feat: refactor analyzer to not realy only on debug symbols (#53)

fix: patching NOP things that code gen already know how to handle feat: added a bug on JAL/J/JAR code gen feat: enhanced tom file fix: fix memory layout for ps2 macros feat: added a lot of not working garbabe to runtime (fix later) feat: some code organization feat: added new tests feat: update readme
2026-10-01 10:26:07 -04:00 · 2026-02-13 15:02:34 -03:00
parent 1d1b79d5f0
commit 8f747334d6
26 changed files with 10987 additions and 2111 deletions
@@ -30,6 +30,51 @@ static Instruction makeNop(uint32_t address)
    return inst;
 }

+static Instruction makeJal(uint32_t address, uint32_t target)
+{
+    Instruction inst{};
+    inst.address = address;
+    inst.opcode = OPCODE_JAL;
+    inst.target = (target >> 2) & 0x3FFFFFF;
+    inst.hasDelaySlot = true;
+    inst.raw = (OPCODE_JAL << 26) | inst.target;
+    return inst;
+}
+
+static Instruction makeJalr(uint32_t address, uint8_t rs, uint8_t rd)
+{
+    Instruction inst{};
+    inst.address = address;
+    inst.opcode = OPCODE_SPECIAL;
+    inst.function = SPECIAL_JALR;
+    inst.rs = rs;
+    inst.rd = rd; // Destination for link address (default 31)
+    inst.hasDelaySlot = true;
+    inst.raw = (OPCODE_SPECIAL << 26) | (rs << 21) | (0 << 16) | (rd << 11) | (0 << 6) | SPECIAL_JALR;
+    return inst;
+}
+
+static Instruction makeJr(uint32_t address, uint8_t rs)
+{
+    Instruction inst{};
+    inst.address = address;
+    inst.opcode = OPCODE_SPECIAL;
+    inst.function = SPECIAL_JR;
+    inst.rs = rs;
+    inst.hasDelaySlot = true;
+    inst.raw = (OPCODE_SPECIAL << 26) | (rs << 21) | SPECIAL_JR;
+    return inst;
+}
+
+static void printGeneratedCode(const std::string& name, const std::string& code)
+{
+#ifdef PRINT_GENERATED_CODE
+    std::cout << "=== Generated Code for " << name << " ===" << std::endl;
+    std::cout << code << std::endl;
+    std::cout << "========================================" << std::endl;
+#endif
+}
+
 void register_code_generator_tests()
 {
    MiniTest::Case("CodeGenerator", [](TestCase &tc)
@@ -57,6 +102,7 @@ void register_code_generator_tests()

        CodeGenerator gen({});
        std::string generated = gen.generateFunction(func, instructions, false);
+        printGeneratedCode("emits labels and gotos for internal branches", generated);

        t.IsTrue(generated.find("label_100c:") != std::string::npos, "branch target should emit a label");
        t.IsTrue(generated.find("goto label_100c;") != std::string::npos, "internal branch should jump via goto");
@@ -79,6 +125,7 @@ void register_code_generator_tests()

        CodeGenerator gen({});
        std::string generated = gen.generateFunction(func, instructions, false);
+        printGeneratedCode("labels delay slot when it is a branch target", generated);

        t.IsTrue(generated.find("label_2004:") != std::string::npos, "delay slot that is a target should emit a label");
        t.IsTrue(generated.find("goto label_2004;") != std::string::npos, "branch to delay slot should use goto");
@@ -102,6 +149,7 @@ void register_code_generator_tests()

        CodeGenerator gen({});
        std::string generated = gen.generateFunction(func, instructions, false);
+        printGeneratedCode("branches outside function still set pc", generated);

        t.IsTrue(generated.find("ctx->pc = 0x") != std::string::npos, "external branch should set ctx->pc");
        t.IsTrue(generated.find("goto label_") == std::string::npos, "external branch should not use goto");
@@ -133,6 +181,7 @@ void register_code_generator_tests()

        CodeGenerator gen({targetSym});
        std::string generated = gen.generateFunction(func, instructions, false);
+        printGeneratedCode("jumps to known symbols call by name", generated);

        t.IsTrue(generated.find("target_func(rdram, ctx, runtime); return;") != std::string::npos,
                 "jump to known function should emit direct call");
@@ -158,6 +207,7 @@ void register_code_generator_tests()

