Add utils to read elf binary and get Build id

Adds code to parse elf binary to get build id from
.note.gnu.build-id section. The build ID is needed to symbolize
heap dumps from the users.

BUG=734705

Change-Id: If4365e232c060ba96071ba1e1c43618f9807e39c
Reviewed-on: https://chromium-review.googlesource.com/1028995
Commit-Queue: Siddhartha S <ssid@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
Cr-Commit-Position: refs/heads/master@{#554304}

base/BUILD.gn

@@ -1164,7 +1164,11 @@ jumbo_component("base") {
     ]
 
     if (is_linux) {
-      sources += [ "base_paths_posix.cc" ]
+      sources += [
+        "base_paths_posix.cc",
+        "debug/elf_reader_linux.cc",
+        "debug/elf_reader_linux.h",
+      ]
     }
   }
 
@@ -1264,6 +1268,8 @@ jumbo_component("base") {
     # Android uses some Linux sources, put those back.
     set_sources_assignment_filter([])
     sources += [
+      "debug/elf_reader_linux.cc",
+      "debug/elf_reader_linux.h",
       "debug/proc_maps_linux.cc",
       "debug/proc_maps_linux.h",
       "files/file_path_watcher_linux.cc",
@@ -2137,6 +2143,7 @@ test("base_unittests") {
     "debug/alias_unittest.cc",
     "debug/crash_logging_unittest.cc",
     "debug/debugger_unittest.cc",
+    "debug/elf_reader_linux_unittest.cc",
     "debug/leak_tracker_unittest.cc",
     "debug/proc_maps_linux_unittest.cc",
     "debug/stack_trace_unittest.cc",
@@ -2578,6 +2585,7 @@ test("base_unittests") {
     deps += [ "//testing/android/native_test:native_test_native_code" ]
     set_sources_assignment_filter([])
     sources += [
+      "debug/elf_reader_linux_unittest.cc",
       "debug/proc_maps_linux_unittest.cc",
       "trace_event/trace_event_android_unittest.cc",
     ]

base/debug/elf_reader_linux.cc

@@ -0,0 +1,102 @@
+// Copyright 2018 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "base/debug/elf_reader_linux.h"
+
+#include <arpa/inet.h>
+#include <elf.h>
+
+#include <vector>
+
+#include "base/bits.h"
+#include "base/containers/span.h"
+#include "base/sha1.h"
+#include "base/strings/stringprintf.h"
+
+extern char __executable_start;
+
+namespace base {
+namespace debug {
+
+namespace {
+
+#if __SIZEOF_POINTER__ == 4
+using Ehdr = Elf32_Ehdr;
+using Half = Elf32_Half;
+using Nhdr = Elf32_Nhdr;
+using Phdr = Elf32_Phdr;
+#else
+using Ehdr = Elf64_Ehdr;
+using Half = Elf64_Half;
+using Nhdr = Elf64_Nhdr;
+using Phdr = Elf64_Phdr;
+#endif
+
+using ElfSegment = span<const char>;
+
+Optional<std::string> ElfSegmentBuildIDNoteAsString(const ElfSegment& segment) {
+  const void* section_end = segment.data() + segment.size_bytes();
+  const Nhdr* note_header = reinterpret_cast<const Nhdr*>(segment.data());
+  while (note_header < section_end) {
+    if (note_header->n_type == NT_GNU_BUILD_ID)
+      break;
+    note_header = reinterpret_cast<const Nhdr*>(
+        reinterpret_cast<const char*>(note_header) + sizeof(Nhdr) +
+        bits::Align(note_header->n_namesz, 4) +
+        bits::Align(note_header->n_descsz, 4));
+  }
+
+  if (note_header >= section_end || note_header->n_descsz != kSHA1Length)
+    return nullopt;
+
+  const uint8_t* guid = reinterpret_cast<const uint8_t*>(note_header) +
+                        sizeof(Nhdr) + bits::Align(note_header->n_namesz, 4);
+
+  uint32_t dword = htonl(*reinterpret_cast<const int32_t*>(guid));
+  uint16_t word1 = htons(*reinterpret_cast<const int16_t*>(guid + 4));
+  uint16_t word2 = htons(*reinterpret_cast<const int16_t*>(guid + 6));
+  std::string identifier;
+  identifier.reserve(kSHA1Length * 2);  // as hex string
+  SStringPrintf(&identifier, "%08X%04X%04X", dword, word1, word2);
+  for (size_t i = 8; i < note_header->n_descsz; ++i)
+    StringAppendF(&identifier, "%02X", guid[i]);
+
+  return identifier;
+}
+
+std::vector<ElfSegment> FindElfSegments(const char* elf_base,
+                                        uint32_t segment_type) {
+  if (strncmp(static_cast<const char*>(elf_base), ELFMAG, SELFMAG) != 0)
+    return std::vector<ElfSegment>();
+
+  const Ehdr* elf_header = reinterpret_cast<const Ehdr*>(elf_base);
+  const Phdr* phdrs =
+      reinterpret_cast<const Phdr*>(elf_base + elf_header->e_phoff);
+  std::vector<ElfSegment> segments;
+  for (Half i = 0; i < elf_header->e_phnum; ++i) {
+    if (phdrs[i].p_type == segment_type)
+      segments.push_back({elf_base + phdrs[i].p_offset, phdrs[i].p_filesz});
+  }
+  return segments;
+}
+
+}  // namespace
+
+Optional<std::string> ReadElfBuildId() {
+  // Elf program headers can have multiple PT_NOTE arrays.
+  std::vector<ElfSegment> segs = FindElfSegments(&__executable_start, PT_NOTE);
+  if (segs.empty())
+    return nullopt;
+  Optional<std::string> id;
+  for (const ElfSegment& seg : segs) {
+    id = ElfSegmentBuildIDNoteAsString(seg);
+    if (id)
+      return id;
+  }
+
+  return nullopt;
+}
+
+}  // namespace debug
+}  // namespace base

base/debug/elf_reader_linux.h

@@ -0,0 +1,23 @@
+// Copyright 2018 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef BASE_DEBUG_ELF_READER_LINUX_H_
+#define BASE_DEBUG_ELF_READER_LINUX_H_
+
+#include <string>
+
+#include "base/base_export.h"
+#include "base/optional.h"
+
+namespace base {
+namespace debug {
+
+// Returns the ELF section .note.gnu.build-id from current executable file, if
+// present.
+Optional<std::string> BASE_EXPORT ReadElfBuildId();
+
+}  // namespace debug
+}  // namespace base
+
+#endif  // BASE_DEBUG_ELF_READER_LINUX_H_

base/debug/elf_reader_linux_unittest.cc

@@ -0,0 +1,30 @@
+// Copyright 2018 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "base/debug/elf_reader_linux.h"
+
+#include "base/strings/string_util.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+namespace base {
+namespace debug {
+
+// The linker flag --build-id is passed only on official builds. Clang does not
+// enable it by default and we do not have build id section in non-official
+// builds.
+#if defined(OFFICIAL_BUILD)
+TEST(ElfReaderTest, ReadElfBuildId) {
+  Optional<std::string> build_id = ReadElfBuildId();
+  ASSERT_TRUE(build_id);
+  const size_t kGuidBytes = 20;
+  EXPECT_EQ(2 * kGuidBytes, build_id.value().size());
+  for (char c : *build_id) {
+    EXPECT_TRUE(IsHexDigit(c));
+    EXPECT_FALSE(IsAsciiLower(c));
+  }
+}
+#endif
+
+}  // namespace debug
+}  // namespace base