DWARF processing fixes (#224)

Add support for binaries with both .debug_ranges and .debug_rnglists sections.

.debug_rnglists was added in DWARF5 to replace .debug_ranges. This change
handles cases when the binary has compilation units with DWARF5 info and
compilation units with pre-version-5 DWARF info.

Fix processing of DW_AT_ranges attributes to allow DW_FORM_data4 and DW_FORM_data8
This affects DWARF pre-version-4 and was broken by
dd1395a .

This patch also adds a test with a binary from #154 .

Fixes #154

CMakeLists.txt

@@ -21,8 +21,12 @@ endfunction()
 function (build_gcov)
   add_subdirectory(third_party/abseil)
   add_subdirectory(third_party/glog)
+  add_subdirectory(third_party/googletest)
 
   add_custom_target(exclude_extlib_tests ALL
+    COMMAND rm -f ${gtest_BINARY_DIR}/CTestTestfile.cmake
+    COMMAND rm -f ${gmock_BINARY_DIR}/CTestTestfile.cmake
+    COMMAND rm -f ${googletest-distribution_BINARY_DIR}/CTestTestfile.cmake
     COMMAND rm -f ${glog_BINARY_DIR}/CTestTestfile.cmake
     COMMAND rm -f ${absl_BINARY_DIR}/CTestTestfile.cmake)
 
@@ -31,6 +35,8 @@ function (build_gcov)
     third_party/abseil
     third_party/perf_data_converter/src
     third_party/perf_data_converter/src/quipper
+    third_party/googletest/googletest/include
+    third_party/googletest/googlemock/include
     util
     ${PROJECT_BINARY_DIR}
     ${PROJECT_BINARY_DIR}/third_party/glog
@@ -39,23 +45,26 @@ function (build_gcov)
   find_library (LIBELF_LIBRARIES NAMES elf REQUIRED)
   find_library (LIBCRYPTO_LIBRARIES NAMES crypto REQUIRED)
 
+  add_library(addr2line_lib OBJECT
+    legacy_addr2line.cc
+    util/symbolize/addr2line_inlinestack.cc
+    util/symbolize/bytereader.cc
+    util/symbolize/functioninfo.cc
+    util/symbolize/dwarf2reader.cc
+    util/symbolize/dwarf3ranges.cc
+    util/symbolize/elf_reader.cc
+    util/symbolize/index_helper.cc
+  )
+
   add_library(create_gcov_lib OBJECT
     create_gcov.cc
     gcov.cc
     instruction_map.cc
-    legacy_addr2line.cc
     profile.cc
     profile_creator.cc
     profile_writer.cc
     sample_reader.cc
     symbol_map.cc
-    util/symbolize/addr2line_inlinestack.cc
-    util/symbolize/bytereader.cc
-    util/symbolize/functioninfo.cc
-    util/symbolize/dwarf2reader.cc
-    util/symbolize/dwarf3ranges.cc
-    util/symbolize/elf_reader.cc
-    util/symbolize/index_helper.cc
   )
   target_link_libraries(create_gcov_lib perf_proto)
 
@@ -94,6 +103,7 @@ function (build_gcov)
     absl::flags
     absl::flags_parse
     create_gcov_lib
+    addr2line_lib
     glog
     quipper_perf
   )
@@ -136,6 +146,24 @@ function (build_gcov)
     dump_gcov_lib
     glog)
 
+    add_executable(addr2line_test addr2line_test.cc)
+    target_link_libraries(addr2line_test
+      absl::base
+      gtest
+      gtest_main
+      absl::flags
+      absl::flags_parse
+      addr2line_lib
+      glog
+      quipper_perf)
+    add_test(NAME addr2line_test COMMAND addr2line_test)
+
+    add_custom_command(PRE_BUILD
+      OUTPUT prepare_cmds
+      COMMAND ln -s -f ${CMAKE_HOME_DIRECTORY}/testdata)
+    add_custom_target(prepare ALL
+      DEPENDS prepare_cmds)
+
 endfunction()
 
 function (build_llvm)

addr2line_test.cc

@@ -0,0 +1,18 @@
+#include "addr2line.h"
+#include "third_party/abseil/absl/strings/str_cat.h"
+#include "util/testing/status_matchers.h"
+
+namespace {
+
+using ::devtools_crosstool_autofdo::Addr2line;
+
+TEST(Addr2lineTest, Dwarf2Dwarf5Binary) {
+  const std::string binary =
+      absl::StrCat(::testing::SrcDir(),
+                   "/testdata/"
+                   "dwarf2_dwarf5.bin");
+
+  Addr2line* addr2line = Addr2line::Create(binary);
+  EXPECT_TRUE(addr2line != NULL);
+}
+}  // namespace

