From d33896ecd246000dc43024d72d2e35af996100f3 Mon Sep 17 00:00:00 2001
From: York Jasper Niebuhr <yjnworkstation@gmail.com>
Date: Mon, 27 Oct 2025 18:27:59 +0100
Subject: [PATCH] Finalizer spslr file parsing

---
 finalize/CMakeLists.txt           |   3 +-
 finalize/accumulation.cpp         | 232 ++++++++++++++++++++++++++++++
 finalize/accumulation.h           |  48 +++++++
 finalize/finalize.cpp             |  63 +++-----
 pinpoint/final/on_finish_unit.cpp |   5 +-
 5 files changed, 307 insertions(+), 44 deletions(-)
 create mode 100644 finalize/accumulation.cpp
 create mode 100644 finalize/accumulation.h

diff --git a/finalize/CMakeLists.txt b/finalize/CMakeLists.txt
index b3d0e7d..7bb8957 100644
--- a/finalize/CMakeLists.txt
+++ b/finalize/CMakeLists.txt
@@ -1,4 +1,5 @@
 find_package(LIEF REQUIRED)
 
-add_executable(spslr_finalize finalize.cpp)
+add_executable(spslr_finalize finalize.cpp accumulation.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)
diff --git a/finalize/accumulation.cpp b/finalize/accumulation.cpp
new file mode 100644
index 0000000..51ca6e1
--- /dev/null
+++ b/finalize/accumulation.cpp
@@ -0,0 +1,232 @@
+#include "accumulation.h"
+
+#include <iostream>
+#include <filesystem>
+#include <fstream>
+
+namespace fs = std::filesystem;
+
+static std::size_t next_global_target_uid = 0;
+
+std::unordered_map<std::size_t, TARGET> targets;
+std::unordered_map<std::string, DPIN> global_dpins;
+std::unordered_map<std::string, CU> units;
+
+static bool global_target_field_cmp(const TARGET& a, const TARGET& b) {
+	// TODO
+	return true;
+}
+
+static bool global_target_cmp(const TARGET& a, const TARGET& b) {
+	// Note -> Later, use hash of type (generated by pinpoint plugin)
+
+	if (a.name != b.name)
+		return false;
+
+	if (a.size != b.size || !global_target_field_cmp(a, b)) {
+		std::cerr << "WARNING: Got different definitions of \"" << a.name << "\" -> detached randomization" << std::endl;
+		return false;
+	}
+
+	return true;
+}
+
+static std::size_t accumulate_global_target(TARGET&& target) {
+	for (const auto& [guid, gtarget] : targets) {
+		if (global_target_cmp(gtarget, target))
+			return guid;
+	}
+
+	std::size_t guid = next_global_target_uid++;
+	targets.emplace(guid, std::move(target));
+	return guid;
+}
+
+/*
+<sourcefile>.spslr:
+
+SPSLR <CU filename> <CU uid symbol>
+target <name> <local uid> <size> <field count>
+f <offset> <size> <alignment> <flags>
+f <offset> <size> <alignment> <flags>
+...
+ipin <label> <target uid> <field offset>
+ipin <label> <target uid> <field offset>
+ipin <label> <target uid> <field offset>
+...
+dpin <local/global> <symbol> <offset> <level> <target uid>
+dpin <local/global> <symbol> <offset> <level> <target uid>
+...
+*/
+static bool accumulate_file(const fs::path& path) {
+	std::ifstream infile(path);
+	if (!infile)
+		return false;
+
+	std::string hdr_line;
+	if (!std::getline(infile, hdr_line))
+		return false;
+
+	std::istringstream hdr_iss(hdr_line);
+
+	std::string hdr_magic, hdr_cu_file, hdr_cu_uid;
+
+	if (!(hdr_iss >> hdr_magic) || hdr_magic != "SPSLR")
+		return false;
+
+	if (!(hdr_iss >> hdr_cu_file) || !(hdr_iss >> hdr_cu_uid))
+		return false;
+
+	std::cout << "Parsing " << path << std::endl;
+	std::cout << "    Source file -> " << hdr_cu_file << std::endl;
+	std::cout << "    Unit UID    -> " << hdr_cu_uid << std::endl;
+
+	if (units.contains(hdr_cu_uid)) {
+		std::cerr << "Duplicate unit UID -> " << hdr_cu_uid << std::endl;
+		return false;
+	}
+
+	units.emplace(hdr_cu_uid, CU{});
+	CU& cu = units.at(hdr_cu_uid);
+
+	std::string line;
+	while (std::getline(infile, line)) {
+		if (line.empty())
+			continue;
+
+		std::istringstream iss(line);
+		std::string type;
+
+		if (!(iss >> type))
+			return false;
+
+		if (type == "target") {
+			TARGET target;
+
+			std::size_t local_uid, field_count;
+
+			if (!(iss >> target.name) || !(iss >> local_uid) || !(iss >> target.size) || !(iss >> field_count)) {
+				std::cerr << "Invalid target line" << std::endl;
+				return false;
+			}
+
+			if (cu.local_targets.contains(local_uid)) {
+				std::cerr << "Duplicate local target uid -> " << local_uid << std::endl;
+				return false;
+			}
+
+			for (std::size_t i = 0; i < field_count; i++) {
+				std::string fline;
+				if (!std::getline(infile, fline)) {
+					std::cerr << "Missing field declaration line!" << std::endl;
+					return false;
+				}
+
+				std::istringstream fiss(fline);
+
+				std::string ftype;
+				if (!(fiss >> ftype) || ftype != "f") {
