feat: integrate Rabbitizer via FetchContent

ps2xAnalyzer/CMakeLists.txt

@@ -3,23 +3,24 @@ project(PS2Analyzer VERSION 0.1.0 LANGUAGES CXX)
 
 set(CMAKE_CXX_STANDARD 20)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
- 
+
 file(GLOB_RECURSE PS2ANALYZER_SOURCES
     "src/*.cpp"
 )
- 
+
 add_executable(ps2_analyzer ${PS2ANALYZER_SOURCES})
- 
+
 target_include_directories(ps2_analyzer PRIVATE
     ${CMAKE_CURRENT_SOURCE_DIR}/include
     ${CMAKE_SOURCE_DIR}/ps2xRecomp/include
 )
- 
+
 target_link_libraries(ps2_analyzer PRIVATE
     fmt::fmt
     ps2_recomp
+    rabbitizer
 )
- 
+
 install(TARGETS ps2_analyzer
     RUNTIME DESTINATION bin
 )

ps2xAnalyzer/include/ps2recomp/elf_analyzer.h

@@ -2,8 +2,8 @@
 #define PS2RECOMP_ELF_ANALYZER_H
 
 #include "ps2recomp/elf_parser.h"
-#include "ps2recomp/r5900_decoder.h"
 #include "ps2recomp/types.h"
+#include "ps2recomp/instructions.h"
 #include <string>
 #include <vector>
 #include <unordered_set>
@@ -26,25 +26,24 @@ namespace ps2recomp
     private:
         std::string m_elfPath;
         std::unique_ptr<ElfParser> m_elfParser;
-        std::unique_ptr<R5900Decoder> m_decoder;
 
         std::vector<Function> m_functions;
         std::vector<Symbol> m_symbols;
         std::vector<Section> m_sections;
         std::vector<Relocation> m_relocations;
- 
+
         std::unordered_set<std::string> m_libFunctions;
         std::unordered_set<std::string> m_skipFunctions;
         std::unordered_map<std::string, std::set<std::string>> m_functionDataUsage;
         std::unordered_map<uint32_t, std::string> m_commonDataAccess;
- 
+
         std::map<uint32_t, uint32_t> m_patches;
         std::map<uint32_t, std::string> m_patchReasons;
- 
+
         std::unordered_map<uint32_t, CFG> m_functionCFGs;
         std::vector<JumpTable> m_jumpTables;
         std::unordered_map<uint32_t, std::vector<FunctionCall>> m_functionCalls;
- 
+
         void initializeLibraryFunctions();
         void analyzeEntryPoint();
         void analyzeLibraryFunctions();
@@ -57,12 +56,12 @@ namespace ps2recomp
         void analyzeRegisterUsage();
         void analyzeFunctionSignatures();
         void optimizePatches();
- 
+
         bool identifyMemcpyPattern(const Function &func);
         bool identifyMemsetPattern(const Function &func);
         bool identifyStringOperationPattern(const Function &func);
         bool identifyMathPattern(const Function &func);
- 
+
         bool isSystemFunction(const std::string &name) const;
         bool isLibraryFunction(const std::string &name) const;
         std::vector<Instruction> decodeFunction(const Function &function);

ps2xAnalyzer/src/elf_analyzer.cpp

@@ -1,4 +1,5 @@
 #include "ps2recomp/elf_analyzer.h"
+#include <rabbitizer.h>
 #include <iostream>
 #include <sstream>
 #include <algorithm>
@@ -20,7 +21,6 @@ namespace ps2recomp
         : m_elfPath(elfPath)
     {
         m_elfParser = std::make_unique<ElfParser>(elfPath);
-        m_decoder = std::make_unique<R5900Decoder>();
 
         initializeLibraryFunctions();
     }
@@ -192,7 +192,7 @@ namespace ps2recomp
         const std::vector<std::string> stdLibFuncs = {
             // I/O functions
             "printf", "sprintf", "snprintf", "fprintf", "vprintf", "vfprintf", "vsprintf", "vsnprintf",
-            "puts", "putchar", "getchar", "gets", "fgets", "fputs", "scanf", "fscanf", "sscanf", 
+            "puts", "putchar", "getchar", "gets", "fgets", "fputs", "scanf", "fscanf", "sscanf",
             "sprint", "sbprintf",
 