            CodeGenerator gen({});
            std::string generated = gen.generateFunction(func, instructions, false);
+            printGeneratedCode("jump to unknown target sets pc", generated);

            t.IsTrue(generated.find("ctx->pc = 0x") != std::string::npos, "unknown jump target should set ctx->pc");
            t.IsTrue(generated.find("goto label_") == std::string::npos, "external jump should not use goto");
@@ -183,6 +233,7 @@ void register_code_generator_tests()
            gen.setRenamedFunctions({{0x8000, "renamed_target"}});

            std::string sw = gen.generateJumpTableSwitch(inst, 0x0, entries);
+            printGeneratedCode("renamed function used in jump table", sw);

        t.IsTrue(sw.find("renamed_target(rdram, ctx, runtime);") != std::string::npos,
                 "jump table should use renamed function name");
@@ -215,10 +266,186 @@ void register_code_generator_tests()
            gen.setRenamedFunctions({{targetSym.address, "ps2___is_pointer"}});

            std::string generated = gen.generateFunction(func, instructions, false);
+            printGeneratedCode("reserved identifiers are sanitized and used in calls", generated);

            t.IsTrue(generated.find("void ps2___is_pointer(") != std::string::npos,
                     "definition should use sanitized name");
            t.IsTrue(generated.find("ps2___is_pointer(rdram, ctx, runtime); return;") != std::string::npos,
-                     "call should use sanitized name");
-        }); });
+                "call should use sanitized name but got: " + generated);
+        });
+
+        tc.Run("JAL to known function emits call and check", [](TestCase &t) {
+            Function func;
+            func.name = "jal_test";
+            func.start = 0xA000;
+            func.end = 0xA020;
+            func.isRecompiled = true;
+            func.isStub = false;
+
+            Symbol targetSym;
+            targetSym.name = "some_func";
+            targetSym.address = 0xB000;
+            targetSym.isFunction = true;
+
+            // 0xA000: JAL 0xB000
+            // 0xA004: NOP (delay slot)
+            Instruction jal = makeJal(0xA000, 0xB000);
+            Instruction delay = makeNop(0xA004);
+            
+            CodeGenerator gen({targetSym});
+            std::string generated = gen.generateFunction(func, {jal, delay}, false);
+            printGeneratedCode("JAL to known function emits call and check", generated);
+
+            // Expect:
+            // SET_GPR_U32(ctx, 31, 0xA008u);
+            // ctx->pc = 0xA004u;
+            // ... delay slot ...
+            // some_func(rdram, ctx, runtime);
+            // if (ctx->pc != 0xA008u) { return; }
+
+            t.IsTrue(generated.find("SET_GPR_U32(ctx, 31, 0xA008u);") != std::string::npos, "JAL should set RA");
+            t.IsTrue(generated.find("some_func(rdram, ctx, runtime);") != std::string::npos, "JAL should call function");
+            t.IsTrue(generated.find("if (ctx->pc != 0xA008u) { return; }") != std::string::npos, "JAL should check return PC");
+        });
+
+        tc.Run("JAL to internal target becomes goto", [](TestCase &t) {
+            Function func;
+            func.name = "jal_internal";
+            func.start = 0xC000;
+            func.end = 0xC020;
+            func.isRecompiled = true;
+            func.isStub = false;
+
+            // 0xC000: JAL 0xC010
+            // 0xC004: NOP
+            // ...
+            // 0xC010: NOP
+            Instruction jal = makeJal(0xC000, 0xC010);
+            Instruction delay = makeNop(0xC004);
+            Instruction targetInst = makeNop(0xC010);
+
+            CodeGenerator gen({});
+            std::string generated = gen.generateFunction(func, {jal, delay, targetInst}, false);
+            printGeneratedCode("JAL to internal target becomes goto", generated);
+
+            t.IsTrue(generated.find("SET_GPR_U32(ctx, 31, 0xC008u);") != std::string::npos, "Internal JAL should set RA");
+            t.IsTrue(generated.find("goto label_c010;") != std::string::npos, "Internal JAL should use goto");
+        });
+