+					std::cerr << "Missing field declaration line, got \"" << ftype << "\" instead!" << std::endl;
+					return false;
+				}
+
+				FIELD field;
+				if (!(fiss >> field.offset) || !(fiss >> field.size) || !(fiss >> field.alignment) || !(fiss >> field.flags)) {
+					std::cerr << "Failed to parse field declaration line!" << std::endl;
+					return false;
+				}
+
+				// Note -> could do sanity checks here
+
+				target.fields.emplace(field.offset, field);
+			}
+
+			std::size_t global_target_uid = accumulate_global_target(std::move(target));
+			cu.local_targets.emplace(local_uid, global_target_uid);
+			continue;
+		} else if (type == "ipin") {
+			IPIN ipin;
+			if (!(iss >> ipin.symbol) || !(iss >> ipin.local_target) || !(iss >> ipin.field_offset)) {
+				std::cerr << "Failed to parse ipin declaration!" << std::endl;
+				return false;
+			}
+
+			if (cu.ipins.contains(ipin.symbol)) {
+				std::cerr << "Duplicate ipin for symbol \"" << ipin.symbol << "\"!" << std::endl;
+				return false;
+			}
+
+			cu.ipins.emplace(ipin.symbol, ipin);
+			continue;
+		} else if (type == "dpin") {
+			std::string symbol;
+			std::string scope;
+			DPIN::COMPONENT comp;
+			std::size_t local_target;
+
+			if (!(iss >> scope) || !(iss >> symbol) || !(iss >> comp.offset) ||
+					!(iss >> comp.level) || !(iss >> local_target)) {
+				std::cerr << "Failed to parse dpin declaration!" << std::endl;
+				return false;
+			}
+
+			if (scope == "l" || scope == "L") {
+				if (!cu.dpins.contains(symbol))
+					cu.dpins.emplace(symbol, DPIN{ .symbol=symbol });
+
+				DPIN& dpin = cu.dpins.at(symbol);
+
+				if (!cu.local_targets.contains(local_target)) {
+					std::cerr << "Local datapin references target that has not yet been parsed!" << std::endl;
+					return false;
+				}
+
+				comp.target = local_target;
+				dpin.components.push_back(comp);
+			} else if (scope == "g" || scope == "G") {
+				if (!global_dpins.contains(symbol))
+					global_dpins.emplace(symbol, DPIN{ .symbol=symbol });
+
+				DPIN& dpin = global_dpins.at(symbol);
+
+				if (!cu.local_targets.contains(local_target)) {
+					std::cerr << "Global datapin references target that has not yet been parsed!" << std::endl;
+					return false;
+				}
+
+				comp.target = cu.local_targets.at(local_target);
+				dpin.components.push_back(comp);
+			} else {
+				std::cerr << "Unexpected scope of dpin (\"" << scope << "\"), expected \"l\" or \"g\"!" << std::endl;
+				return false;
+			}
+
+			continue;
+		} else {
+			std::cerr << "Invalid spslr file line beginning: \"" << type << "\"" << std::endl;
+			return false;
+		}
+	}
+
+	return true;
+}
+
+bool accumulate(const std::string& spslr_dir) {
+	fs::path spslr_root { spslr_dir };
+
+	if (!fs::exists(spslr_root) || !fs::is_directory(spslr_root))
+		return false;
+
+	for (const auto& dir_entry : fs::recursive_directory_iterator(spslr_root)) {
+		if (!dir_entry.is_regular_file())
+			continue;
+
+		if (!dir_entry.path().string().ends_with(".spslr"))
+			continue;
+
+		if (!accumulate_file(dir_entry.path()))
+			return false;
+	}
+
+	return true;
+}
diff --git a/finalize/accumulation.h b/finalize/accumulation.h
new file mode 100644
index 0000000..f9a858e
--- /dev/null
+++ b/finalize/accumulation.h
@@ -0,0 +1,48 @@
+#pragma once
+#include <string>
+#include <cstddef>
+#include <unordered_map>
+#include <map>
+#include <list>
+
+struct IPIN {
+	std::string symbol;
+	std::size_t local_target;
+	std::size_t field_offset;
+};
+
+struct DPIN {
+	struct COMPONENT {
+		std::size_t offset;
+		std::size_t level;
+		std::size_t target; // local pin -> local target, global pin -> global target
+	};
+
+	std::string symbol;
+	std::list<COMPONENT> components;
+};
+
+struct FIELD {
+	std::size_t offset;
+	std::size_t size;
+	std::size_t alignment;
+	std::size_t flags;
+};
+
+struct TARGET {
+	std::string name;
+	std::size_t size;
+	std::map<std::size_t, FIELD> fields;
+};
+
+struct CU {
+	std::unordered_map<std::size_t, std::size_t> local_targets;
+	std::unordered_map<std::string, IPIN> ipins;
+	std::unordered_map<std::string, DPIN> dpins;
+};
+
+extern std::unordered_map<std::size_t, TARGET> targets;
+extern std::unordered_map<std::string, DPIN> global_dpins;
+extern std::unordered_map<std::string, CU> units;
+
+bool accumulate(const std::string& spslr_dir);
diff --git a/finalize/finalize.cpp b/finalize/finalize.cpp
index 278af21..2c4c711 100644
--- a/finalize/finalize.cpp
+++ b/finalize/finalize.cpp
@@ -2,48 +2,14 @@
 #include <string>
 #include <getopt.h>
 
