issue 243 - Linux dumper should use build id produced by ld --build-id if available

R=thestig at http://breakpad.appspot.com/185001

git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@830 4c0a9323-5329-0410-9bdc-e9ce6186880e

5 files changed, 260 insertions, 96 deletions

diff --git a/src/common/linux/file_id.cc b/src/common/linux/file_id.cc
index 2227c83b..acd448a9 100644
--- a/src/common/linux/file_id.cc
+++ b/src/common/linux/file_id.cc
@@ -56,39 +56,46 @@
 
 namespace google_breakpad {
 
+#ifndef NT_GNU_BUILD_ID
+#define NT_GNU_BUILD_ID 3
+#endif
+
 FileID::FileID(const char* path) {
   strncpy(path_, path, sizeof(path_));
 }
 
 struct ElfClass32 {
   typedef Elf32_Ehdr Ehdr;
+  typedef Elf32_Nhdr Nhdr;
   typedef Elf32_Shdr Shdr;
   static const int kClass = ELFCLASS32;
 };
 
 struct ElfClass64 {
   typedef Elf64_Ehdr Ehdr;
+  typedef Elf64_Nhdr Nhdr;
   typedef Elf64_Shdr Shdr;
   static const int kClass = ELFCLASS64;
 };
 
-// These three functions are also used inside the crashed process, so be safe
+// These six functions are also used inside the crashed process, so be safe
 // and use the syscall/libc wrappers instead of direct syscalls or libc.
 template<typename ElfClass>
-static void FindElfClassTextSection(const char *elf_base,
-                                    const void **text_start,
-                                    int *text_size) {
+static void FindElfClassSection(const char *elf_base,
+                                const char *section_name,
+                                uint32_t section_type,
+                                const void **section_start,
+                                int *section_size) {
   typedef typename ElfClass::Ehdr Ehdr;
   typedef typename ElfClass::Shdr Shdr;
 
   assert(elf_base);
-  assert(text_start);
-  assert(text_size);
+  assert(section_start);
+  assert(section_size);
 
   assert(my_strncmp(elf_base, ELFMAG, SELFMAG) == 0);
 
-  const char* text_section_name = ".text";
-  int name_len = my_strlen(text_section_name);
+  int name_len = my_strlen(section_name);
 
   const Ehdr* elf_header = reinterpret_cast<const Ehdr*>(elf_base);
   assert(elf_header->e_ident[EI_CLASS] == ElfClass::kClass);
@@ -97,30 +104,41 @@ static void FindElfClassTextSection(const char *elf_base,
       reinterpret_cast<const Shdr*>(elf_base + elf_header->e_shoff);
   const Shdr* string_section = sections + elf_header->e_shstrndx;
 
-  const Shdr* text_section = NULL;
+  const Shdr* section = NULL;
   for (int i = 0; i < elf_header->e_shnum; ++i) {
-    if (sections[i].sh_type == SHT_PROGBITS) {
+    if (sections[i].sh_type == section_type) {
       const char* section_name = (char*)(elf_base +
                                          string_section->sh_offset +
                                          sections[i].sh_name);
-      if (!my_strncmp(section_name, text_section_name, name_len)) {
-        text_section = &sections[i];
+      if (!my_strncmp(section_name, section_name, name_len)) {
+        section = &sections[i];
         break;
       }
     }
   }
-  if (text_section != NULL && text_section->sh_size > 0) {
-    *text_start = elf_base + text_section->sh_offset;
-    *text_size = text_section->sh_size;
+  if (section != NULL && section->sh_size > 0) {
+    *section_start = elf_base + section->sh_offset;
+    *section_size = section->sh_size;
   }
 }
 
-static bool FindElfTextSection(const void *elf_mapped_base,
-                               const void **text_start,
-                               int *text_size) {
+// Attempt to find a section named |section_name| of type |section_type|
+// in the ELF binary data at |elf_mapped_base|. On success, returns true
+// and sets |*section_start| to point to the start of the section data,
+// and |*section_size| to the size of the section's data. If |elfclass|
+// is not NULL, set |*elfclass| to the ELF file class.
+static bool FindElfSection(const void *elf_mapped_base,
+                           const char *section_name,
+                           uint32_t section_type,
+                           const void **section_start,
+                           int *section_size,
+                           int *elfclass) {
   assert(elf_mapped_base);
-  assert(text_start);
-  assert(text_size);
+  assert(section_start);
+  assert(section_size);
+
+  *section_start = NULL;
+  *section_size = 0;
 
   const char* elf_base =
     static_cast<const char*>(elf_mapped_base);
@@ -129,14 +147,89 @@ static bool FindElfTextSection(const void *elf_mapped_base,
   if (my_strncmp(elf_base, ELFMAG, SELFMAG) != 0)
     return false;
 