legacy_addr2line.cc

@@ -105,27 +105,28 @@ bool Google3Addr2line::Prepare() {
   size_t debug_info_size = 0;
   size_t debug_addr_size = 0;
   size_t debug_ranges_size = 0;
+  size_t debug_rnglists_size = 0;
   const char *debug_info_data = NULL;
   const char *debug_addr_data = NULL;
   const char *debug_ranges_data = NULL;
+  const char *debug_rnglists_data = NULL;
   bool is_rnglists_section = false;
   GetSection(sections, ".debug_info", &debug_info_data, &debug_info_size);
   if (debug_info_data == NULL)
     LOG(WARNING) << "File '" << binary_name_ << "' does not have .debug_info section.";
   GetSection(sections, ".debug_addr", &debug_addr_data, &debug_addr_size);
   GetSection(sections, ".debug_ranges", &debug_ranges_data, &debug_ranges_size);
-  if (debug_ranges_data == NULL) {
-      GetSection(sections, ".debug_rnglists", &debug_ranges_data, &debug_ranges_size);
-      if (debug_ranges_data != NULL) 
-        is_rnglists_section = true;
-      else 
-        LOG(WARNING) << "File '" << binary_name_ << "' does not have .debug_ranges nor .debug_rnglists sections.";
+  GetSection(sections, ".debug_rnglists", &debug_rnglists_data, &debug_rnglists_size);
+
+  if (debug_ranges_data == NULL && debug_rnglists_data == NULL) {
+    LOG(WARNING) << "File '" << binary_name_ << "' does not have .debug_ranges nor .debug_rnglists sections.";
   }
 
   AddressRangeList debug_ranges(debug_ranges_data,
                                 debug_ranges_size,
+                                debug_rnglists_data,
+                                debug_rnglists_size,
                                 &reader,
-                                is_rnglists_section, 
                                 debug_addr_data, 
                                 debug_addr_size);
   inline_stack_handler_ = new InlineStackHandler(

util/symbolize/addr2line_inlinestack.cc

@@ -99,7 +99,7 @@ bool InlineStackHandler::StartCompilationUnit(uint64 offset,
                                               uint8 /*address_size*/,
                                               uint8 /*offset_size*/,
                                               uint64 /*cu_length*/,
-                                              uint8 /*dwarf_version*/) {
+                                              uint8 dwarf_version) {
   CHECK(subprogram_stack_.empty());
   compilation_unit_offset_ = offset;
   compilation_unit_base_ = 0;
@@ -109,6 +109,7 @@ bool InlineStackHandler::StartCompilationUnit(uint64 offset,
     input_file_index_ = 0;
     subprograms_by_offset_maps_.push_back(new SubprogramsByOffsetMap);
   }
+  dwarf_version_ = dwarf_version;
   return true;
 }
 
@@ -348,13 +349,14 @@ void InlineStackHandler::ProcessAttributeUnsigned(
       case DW_AT_ranges: {
         CHECK_EQ(0, subprogram_stack_.back()->address_ranges()->size());
         AddressRangeList::RangeList ranges;
-        if (form == DW_FORM_sec_offset) {
-            address_ranges_->ReadRangeList(data, compilation_unit_base_, &ranges);
+
+        if (form == DW_FORM_sec_offset || form == DW_FORM_data4 || form == DW_FORM_data8) {
+          address_ranges_->ReadRangeList(data, compilation_unit_base_, &ranges, dwarf_version_);
         }
         else {
-            CHECK(form == DW_FORM_rnglistx);
-            uint64 address_ = address_ranges_->GetRngListsElementOffsetByIndex(ranges_base_, data);
-            address_ranges_->ReadDwarfRngListwithOffsetArray(address_, compilation_unit_base_, &ranges, addr_base_, ranges_base_);
+          CHECK(form == DW_FORM_rnglistx);
+          uint64 address_ = address_ranges_->GetRngListsElementOffsetByIndex(ranges_base_, data);
+          address_ranges_->ReadDwarfRngListwithOffsetArray(address_, compilation_unit_base_, &ranges, addr_base_, ranges_base_);
         }
 
         if (subprogram_stack_.size() == 1) {
@@ -440,7 +442,7 @@ void InlineStackHandler::ProcessAttributeUnsigned(
         str_offset_base_ = data;
         break;
       case DW_AT_ranges: 
-        CHECK(form == DW_FORM_sec_offset);
+        CHECK(form == DW_FORM_sec_offset || form == DW_FORM_data4 || form == DW_FORM_data8);
         FALLTHROUGH_INTENDED;
       case DW_AT_GNU_ranges_base:
       case DW_AT_rnglists_base:

util/symbolize/addr2line_inlinestack.h

@@ -282,6 +282,7 @@ class InlineStackHandler: public Dwarf2Handler {
   int overlap_count_;
   bool have_two_level_line_tables_;
   bool subprogram_added_by_cu_;
+  uint8 dwarf_version_;
 
   DISALLOW_COPY_AND_ASSIGN(InlineStackHandler);
 };

util/symbolize/dwarf2reader.cc

@@ -1017,15 +1017,17 @@ void CompilationUnit::ProcessSplitDwarf() {
         auto iter = sections.find(".debug_rnglists");
         size_t debug_ranges_size = 0;
         const char *debug_ranges_data = NULL;
+        size_t debug_rnglist_size = 0;
+        const char *debug_rnglist_data = NULL;
         if (iter != sections.end()) {
-          debug_ranges_data = iter->second.first;
-          debug_ranges_size = iter->second.second;
-          const bool is_rnglists_section = true;
+          debug_rnglist_data = iter->second.first;
+          debug_rnglist_size = iter->second.second;
           AddressRangeList *debug_ranges = 
                 new AddressRangeList(debug_ranges_data,
                                       debug_ranges_size,
+                                      debug_rnglist_data,
+                                      debug_rnglist_size,
                                       &reader,
-                                      is_rnglists_section,
                                       addr_buffer_, 
                                       addr_buffer_length_);
           InlineStackHandler *inlinestackhandler = dynamic_cast<InlineStackHandler*>(handler_);