             // Memory management
@@ -255,7 +255,7 @@ namespace ps2recomp
             {
                 if (inst.opcode == OPCODE_JAL)
                 {
-                    uint32_t target = (inst.address & 0xF0000000) | (inst.target << 2);
+                    uint32_t target = inst.target;
 
                     for (const auto &func : m_functions)
                     {
@@ -537,7 +537,7 @@ namespace ps2recomp
 
                     if (nextInst.opcode == OPCODE_J || nextInst.opcode == OPCODE_JAL)
                     {
-                        uint32_t jumpTarget = (nextInst.address & 0xF0000000) | (nextInst.target << 2);
+                        uint32_t jumpTarget = nextInst.target;
 
                         for (const auto &section : m_sections)
                         {
@@ -652,7 +652,7 @@ namespace ps2recomp
 
                     if (inst.opcode == OPCODE_JAL)
                     {
-                        targetAddr = (inst.address & 0xF0000000) | (inst.target << 2);
+                        targetAddr = inst.target;
                     }
                     else
                     {
@@ -1358,7 +1358,7 @@ namespace ps2recomp
                 // Jump target for J/JAL
                 if ((inst.opcode == OPCODE_J || inst.opcode == OPCODE_JAL) && !inst.isCall)
                 {
-                    uint32_t target = (inst.address & 0xF0000000) | (inst.target << 2);
+                    uint32_t target = inst.target;
                     leaders.insert(target);
                 }
             }
@@ -1439,7 +1439,7 @@ namespace ps2recomp
                 if (lastInst.opcode == OPCODE_J || lastInst.opcode == OPCODE_JAL)
                 {
                     // Direct jump
-                    uint32_t targetAddr = (lastInst.address & 0xF0000000) | (lastInst.target << 2);
+                    uint32_t targetAddr = lastInst.target;
 
                     // Only add successor if it's within this function
                     if (targetAddr >= function.start && targetAddr < function.end &&
@@ -1580,16 +1580,33 @@ namespace ps2recomp
 
             uint32_t rawInstruction = m_elfParser->readWord(addr);
 
-            try
-            {
-                Instruction inst = m_decoder->decodeInstruction(addr, rawInstruction);
-                instructions.push_back(inst);
-            }
-            catch (const std::exception &e)
-            {
-                std::cerr << "Error decoding instruction at " << formatAddress(addr)
-                          << ": " << e.what() << std::endl;
-            }
+            RabbitizerInstruction instr;
+            RabbitizerInstruction_init(&instr, rawInstruction, addr);
+
+            Instruction inst;
+            inst.address = addr;
+            inst.raw = rawInstruction;
+            inst.opcode = RAB_INSTR_GET_opcode(&instr);
+            inst.rs = RAB_INSTR_GET_rs(&instr);
+            inst.rt = RAB_INSTR_GET_rt(&instr);
+            inst.rd = RAB_INSTR_GET_rd(&instr);
+            inst.sa = RAB_INSTR_GET_sa(&instr);
+            inst.function = RAB_INSTR_GET_function(&instr);
+            inst.immediate = RAB_INSTR_GET_immediate(&instr);
+            inst.simmediate = static_cast<uint32_t>(RabbitizerInstruction_getProcessedImmediate(&instr));
+            inst.target = RabbitizerInstruction_getInstrIndexAsVram(&instr);
+            inst.hasDelaySlot = RabbitizerInstruction_hasDelaySlot(&instr);
+            inst.isCall = RabbitizerInstruction_isFunctionCall(&instr);
+            inst.isReturn = RabbitizerInstruction_isReturn(&instr);
+            inst.isBranch = (RabbitizerInstruction_getBranchOffsetGeneric(&instr) != 0) ||
+                            RabbitizerInstruction_isJumptableJump(&instr);
+            inst.isJump = RabbitizerInstruction_isUnconditionalBranch(&instr) ||
+                          RabbitizerInstruction_isJumptableJump(&instr);
+            inst.isMMI = (inst.opcode == OPCODE_MMI);
+            inst.isVU = (inst.opcode == OPCODE_COP2);
+
+            instructions.push_back(inst);
+            RabbitizerInstruction_destroy(&instr);
         }
 
         return instructions;

ps2xRecomp/CMakeLists.txt

@@ -29,6 +29,28 @@ FetchContent_Declare(
 )
 FetchContent_MakeAvailable(fmt)
 
