Try to find a build-id through PT_NOTE program headers

A=Mike Hommey <mh@glandium.org>
R=ted at https://breakpad.appspot.com/544003/

git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@1143 4c0a9323-5329-0410-9bdc-e9ce6186880e

4 files changed, 138 insertions, 10 deletions

diff --git a/src/common/linux/elfutils.cc b/src/common/linux/elfutils.cc
index ee2f4ac0..1fd504d9 100644
--- a/src/common/linux/elfutils.cc
+++ b/src/common/linux/elfutils.cc
@@ -75,6 +75,35 @@ void FindElfClassSection(const char *elf_base,
   }
 }
 
+template<typename ElfClass>
+void FindElfClassSegment(const char *elf_base,
+                         typename ElfClass::Word segment_type,
+                         const void **segment_start,
+                         int *segment_size) {
+  typedef typename ElfClass::Ehdr Ehdr;
+  typedef typename ElfClass::Phdr Phdr;
+
+  assert(elf_base);
+  assert(segment_start);
+  assert(segment_size);
+
+  assert(my_strncmp(elf_base, ELFMAG, SELFMAG) == 0);
+
+  const Ehdr* elf_header = reinterpret_cast<const Ehdr*>(elf_base);
+  assert(elf_header->e_ident[EI_CLASS] == ElfClass::kClass);
+
+  const Phdr* phdrs =
+    GetOffset<ElfClass,Phdr>(elf_header, elf_header->e_phoff);
+
+  for (int i = 0; i < elf_header->e_phnum; ++i) {
+    if (phdrs[i].p_type == segment_type) {
+      *segment_start = elf_base + phdrs[i].p_offset;
+      *segment_size = phdrs[i].p_filesz;
+      return;
+    }
+  }
+}
+
 }  // namespace
 
 bool IsValidElf(const void* elf_base) {
@@ -126,4 +155,40 @@ bool FindElfSection(const void *elf_mapped_base,
   return false;
 }
 
+bool FindElfSegment(const void *elf_mapped_base,
+                    uint32_t segment_type,
+                    const void **segment_start,
+                    int *segment_size,
+                    int *elfclass) {
+  assert(elf_mapped_base);
+  assert(segment_start);
+  assert(segment_size);
+
+  *segment_start = NULL;
+  *segment_size = 0;
+
+  if (!IsValidElf(elf_mapped_base))
+    return false;
+
+  int cls = ElfClass(elf_mapped_base);
+  if (elfclass) {
+    *elfclass = cls;
+  }
+
+  const char* elf_base =
+    static_cast<const char*>(elf_mapped_base);
+
+  if (cls == ELFCLASS32) {
+    FindElfClassSegment<ElfClass32>(elf_base, segment_type,
+                                    segment_start, segment_size);
+    return *segment_start != NULL;
+  } else if (cls == ELFCLASS64) {
+    FindElfClassSegment<ElfClass64>(elf_base, segment_type,
+                                    segment_start, segment_size);
+    return *segment_start != NULL;
+  }
+
+  return false;
+}
+
 }  // namespace google_breakpad
diff --git a/src/common/linux/elfutils.h b/src/common/linux/elfutils.h
index 748da988..fe12e251 100644
--- a/src/common/linux/elfutils.h
+++ b/src/common/linux/elfutils.h
@@ -93,6 +93,17 @@ FindElfSectionByName(const char* name,
                      const char* names_end,
                      int nsection);
 
+// Attempt to find the first segment of type |segment_type| in the ELF
+// binary data at |elf_mapped_base|. On success, returns true and sets
+// |*segment_start| to point to the start of the segment data, and
+// and |*segment_size| to the size of the segment's data. If |elfclass|
+// is not NULL, set |*elfclass| to the ELF file class.
+bool FindElfSegment(const void *elf_mapped_base,
+                    uint32_t segment_type,
+                    const void **segment_start,
+                    int *segment_size,
+                    int *elfclass);
+
 // Convert an offset from an Elf header into a pointer to the mapped
 // address in the current process. Takes an extra template parameter
 // to specify the return type to avoid having to dynamic_cast the
diff --git a/src/common/linux/file_id.cc b/src/common/linux/file_id.cc
index 4e380b06..ca5a3e1f 100644
--- a/src/common/linux/file_id.cc
+++ b/src/common/linux/file_id.cc
@@ -55,22 +55,34 @@ FileID::FileID(const char* path) {
   strncpy(path_, path, sizeof(path_));
 }
 