+  if (elfclass) {
+    *elfclass = elf_header->e_ident[EI_CLASS];
+  }
+
   if (elf_header->e_ident[EI_CLASS] == ELFCLASS32) {
-    FindElfClassTextSection<ElfClass32>(elf_base, text_start, text_size);
+    FindElfClassSection<ElfClass32>(elf_base, section_name, section_type,
+                                    section_start, section_size);
+    return *section_start != NULL;
   } else if (elf_header->e_ident[EI_CLASS] == ELFCLASS64) {
-    FindElfClassTextSection<ElfClass64>(elf_base, text_start, text_size);
-  } else {
+    FindElfClassSection<ElfClass64>(elf_base, section_name, section_type,
+                                    section_start, section_size);
+    return *section_start != NULL;
+  }
+
+  return false;
+}
+
+template<typename ElfClass>
+static bool ElfClassBuildIDNoteIdentifier(const void *section,
+                                          uint8_t identifier[kMDGUIDSize])
+{
+  typedef typename ElfClass::Nhdr Nhdr;
+
+  const Nhdr* note_header = reinterpret_cast<const Nhdr*>(section);
+  if (note_header->n_type != NT_GNU_BUILD_ID ||
+      note_header->n_descsz == 0) {
     return false;
   }
 
+  const char* build_id = reinterpret_cast<const char*>(section) +
+    sizeof(Nhdr) + note_header->n_namesz;
+  // Copy as many bits of the build ID as will fit
+  // into the GUID space.
+  my_memset(identifier, 0, kMDGUIDSize);
+  memcpy(identifier, build_id,
+         std::min(kMDGUIDSize, (size_t)note_header->n_descsz));
+
+  return true;
+}
+
+// Attempt to locate a .note.gnu.build-id section in an ELF binary
+// and copy as many bytes of it as will fit into |identifier|.
+static bool FindElfBuildIDNote(const void *elf_mapped_base,
+                               uint8_t identifier[kMDGUIDSize])
+{
+  void* note_section;
+  int note_size, elfclass;
+  if (!FindElfSection(elf_mapped_base, ".note.gnu.build-id", SHT_NOTE,
+                      (const void**)&note_section, &note_size, &elfclass) ||
+      note_size == 0) {
+    return false;
+  }
+
+  if (elfclass == ELFCLASS32) {
+    return ElfClassBuildIDNoteIdentifier<ElfClass32>(note_section, identifier);
+  } else if (elfclass == ELFCLASS64) {
+    return ElfClassBuildIDNoteIdentifier<ElfClass64>(note_section, identifier);
+  }
+
+  return false;
+}
+
+// Attempt to locate the .text section of an ELF binary and generate
+// a simple hash by XORing the first page worth of bytes into |identifier|.
+static bool HashElfTextSection(const void *elf_mapped_base,
+                               uint8_t identifier[kMDGUIDSize]) {
+
+  void* text_section;
+  int text_size;
+  if (!FindElfSection(elf_mapped_base, ".text", SHT_PROGBITS,
+                      (const void**)&text_section, &text_size, NULL) ||
+      text_size == 0) {
+    return false;
+  }
+
+  my_memset(identifier, 0, kMDGUIDSize);
+  const uint8_t* ptr = reinterpret_cast<const uint8_t*>(text_section);
+  const uint8_t* ptr_end = ptr + std::min(text_size, 4096);
+  while (ptr < ptr_end) {
+    for (unsigned i = 0; i < kMDGUIDSize; i++)
+      identifier[i] ^= ptr[i];
+    ptr += kMDGUIDSize;
+  }
   return true;
 }
 
