Finalizer scanning for symbols

finalize/CMakeLists.txt

@@ -1,5 +1,5 @@
 find_package(LIEF REQUIRED)
 
-add_executable(spslr_finalize finalize.cpp accumulation.cpp)
+add_executable(spslr_finalize finalize.cpp accumulation.cpp symbol_collection.cpp)
 target_link_libraries(spslr_finalize PRIVATE LIEF::LIEF)
 set_target_properties(spslr_finalize PROPERTIES CXX_STANDARD 20 CXX_STANDARD_REQUIRED YES CXX_EXTENSIONS NO)

finalize/finalize.cpp

@@ -2,7 +2,10 @@
 #include <string>
 #include <getopt.h>
 
+#include <LIEF/LIEF.hpp>
+
 #include "accumulation.h"
+#include "symbol_collection.h"
 
 /*
 Notes:
@@ -75,13 +78,21 @@ int main(int argc, char** argv) {
 	std::cout << "Gathered a total of " << targets.size() << " distinct targets from "
 		<< units.size() << " compilation units!" << std::endl;
 
+	std::unique_ptr<LIEF::ELF::Binary> bin = LIEF::ELF::Parser::parse(bin_file);
+	if (!bin) {
+		std::cerr << "Failed to parse binary \"" << bin_file << "\"!" << std::endl;
+		return 1;
+	}
+
+	if (!collect_symbols(bin)) {
+		std::cerr << "Failed to collect symbols from \"" << bin_file << "\"!" << std::endl;
+		return 1;
+	}
+
 	// TODO
 	/*
-	1. Loop over all symbols of the binary
-		-> associate blocks via CU uid symbol
-		-> find __spslr_program symbol (spslr vaddr pivot)
-	2. Find virtual address and file address for all pins
-	3. Emit patcher program into final executable and set __spslr_program
+	1. Find virtual address and file address for all pins
+	2. Emit patcher program into final executable and set __spslr_program
 		-> make all patcher addresses relative to &__spslr_program
 		-> dpatch entries ordered by level (high levels/nest depths first)
 	*/

finalize/symbol_collection.cpp

@@ -0,0 +1,93 @@
+#include "symbol_collection.h"
+
+#include <iostream>
+
+#define SPSLR_PINPOINT_CU_UID_LABEL "__spslr_cu_" /* suffixed with "<uid>" */
+
+/*
+struct GlobalSymtab {
+    std::unordered_map<std::string, VADDR> symbols;
+};
+
+struct LocalSymtab {
+    std::string cu;
+    std::unordered_map<std::string, VADDR> symbols;
+};
+*/
+
+using namespace LIEF::ELF;
+
+GlobalSymtab global_syms;
+std::unordered_map<std::string, LocalSymtab> local_syms;
+
+bool collect_symbols(const std::unique_ptr<LIEF::ELF::Binary>& bin) {
+	if (!bin)
+		return false;
+
+	bool cu_uid_found = false;
+	LocalSymtab tmp_local_syms;
+
+	auto finish_cu_symbols = [&]() -> bool {
+		if (cu_uid_found) {
+			if (local_syms.contains(tmp_local_syms.cu)) {
+				std::cerr << "Duplicate CU UID -> " << tmp_local_syms.cu << std::endl;
+				return false;
+			}
+
+			local_syms.emplace(tmp_local_syms.cu, std::move(tmp_local_syms));
+		}
+
+		tmp_local_syms.cu.clear();
+		tmp_local_syms.symbols.clear();
+		cu_uid_found = false;
+		return true;
+	};
+
+	for (const Symbol& sym : bin->symbols()) {
+		if (sym.type() == Symbol::TYPE::FILE) {
+			if (!finish_cu_symbols())
+				return false;
+		}
+
+		if (sym.name().empty())
+			continue;
+
+		if (sym.type() != Symbol::TYPE::OBJECT && sym.type() != Symbol::TYPE::NOTYPE)
+			continue;
+
+		VADDR vaddr = sym.value();
+
+		if (sym.binding() == Symbol::BINDING::GLOBAL) {
+			if (global_syms.symbols.contains(sym.name())) {
+				std::cerr << "Duplicate global symbol \"" << sym.name() << "\"!" << std::endl;
+				return false;
+			}
+
+			global_syms.symbols.emplace(sym.name(), vaddr);
+			continue;
+		}
+
+		if (sym.binding() != Symbol::BINDING::LOCAL)
+			continue;
+
+		if (sym.name().starts_with(SPSLR_PINPOINT_CU_UID_LABEL)) {
+			tmp_local_syms.cu = sym.name().substr(std::strlen(SPSLR_PINPOINT_CU_UID_LABEL));
+			cu_uid_found = true;
+			continue;
+		}
+
+		if (tmp_local_syms.symbols.contains(sym.name())) {
+			std::cerr << "Duplicate local symbol \"" << sym.name() << "\"!" << std::endl;
+			return false;
+		}
+
+		tmp_local_syms.symbols.emplace(sym.name(), vaddr);
+		continue;
+	}
+
+	if (!finish_cu_symbols())
+		return false;
+
+	return true;
+}
+

finalize/symbol_collection.h

@@ -0,0 +1,22 @@
+#pragma once
+#include <LIEF/LIEF.hpp>
+#include <string>
+#include <unordered_map>
+#include <cstdint>
+
+using VADDR = uint64_t;
+
+struct GlobalSymtab {
+	std::unordered_map<std::string, VADDR> symbols;
+};
+
+struct LocalSymtab {
+	std::string cu;
+	std::unordered_map<std::string, VADDR> symbols;
+};
+
+extern GlobalSymtab global_syms;
+extern std::unordered_map<std::string, LocalSymtab> local_syms;
+
+bool collect_symbols(const std::unique_ptr<LIEF::ELF::Binary>& bin);
+

selfpatch/src/selfpatch.c

@@ -13,7 +13,7 @@ With ASLR, there are 2 options to make it function correctly:
 	1. Make sure __spslr_program is relocated with program image shift (preferred)
 	2. __spslr_program = spslr_ptr_absolute((uint64_t)__spslr_program) and postprocessor inserts relative value (kinda weird)
 */
-static const uint8_t* __spslr_program = NULL;
+const uint8_t* __spslr_program = NULL;
 
 static void* spslr_ptr_absolute(uint64_t relative) {
 	// To allow ASLR, all patcher addresses are relative to &__spslr_program