+FetchContent_Declare(
+    rabbitizer
+    GIT_REPOSITORY https://github.com/Decompollaborate/rabbitizer.git
+    GIT_TAG 1.x
+)
+FetchContent_MakeAvailable(rabbitizer)
+
+file(GLOB_RECURSE RABBITIZER_SOURCES CONFIGURE_DEPENDS
+    "${rabbitizer_SOURCE_DIR}/src/analysis/*.c"
+    "${rabbitizer_SOURCE_DIR}/src/common/*.c"
+    "${rabbitizer_SOURCE_DIR}/src/instructions/*.c"
+    "${rabbitizer_SOURCE_DIR}/src/instructions/*/*.c"
+)
+
+add_library(rabbitizer STATIC ${RABBITIZER_SOURCES})
+
+target_include_directories(rabbitizer PUBLIC
+    "${rabbitizer_SOURCE_DIR}/include"
+    "${rabbitizer_SOURCE_DIR}/tables"
+    "${rabbitizer_SOURCE_DIR}/tables/tables"
+)
+
 file(GLOB_RECURSE PS2RECOMP_SOURCES
     "src/*.cpp"
 )
@@ -60,11 +82,13 @@ target_include_directories(ps2_recomp PRIVATE
 target_link_libraries(ps2recomp PRIVATE
     fmt::fmt
     toml11::toml11
+    rabbitizer
 )
 
 target_link_libraries(ps2_recomp PUBLIC
     fmt::fmt
     toml11::toml11
+    rabbitizer
 )
 
 install(TARGETS ps2recomp ps2_recomp

ps2xRecomp/example_config.toml

@@ -30,7 +30,7 @@ instructions = [
 function = "printf"
 code = '''
 // Custom printf implementation
-void printf(uint8_t* rdram, R5900Context* ctx) {
+void printf(uint8_t* rdram, Ps2CpuContext* ctx) {
     // Implementation here
 }
 '''
@@ -40,7 +40,7 @@ void printf(uint8_t* rdram, R5900Context* ctx) {
 address = "0x100000"
 code = '''
 // Custom implementation for function at 0x100000
-void func_00100000(uint8_t* rdram, R5900Context* ctx) {
+void func_00100000(uint8_t* rdram, Ps2CpuContext* ctx) {
     // Implementation here
 }
 '''

ps2xRecomp/include/ps2recomp/code_generator.h

@@ -120,6 +120,7 @@ namespace ps2recomp
 
         Symbol *findSymbolByAddress(uint32_t address);
         std::string getFunctionName(uint32_t address);
+        std::string sanitizeFunctionName(const std::string &name);
         std::string getGeneratedFunctionName(const Function &function);
     };
 

ps2xRecomp/include/ps2recomp/decoder.h

@@ -0,0 +1,15 @@
+#ifndef PS2RECOMP_DECODER_H
+#define PS2RECOMP_DECODER_H
+
+#include "ps2recomp/types.h"
+#include "ps2recomp/instructions.h"
+#include <cstdint>
+
+namespace ps2recomp
+{
+    Instruction decodeInstruction(uint32_t address, uint32_t rawInstruction);
+    uint32_t getBranchTarget(const Instruction &inst);
+    uint32_t getJumpTarget(const Instruction &inst);
+}
+
+#endif // PS2RECOMP_DECODER_H

ps2xRecomp/include/ps2recomp/ps2_recompiler.h

@@ -3,7 +3,6 @@
 
 #include "ps2recomp/types.h"
 #include "ps2recomp/elf_parser.h"
-#include "ps2recomp/r5900_decoder.h"
 #include "ps2recomp/code_generator.h"
 #include "ps2recomp/config_manager.h"
 #include <string>
@@ -28,7 +27,6 @@ namespace ps2recomp
     private:
         ConfigManager m_configManager;
         std::unique_ptr<ElfParser> m_elfParser;
-        std::unique_ptr<R5900Decoder> m_decoder;
         std::unique_ptr<CodeGenerator> m_codeGenerator;
         RecompilerConfig m_config;