-/*
-TODO
-1. Recursively gather all spslr CU files
-	cu uid -> data {
-		target uid -> fields
-		ipin label -> data
-		dpin symbol -> data (list of sub dpins)
-	}
-2. Merge types between CUs (per CU mapping to global target UIDs)
-3. Find dpatch application order based on levels (high level (very nested) to low level (root))
-4. Loop over all symbols of the binary
-	-> associate blocks via CU uid symbol
-	-> find __spslr_program symbol (spslr vaddr pivot)
-5. Find virtual address and file address for all pins
-6. Emit patcher program into final executable and set __spslr_program
-*/
+#include "accumulation.h"
 
 /*
-<sourcefile>.spslr:
-
-SPSLR <CU filename> <CU uid symbol>
-target <name> <local uid> <size> <field count>
-f <offset> <size> <flags>
-f <offset> <size> <flags>
-...
-ipin <label> <target uid> <field offset>
-ipin <label> <target uid> <field offset>
-ipin <label> <target uid> <field offset>
-...
-dpin <local/global> <symbol> <offset> <level> <target uid>
-dpin <local/global> <symbol> <offset> <level> <target uid>
-...
-
-
-
-Datapins for same var/symbol are randomized in order of their level, from bottom of nest to top
-The CU uid symbol helps differentiating between e.g. "file.c" and "sub/file.c" (symbtab has no idea)
-Between CUs, types with the same name HAVE TO HAVE the same layout -> randomized together
-To begin with, anonymous types are not allowed for randomization (later solved with hash(type) instead of name)!
-
-Note -> field alignment should probably be gathered by pinpoint plugin!
-
+Notes:
+	Datapins for same var/symbol are randomized in order of their level, from bottom of nest to top
+	The CU uid symbol helps differentiating between e.g. "file.c" and "sub/file.c" (symbtab has no idea)
+	Between CUs, types with the same name HAVE TO HAVE the same layout -> randomized together
+	To begin with, anonymous types are not allowed for randomization (later solved with hash(type) instead of name)!
 */
 
 int main(int argc, char** argv) {
@@ -101,7 +67,22 @@ int main(int argc, char** argv) {
 		return 1;
 	}
 
+	if (!accumulate(spslr_dir)) {
+		std::cerr << "Failed to accumulate data from spslr directory!" << std::endl;
+		return 1;
+	}
+
+	std::cout << "Gathered a total of " << targets.size() << " distinct targets from "
+		<< units.size() << " compilation units!" << std::endl;
+
 	// TODO
-	std::cout << "Hello World!" << std::endl;
+	/*
+	1. Find dpatch application order based on levels (high level (very nested) to low level (root))
+	2. Loop over all symbols of the binary
+		-> associate blocks via CU uid symbol
+		-> find __spslr_program symbol (spslr vaddr pivot)
+	3. Find virtual address and file address for all pins
+	4. Emit patcher program into final executable and set __spslr_program
+	*/
 }
 
diff --git a/pinpoint/final/on_finish_unit.cpp b/pinpoint/final/on_finish_unit.cpp
index c408673..97603a2 100644
--- a/pinpoint/final/on_finish_unit.cpp
+++ b/pinpoint/final/on_finish_unit.cpp
@@ -135,8 +135,9 @@ void on_finish_unit(void* plugin_data, void* user_data) {
 			pinpoint_fatal("on_finish_unit encountered incomplete target type");
 
 		for (const auto& [off, field] : target.fields()) {
-			// f <offset> <size> <flags>
-			out << "f " << field.offset << " " << field.size << " " << field.flags << std::endl;
+			// f <offset> <size> <alignment> <flags>
+			std::size_t field_alignment = field.size; // TODO
+			out << "f " << field.offset << " " << field.size << " " << field_alignment << " " << field.flags << std::endl;
 		}
 	}