Fix parsing of COFF debug sections

crosire · Dec 11, 2019 · 168a4cf · 168a4cf
1 parent cb5f139
commit 168a4cf
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 25 deletions.
diff --git a/source/blink.cpp b/source/blink.cpp
@@ -340,30 +340,42 @@ std::string blink::application::build_compile_command_line(const std::filesystem
 				SetFilePointer(file, section->PointerToRawData, nullptr, FILE_BEGIN);
 				ReadFile(file, debug_data.data(), section->SizeOfRawData, &read, nullptr);
 
-				// Replace command-line with the one in the debug section
-				cmdline.clear();
-
 				// Skip header in front of CodeView records (version, ...)
 				stream_reader stream(std::move(debug_data));
-				stream.skip(4 * 3);
+				stream.skip(4); // Skip 32-bit signature (this should be CV_SIGNATURE_C13, aka 4)
 
-				parse_code_view_records(stream, [&](uint16_t tag) {
-					if (tag != 0x113d) // S_ENVBLOCK
-						return; // Skip all records that are not about the compiler environment
-					stream.skip(1);
-					while (stream.tell() < stream.size() && *stream.data() != '\0')
+				while (stream.tell() < stream.size() && cmdline.empty())
+				{
+					// CV_DebugSSubsectionHeader_t
+					const auto subsection_type = stream.read<uint32_t>();
+					const auto subsection_length = stream.read<uint32_t>();
+					if (subsection_type != 0xf1 /*DEBUG_S_SYMBOLS*/)
 					{
-						const auto key = stream.read_string();
-						const std::string value(stream.read_string());
-
-						if (key == "cwd")
-							cmdline += "cd /D \"" + value + "\"\n";
-						else if (key == "cl") // Add compiler directories to path, so that 'mspdbcore.dll' is found
-							cmdline += "set PATH=%PATH%;" + value + "\\..\\..\\x86;" + value + "\\..\\..\\x64\n\"" + value + "\" ";
-						else if (key == "cmd")
-							cmdline += value;
+						stream.skip(subsection_length);
+						stream.align(4);
+						continue;
 					}
-				});
+
+					parse_code_view_records(stream, subsection_length, [&](uint16_t tag) {
+						if (tag != 0x113d) // S_ENVBLOCK
+							return; // Skip all records that are not about the compiler environment
+						stream.skip(1);
+						while (stream.tell() < stream.size() && *stream.data() != '\0')
+						{
+							const auto key = stream.read_string();
+							const std::string value(stream.read_string());
+
+							if (key == "cwd")
+								cmdline += "cd /D \"" + value + "\"\n";
+							else if (key == "cl") // Add compiler directories to path, so that 'mspdbcore.dll' is found
+								cmdline += "set PATH=%PATH%;" + value + "\\..\\..\\x86;" + value + "\\..\\..\\x64\n\"" + value + "\" ";
+							else if (key == "cmd")
+								cmdline += value;
+						}
+					});
+
+					stream.align(4); // Subsection headers are 4-byte aligned
+				}
 			}
 		}
 	}

diff --git a/source/pdb_reader.cpp b/source/pdb_reader.cpp
@@ -197,7 +197,7 @@ void blink::pdb_reader::read_symbol_table(uint8_t *image_base, std::unordered_ma
 	// Read symbol table records in CodeView format
 	stream = msf_reader::stream(header.symbol_record_stream);
 
-	parse_code_view_records(stream, [&](uint16_t tag) {
+	parse_code_view_records(stream, stream.size(), [&](uint16_t tag) {
 		if (tag != 0x110E) // S_PUB32
 			return; // Skip all records that are not about public symbols
 
@@ -242,9 +242,9 @@ void blink::pdb_reader::read_object_files(std::vector<std::filesystem::path> &ob
 
 				// Look up current working directory in symbol stream https://llvm.org/docs/PDB/ModiStream.html
 				stream_reader stream(msf_reader::stream(info.symbol_stream));
-				stream.skip(4); // Skip 32-bit signature
-				
-				parse_code_view_records(stream, [&](uint16_t tag) {
+				stream.skip(4); // Skip 32-bit signature (this should be CV_SIGNATURE_C13, aka 4)
+
+				parse_code_view_records(stream, info.symbol_byte_size - 4, [&](uint16_t tag) {
 					if (tag == 0x113d) // S_ENVBLOCK
 					{
 						stream.skip(1);

diff --git a/source/pdb_reader.hpp b/source/pdb_reader.hpp
@@ -184,10 +184,12 @@ namespace blink
 	/// Helper function that parses a CodeView stream and calls a callback function for every record in it
 	/// </summary>
 	template <typename L>
-	void parse_code_view_records(stream_reader &stream, L callback, size_t alignment = 1)
+	void parse_code_view_records(stream_reader &stream, size_t length, L callback, size_t alignment = 1)
 	{
+		const size_t end = stream.tell() + length;
+
 		// A list of records in CodeView format
-		while (stream.tell() < stream.size())
+		while (stream.tell() < end)
 		{
 			// Each records starts with 2 bytes containing the size of the record after this element
 			const auto size = stream.read<uint16_t>();