@@ -144,21 +237,12 @@ static bool FindElfTextSection(const void *elf_mapped_base,
 bool FileID::ElfFileIdentifierFromMappedFile(void* base,
                                              uint8_t identifier[kMDGUIDSize])
 {
-  const void* text_section = NULL;
-  int text_size = 0;
-  bool success = false;
-  if (FindElfTextSection(base, &text_section, &text_size) && (text_size > 0)) {
-    my_memset(identifier, 0, kMDGUIDSize);
-    const uint8_t* ptr = reinterpret_cast<const uint8_t*>(text_section);
-    const uint8_t* ptr_end = ptr + std::min(text_size, 4096);
-    while (ptr < ptr_end) {
-      for (unsigned i = 0; i < kMDGUIDSize; i++)
-        identifier[i] ^= ptr[i];
-      ptr += kMDGUIDSize;
-    }
-    success = true;
-  }
-  return success;
+  // Look for a build id note first.
+  if (FindElfBuildIDNote(base, identifier))
+    return true;
+
+  // Fall back on hashing the first page of the text section.
+  return HashElfTextSection(base, identifier);
 }
 
 bool FileID::ElfFileIdentifier(uint8_t identifier[kMDGUIDSize]) {
diff --git a/src/common/linux/file_id.h b/src/common/linux/file_id.h
index 2cd4953e..4d5b52dc 100644
--- a/src/common/linux/file_id.h
+++ b/src/common/linux/file_id.h
@@ -49,8 +49,9 @@ class FileID {
   // Load the identifier for the elf file path specified in the constructor into
   // |identifier|.  Return false if the identifier could not be created for the
   // file.
-  // The current implementation will XOR the first 4096 bytes of the
-  // .text section to generate an identifier.
+  // The current implementation will look for a .note.gnu.build-id
+  // section and use that as the file id, otherwise it falls back to
+  // XORing the first 4096 bytes of the .text section to generate an identifier.
   bool ElfFileIdentifier(uint8_t identifier[kMDGUIDSize]);
 
   // Load the identifier for the elf file mapped into memory at |base| into
diff --git a/src/common/linux/file_id_unittest.cc b/src/common/linux/file_id_unittest.cc
index f5298b12..24208983 100644
--- a/src/common/linux/file_id_unittest.cc
+++ b/src/common/linux/file_id_unittest.cc
@@ -33,18 +33,20 @@
 #include <stdlib.h>
 
 #include "common/linux/file_id.h"
+#include "common/linux/synth_elf.h"
+#include "common/test_assembler.h"
 #include "breakpad_googletest_includes.h"
 
 using namespace google_breakpad;
-
-namespace {
-typedef testing::Test FileIDTest;
-}
-
-TEST(FileIDTest, FileIDStrip) {
-  // Calculate the File ID of our binary using
-  // FileID::ElfFileIdentifier, then make a copy of our binary,
-  // strip it, and ensure that we still get the same result.
+using google_breakpad::synth_elf::BuildIDNote;
+using google_breakpad::synth_elf::ELF;
+using google_breakpad::test_assembler::kLittleEndian;
+using google_breakpad::test_assembler::Section;
+
+TEST(FileIDStripTest, StripSelf) {
+  // Calculate the File ID of this binary using
+  // FileID::ElfFileIdentifier, then make a copy of this binary,
+  // strip it, and ensure that the result is the same.
   char exe_name[PATH_MAX];
   ssize_t len = readlink("/proc/self/exe", exe_name, PATH_MAX - 1);
   ASSERT_NE(len, -1);
@@ -75,58 +77,40 @@ TEST(FileIDTest, FileIDStrip) {
   unlink(templ);
 }
 
-struct ElfClass32 {
-  typedef Elf32_Ehdr Ehdr;
-  typedef Elf32_Shdr Shdr;
-  static const int kClass = ELFCLASS32;
-};
+class FileIDTest : public testing::Test {
+public:
+  void GetElfContents(ELF& elf) {
+    string contents;
+    ASSERT_TRUE(elf.GetContents(&contents));
+    ASSERT_LT(0, contents.size());
 
