From dbd454dbe47e584571388fc3533193416bdce67f Mon Sep 17 00:00:00 2001
From: Sterling Augustine <saugustine@google.com>
Date: Fri, 13 Dec 2019 16:54:50 -0800
Subject: Add dwarf5 compilation-unit header handling.

Fix a small typo in DW_LANG__Python.

Change-Id: I1fe54e501a5e8da5057ecc9ac00c7e7259a9bb3f
Reviewed-on: https://chromium-review.googlesource.com/c/breakpad/breakpad/+/1967975
Reviewed-by: Mark Mentovai <mark@chromium.org>
---
 src/common/dwarf/dwarf2reader.cc | 84 +++++++++++++++++++++++++++++++++-------
 1 file changed, 70 insertions(+), 14 deletions(-)

(limited to 'src/common/dwarf/dwarf2reader.cc')

diff --git a/src/common/dwarf/dwarf2reader.cc b/src/common/dwarf/dwarf2reader.cc
index 3e6a3e89..7b8e043d 100644
--- a/src/common/dwarf/dwarf2reader.cc
+++ b/src/common/dwarf/dwarf2reader.cc
@@ -243,11 +243,51 @@ const uint8_t *CompilationUnit::SkipAttribute(const uint8_t *start,
   return NULL;
 }
 
-// Read a DWARF2/3 header.
-// The header is variable length in DWARF3 (and DWARF2 as extended by
+// Read the abbreviation offset from a compilation unit header.
+int CompilationUnit::ReadAbbrevOffset(const uint8_t *headerptr) {
+  assert(headerptr + reader_->OffsetSize() < buffer_ + buffer_length_);
+  header_.abbrev_offset = reader_->ReadOffset(headerptr);
+  return reader_->OffsetSize();
+}
+
+// Read the address size from a compilation unit header.
+int CompilationUnit::ReadAddressSize(const uint8_t *headerptr) {
+  // Compare against less than or equal because this may be the last
+  // section in the file.
+  assert(headerptr + 1 <= buffer_ + buffer_length_);
+  header_.address_size = reader_->ReadOneByte(headerptr);
+  reader_->SetAddressSize(header_.address_size);
+  return 1;
+}
+
+// Read the DWO id from a split or skeleton compilation unit header.
+int CompilationUnit::ReadDwoId(const uint8_t *headerptr) {
+  assert(headerptr + 8 <= buffer_ + buffer_length_);
+  dwo_id_ = reader_->ReadEightBytes(headerptr);
+  return 8;
+}
+
+// Read the type signature from a type or split type compilation unit header.
+int CompilationUnit::ReadTypeSignature(const uint8_t *headerptr) {
+  assert(headerptr + 8 <= buffer_ + buffer_length_);
+  type_signature_ = reader_->ReadEightBytes(headerptr);
+  return 8;
+}
+
+// Read the DWO id from a split or skeleton compilation unit header.
+int CompilationUnit::ReadTypeOffset(const uint8_t *headerptr) {
+  assert(headerptr + reader_->OffsetSize() < buffer_ + buffer_length_);
+  type_offset_ = reader_->ReadOffset(headerptr);
+  return reader_->OffsetSize();
+}
+
+
+// Read a DWARF header.
+// The header is variable length in DWARF3 and DWARF4 (and DWARF2 as extended by
 // most compilers), and consists of an length field, a version number,
 // the offset in the .debug_abbrev section for our abbrevs, and an
-// address size.
+// address size. DWARF5 adds a unit_type to distinguish between
+// partial-, full-, skeleton-, split-, and type- compilation units.
 void CompilationUnit::ReadHeader() {
   const uint8_t *headerptr = buffer_;
   size_t initial_length_size;
@@ -262,17 +302,33 @@ void CompilationUnit::ReadHeader() {
   header_.version = reader_->ReadTwoBytes(headerptr);
   headerptr += 2;
 
-  assert(headerptr + reader_->OffsetSize() < buffer_ + buffer_length_);
-  header_.abbrev_offset = reader_->ReadOffset(headerptr);
-  headerptr += reader_->OffsetSize();
-
-  // Compare against less than or equal because this may be the last
-  // section in the file.
-  assert(headerptr + 1 <= buffer_ + buffer_length_);
-  header_.address_size = reader_->ReadOneByte(headerptr);
-  reader_->SetAddressSize(header_.address_size);
-  headerptr += 1;
-
+  if (header_.version <= 4) {
+    // Older versions of dwarf have a relatively simple structure.
+    headerptr += ReadAbbrevOffset(headerptr);
+    headerptr += ReadAddressSize(headerptr);
+  } else {
+    // DWARF5 adds a unit_type field, and various fields based on unit_type.
+    assert(headerptr + 1 < buffer_ + buffer_length_);
+    int unit_type = reader_->ReadOneByte(headerptr);
+    headerptr += 1;
+    headerptr += ReadAddressSize(headerptr);
+    headerptr += ReadAbbrevOffset(headerptr);
+    switch (unit_type) {
+      case DW_UT_compile:
+      case DW_UT_partial:
+        // nothing else to read
+        break;
+      case DW_UT_skeleton:
+      case DW_UT_split_compile:
+        headerptr += ReadDwoId(headerptr);;
+        break;
+      case DW_UT_type:
+      case DW_UT_split_type:
+        headerptr += ReadTypeSignature(headerptr);;
+        headerptr += ReadTypeOffset(headerptr);;
+        break;
+    }
+  }
   after_header_ = headerptr;
 
   // This check ensures that we don't have to do checking during the
-- 
cgit v1.2.1