-// These six functions are also used inside the crashed process, so be safe
+// ELF note name and desc are 32-bits word padded.
+#define NOTE_PADDING(a) ((a + 3) & ~3)
+
+// These 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 bool ElfClassBuildIDNoteIdentifier(const void *section,
+static bool ElfClassBuildIDNoteIdentifier(const void *section, int length,
                                           uint8_t identifier[kMDGUIDSize]) {
   typedef typename ElfClass::Nhdr Nhdr;
 
+  const void* section_end = reinterpret_cast<const char*>(section) + length;
   const Nhdr* note_header = reinterpret_cast<const Nhdr*>(section);
-  if (note_header->n_type != NT_GNU_BUILD_ID ||
+  while (reinterpret_cast<const void *>(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) +
+                  NOTE_PADDING(note_header->n_namesz) +
+                  NOTE_PADDING(note_header->n_descsz));
+  }
+  if (reinterpret_cast<const void *>(note_header) >= section_end ||
       note_header->n_descsz == 0) {
     return false;
   }
 
-  const char* build_id = reinterpret_cast<const char*>(section) +
-    sizeof(Nhdr) + note_header->n_namesz;
+  const char* build_id = reinterpret_cast<const char*>(note_header) +
+    sizeof(Nhdr) + NOTE_PADDING(note_header->n_namesz);
   // Copy as many bits of the build ID as will fit
   // into the GUID space.
   my_memset(identifier, 0, kMDGUIDSize);
@@ -86,16 +98,21 @@ 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) {
+  if ((!FindElfSegment(elf_mapped_base, PT_NOTE,
+                       (const void**)&note_section, &note_size, &elfclass) ||
+      note_size == 0)  &&
+      (!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);
+    return ElfClassBuildIDNoteIdentifier<ElfClass32>(note_section, note_size,
+                                                     identifier);
   } else if (elfclass == ELFCLASS64) {
-    return ElfClassBuildIDNoteIdentifier<ElfClass64>(note_section, identifier);
+    return ElfClassBuildIDNoteIdentifier<ElfClass64>(note_section, note_size,
+                                                     identifier);
   }
 
   return false;
diff --git a/src/common/linux/file_id_unittest.cc b/src/common/linux/file_id_unittest.cc
index e2e13747..bcd030dc 100644
--- a/src/common/linux/file_id_unittest.cc
+++ b/src/common/linux/file_id_unittest.cc
@@ -175,6 +175,41 @@ TYPED_TEST(FileIDTest, BuildID) {
   EXPECT_STREQ(expected_identifier_string, identifier_string);
 }
 
+TYPED_TEST(FileIDTest, BuildIDPH) {
+  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 elf(EM_386, TypeParam::kClass, kLittleEndian);
+  Section text(kLittleEndian);
+  text.Append(4096, 0);
+  elf.AddSection(".text", text, SHT_PROGBITS);
+  Notes notes(kLittleEndian);
+  notes.AddNote(0, "Linux",
+                reinterpret_cast<const uint8_t *>("\0x42\0x02\0\0"), 4);
+  notes.AddNote(NT_GNU_BUILD_ID, "GNU", kExpectedIdentifier,
+                sizeof(kExpectedIdentifier));
+  int note_idx = elf.AddSection(".note", notes, SHT_NOTE);
+  elf.AddSegment(note_idx, note_idx, PT_NOTE);
+  elf.Finish();
+  this->GetElfContents(elf);
+
+  EXPECT_TRUE(FileID::ElfFileIdentifierFromMappedFile(this->elfdata,
+                                                      identifier));
+
+  FileID::ConvertIdentifierToString(identifier, identifier_string,
+                                    sizeof(identifier_string));
+  EXPECT_STREQ(expected_identifier_string, identifier_string);
+}
+
 // Test to make sure two files with different text sections produce
 // different hashes when not using a build id.
 TYPED_TEST(FileIDTest, UniqueHashes) {