-struct ElfClass64 {
-  typedef Elf64_Ehdr Ehdr;
-  typedef Elf64_Shdr Shdr;
-  static const int kClass = ELFCLASS64;
-};
-
-template<typename ElfClass>
-struct ElfishElf {
-  static const size_t kTextSectionSize = 128;
-  typedef typename ElfClass::Ehdr Ehdr;
-  typedef typename ElfClass::Shdr Shdr;
-
-  Ehdr elf_header;
-  Shdr text_header;
-  Shdr string_header;
-  char text_section[kTextSectionSize];
-  char string_section[8];
-
-  static void Populate(ElfishElf* elf) {
-    memset(elf, 0, sizeof(ElfishElf));
-    memcpy(elf, ELFMAG, SELFMAG);
-    elf->elf_header.e_ident[EI_CLASS] = ElfClass::kClass;
-    elf->elf_header.e_shoff = offsetof(ElfishElf, text_header);
-    elf->elf_header.e_shnum = 2;
-    elf->elf_header.e_shstrndx = 1;
-    elf->text_header.sh_name = 0;
-    elf->text_header.sh_type = SHT_PROGBITS;
-    elf->text_header.sh_offset = offsetof(ElfishElf, text_section);
-    elf->text_header.sh_size = kTextSectionSize;
-    for (size_t i = 0; i < kTextSectionSize; ++i) {
-      elf->text_section[i] = i * 3;
-    }
-    elf->string_header.sh_offset = offsetof(ElfishElf, string_section);
-    strcpy(elf->string_section, ".text");
+    elfdata_v.clear();
+    elfdata_v.insert(elfdata_v.begin(), contents.begin(), contents.end());
+    elfdata = &elfdata_v[0];
   }
+
+  vector<uint8_t> elfdata_v;
+  uint8_t* elfdata;
 };
 
-TEST(FileIDTest, ElfClass) {
+TEST_F(FileIDTest, ElfClass) {
   uint8_t identifier[sizeof(MDGUID)];
   const char expected_identifier_string[] =
       "80808080-8080-0000-0000-008080808080";
   char identifier_string[sizeof(expected_identifier_string)];
+  const size_t kTextSectionSize = 128;
+
+  ELF elf32(EM_386, ELFCLASS32, kLittleEndian);
+  Section text32(kLittleEndian);
+  for (size_t i = 0; i < kTextSectionSize; ++i) {
+    text32.D8(i * 3);
+  }
+  elf32.AddSection(".text", text32, SHT_PROGBITS);
+  elf32.Finish();
+  GetElfContents(elf32);
+
+  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(elfdata, identifier));
 
-  ElfishElf<ElfClass32> elf32;
-  ElfishElf<ElfClass32>::Populate(&elf32);
-  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(&elf32, identifier));
   FileID::ConvertIdentifierToString(identifier, identifier_string,
                                     sizeof(identifier_string));
   EXPECT_STREQ(expected_identifier_string, identifier_string);
@@ -134,9 +118,65 @@ TEST(FileIDTest, ElfClass) {
   memset(identifier, 0, sizeof(identifier));
   memset(identifier_string, 0, sizeof(identifier_string));
 
