y) { // XXX BINARYEN
+  DWARFDebugAddrTable AddrTable;
+  DWARFDataExtractor AddrData(DCtx.getDWARFObj(),
+                              DCtx.getDWARFObj().getAddrSection(),
+                              DCtx.isLittleEndian(), /*AddrSize=*/0);
+  std::vector<DWARFYAML::AddrTable> AddrTables;
+  uint64_t Offset = 0;
+  while (AddrData.isValidOffset(Offset)) {
+    // We ignore any errors that don't prevent parsing the section, since we can
+    // still represent such sections.
+    if (Error E = AddrTable.extract(AddrData, &Offset, DCtx.getMaxVersion(), DCtx.getDWARFObj().getAddressSize(), consumeError)) {
+      errs() << toString(std::move(E)) << '\n';
+      break;
+    }
+    AddrTables.emplace_back();
+    for (uint64_t Addr : AddrTable.getAddressEntries()) {
+      // Currently, the parser doesn't support parsing an address table with non
+      // linear addresses (segment_selector_size != 0). The segment selectors
+      // are specified to be zero.
+      AddrTables.back().Addrs.push_back(Addr);
+    }
+  }
+  Y.DebugAddr = std::move(AddrTables);
+}
+
 void dumpDebugRanges(DWARFContext &DCtx, DWARFYAML::Data &Y) { // XXX BINARYEN
   uint8_t savedAddressByteSize = 4;
   DWARFDataExtractor rangesData(DCtx.getDWARFObj(), DCtx.getDWARFObj().getRangesSection(),
@@ -427,6 +453,7 @@ std::error_code dwarf2yaml(DWARFContext &DCtx, DWARFYAML::Data &Y) {
   dumpDebugARanges(DCtx, Y);
   dumpDebugRanges(DCtx, Y); // XXX BINARYEN
   dumpDebugLoc(DCtx, Y); // XXX BINARYEN
+  dumpDebugAddr(DCtx, Y); // XXX BINARYEN
   dumpDebugPubSections(DCtx, Y);
   dumpDebugInfo(DCtx, Y);
   dumpDebugLines(DCtx, Y);

third_party/llvm-project/include/llvm/DebugInfo/DWARF/DWARFDebugAddr.h

@@ -90,6 +90,9 @@ class DWARFDebugAddrTable {
 
   /// Returns the length of the array of addresses.
   uint32_t getDataSize() const;
+
+  /// Return the parsed addresses of this table.
+  ArrayRef<uint64_t> getAddressEntries() const { return Addrs; }
 };
 
 } // end namespace llvm

third_party/llvm-project/include/llvm/ObjectYAML/DWARFEmitter.h

@@ -34,6 +34,7 @@ void EmitDebugStr(raw_ostream &OS, const Data &DI);
 void EmitDebugAranges(raw_ostream &OS, const Data &DI);
 void EmitDebugRanges(raw_ostream &OS, const Data &DI); // XXX BINARYEN
 void EmitDebugLoc(raw_ostream &OS, const Data &DI); // XXX BINARYEN
+void EmitDebugAddr(raw_ostream &OS, const Data &DI); // XXX BINARYEN
 void EmitPubSection(raw_ostream &OS, const PubSection &Sect,
                     bool IsLittleEndian);
 void EmitDebugInfo(raw_ostream &OS, const Data &DI);

third_party/llvm-project/include/llvm/ObjectYAML/DWARFYAML.h

@@ -94,6 +94,10 @@ struct Loc {
   std::vector<uint8_t> Location;
   uint64_t CompileUnitOffset;
 };
+
+struct AddrTable {
+  std::vector<uint64_t> Addrs;
+};
 // XXX BINARYEN -->
 
 struct PubEntry {
@@ -173,6 +177,7 @@ struct Data {
   std::vector<ARange> ARanges;
   std::vector<Range> Ranges; // XXX BINARYEN
   std::vector<Loc> Locs; // XXX BINARYEN
+  std::vector<AddrTable> DebugAddr; // XXX BINARYEN
   PubSection PubNames;
   PubSection PubTypes;
 
@@ -195,6 +200,7 @@ LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::ARangeDescriptor)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::ARange)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::Range) // XXX BINARYEN
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::Loc) // XXX BINARYEN
+LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::AddrTable) // XXX BINARYEN
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::PubEntry)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::Unit)
 LLVM_YAML_IS_SEQUENCE_VECTOR(llvm::DWARFYAML::FormValue)
@@ -234,6 +240,10 @@ template <> struct MappingTraits<DWARFYAML::Loc> { // XXX BINARYEN
   static void mapping(IO &IO, DWARFYAML::Loc &Loc);
 };
 
+template <> struct MappingTraits<DWARFYAML::AddrTable> { // XXX BINARYEN
+  static void mapping(IO &IO, DWARFYAML::AddrTable &AddrTable);
+};
+
 template <> struct MappingTraits<DWARFYAML::PubEntry> {
   static void mapping(IO &IO, DWARFYAML::PubEntry &Entry);
 };