-  ElfishElf<ElfClass64> elf64;
-  ElfishElf<ElfClass64>::Populate(&elf64);
-  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(&elf64, identifier));
+  ELF elf64(EM_X86_64, ELFCLASS64, kLittleEndian);
+  Section text64(kLittleEndian);
+  for (size_t i = 0; i < kTextSectionSize; ++i) {
+    text64.D8(i * 3);
+  }
+  elf64.AddSection(".text", text64, SHT_PROGBITS);
+  elf64.Finish();
+  GetElfContents(elf64);
+
+  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(elfdata, identifier));
+
+  FileID::ConvertIdentifierToString(identifier, identifier_string,
+                                    sizeof(identifier_string));
+  EXPECT_STREQ(expected_identifier_string, identifier_string);
+}
+
+TEST_F(FileIDTest, BuildID) {
+  const uint8_t kExpectedIdentifier[sizeof(MDGUID)] =
+    {0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,
+     0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F};
+  char expected_identifier_string[] =
+    "00000000-0000-0000-0000-000000000000";
+  FileID::ConvertIdentifierToString(kExpectedIdentifier,
+                                    expected_identifier_string,
+                                    sizeof(expected_identifier_string));
+
+  uint8_t identifier[sizeof(MDGUID)];
+  char identifier_string[sizeof(expected_identifier_string)];
+
+  ELF elf32(EM_386, ELFCLASS32, kLittleEndian);
+  Section text(kLittleEndian);
+  text.Append(4096, 0);
+  elf32.AddSection(".text", text, SHT_PROGBITS);
+  BuildIDNote::AppendSection(elf32,
+                             kExpectedIdentifier,
+                             sizeof(kExpectedIdentifier));
+  elf32.Finish();
+  GetElfContents(elf32);
+
+  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(elfdata, identifier));
+
+  FileID::ConvertIdentifierToString(identifier, identifier_string,
+                                    sizeof(identifier_string));
+  EXPECT_STREQ(expected_identifier_string, identifier_string);
+
+  memset(identifier, 0, sizeof(identifier));
+  memset(identifier_string, 0, sizeof(identifier_string));
+
+  ELF elf64(EM_X86_64, ELFCLASS64, kLittleEndian);
+  // Re-use empty text section from previous test
+  elf64.AddSection(".text", text, SHT_PROGBITS);
+  BuildIDNote::AppendSection(elf64,
+                             kExpectedIdentifier,
+                             sizeof(kExpectedIdentifier));
+  elf64.Finish();
+  GetElfContents(elf64);
+
+  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(elfdata, identifier));
+
   FileID::ConvertIdentifierToString(identifier, identifier_string,
                                     sizeof(identifier_string));
   EXPECT_STREQ(expected_identifier_string, identifier_string);
diff --git a/src/common/linux/synth_elf.cc b/src/common/linux/synth_elf.cc
index 8f8aad32..3e10b691 100644
--- a/src/common/linux/synth_elf.cc
+++ b/src/common/linux/synth_elf.cc
@@ -3,10 +3,15 @@
 #include <assert.h>
 #include <elf.h>
 #include <stdio.h>
+#include <string.h>
 
 namespace google_breakpad {
 namespace synth_elf {
 
+#ifndef NT_GNU_BUILD_ID
+#define NT_GNU_BUILD_ID 3
+#endif
+
 ELF::ELF(uint16_t machine,
          uint8_t file_class,
          Endianness endianness)
@@ -169,6 +174,31 @@ void SymbolTable::AddSymbol(const string& name, uint64_t value,
   D64(size);
 }
 
+BuildIDNote::BuildIDNote(const uint8_t* id_bytes,
+                         size_t id_size,
+                         Endianness endianness) : Section(endianness) {
+  const char kNoteName[] = "GNU";
+  // Elf32_Nhdr and Elf64_Nhdr are exactly the same.
+  Elf32_Nhdr note_header;
+  memset(&note_header, 0, sizeof(note_header));
+  note_header.n_namesz = sizeof(kNoteName);
+  note_header.n_descsz = id_size;
+  note_header.n_type = NT_GNU_BUILD_ID;
+
+  Append(reinterpret_cast<const uint8_t*>(&note_header),
+         sizeof(note_header));
+  AppendCString(kNoteName);
+  Append(id_bytes, id_size);
+}
+
+// static
+void BuildIDNote::AppendSection(ELF& elf,
+                                const uint8_t* id_bytes,
+                                size_t id_size) {
+  const char kBuildIDSectionName[] = ".note.gnu.build-id";
+  BuildIDNote note(id_bytes, id_size, elf.endianness());
+  elf.AddSection(kBuildIDSectionName, note, SHT_NOTE);
+}
+
 }  // namespace synth_elf
 }  // namespace google_breakpad
-
diff --git a/src/common/linux/synth_elf.h b/src/common/linux/synth_elf.h
index 215fd997..90bd8341 100644
--- a/src/common/linux/synth_elf.h
+++ b/src/common/linux/synth_elf.h
@@ -149,6 +149,15 @@ class SymbolTable : public Section {
   StringTable& table_;
 };
 
+// A class to build GNU Build ID note sections
+class BuildIDNote : public Section {
+public:
+  BuildIDNote(const uint8_t* id_bytes, size_t id_size, Endianness endianness);
+
+  // Append a new Build ID note section to |elf|.
+  static void AppendSection(ELF& elf, const uint8_t* id_bytes, size_t id_size);
+};
+
 }  // namespace synth_elf
 }  // namespace google_breakpad