+        tc.Run("JALR emits indirect call", [](TestCase &t) {
+            Function func;
+            func.name = "jalr_test";
+            func.start = 0xD000;
+            func.end = 0xD020;
+            func.isRecompiled = true;
+            func.isStub = false;
+
+            // 0xD000: JALR $4, $31 (call addr in $4, link to $31)
+            // 0xD004: NOP
+            Instruction jalr = makeJalr(0xD000, 4, 31);
+            Instruction delay = makeNop(0xD004);
+
+            CodeGenerator gen({});
+            std::string generated = gen.generateFunction(func, {jalr, delay}, false);
+            printGeneratedCode("JALR emits indirect call", generated);
+
+            t.IsTrue(generated.find("uint32_t jumpTarget = GPR_U32(ctx, 4);") != std::string::npos, "JALR should read target from RS");
+            t.IsTrue(generated.find("SET_GPR_U32(ctx, 31, 0xD008u);") != std::string::npos, "JALR should set link register");
+            t.IsTrue(generated.find("auto targetFn = runtime->lookupFunction(jumpTarget);") != std::string::npos, "JALR should lookup function");
+            t.IsTrue(generated.find("targetFn(rdram, ctx, runtime);") != std::string::npos, "JALR should call function");
+             t.IsTrue(generated.find("if (ctx->pc != 0xD008u) { return; }") != std::string::npos, "JALR should check return PC");
+        }); 
+
+        tc.Run("backward BEQ emits label and goto (sign-extended offset)", [](TestCase &t) {
+            Function func;
+            func.name = "backward_branch";
+            func.start = 0x1100;
+            func.end = 0x1120;
+            func.isRecompiled = true;
+            func.isStub = false;
+
+            // 0x1100: nop
+            // 0x1104: beq $1,$1, target 0x1100 (offset = -2 words)
+            // 0x1108: nop (delay)
+            std::vector<Instruction> instructions;
+            instructions.push_back(makeNop(0x1100));
+
+            Instruction br = makeBranch(0x1104, 0); 
+            br.simmediate = static_cast<uint32_t>(static_cast<int16_t>(-2)); 
+            instructions.push_back(br);
+
+            instructions.push_back(makeNop(0x1108));
+            instructions.push_back(makeNop(0x110c));
+
+            CodeGenerator gen({});
+            std::string generated = gen.generateFunction(func, instructions, false);
+            printGeneratedCode("backward BEQ emits label and goto (sign-extended offset)", generated);
+
+            t.IsTrue(generated.find("label_1100:") != std::string::npos, "target should emit a label");
+            t.IsTrue(generated.find("goto label_1100;") != std::string::npos, "backward internal branch should goto label");
+        });
+
+        tc.Run("branch-likely places delay slot only in taken path", [](TestCase &t) {
+            Function func;
+            func.name = "branch_likely";
+            func.start = 0x1200;
+            func.end = 0x1220;
+            func.isRecompiled = true;
+            func.isStub = false;
+
+            Instruction br{};
+            br.address = 0x1200;
+            br.opcode = OPCODE_BEQL;   // likely
+            br.rs = 1;
+            br.rt = 2;
+            br.simmediate = 1;         // target = 0x1208
+            br.isBranch = true;
+            br.hasDelaySlot = true;
+            br.raw = 0;  
+
+            Instruction delay{};
+            delay.address = 0x1204;
+            delay.opcode = OPCODE_ADDIU;
+            delay.rs = 0;
+            delay.rt = 7;              // make it non-nop so translation is distinctive
+            delay.simmediate = 123;
+            delay.raw = 0;
+
+            Instruction target = makeNop(0x1208);
+
+            CodeGenerator gen({});
+            std::string generated = gen.generateFunction(func, { br, delay, target }, false);
+            printGeneratedCode("branch-likely places delay slot only in taken path", generated);
+ 
+            t.IsTrue(generated.find("SET_GPR_S32(ctx, 7,") != std::string::npos, "delay slot should be translated");
+            t.IsTrue(generated.find("if (branch_taken_0x1200)") != std::string::npos, "should generate branch_taken variable and if for likely branch");
+        });
+
+        tc.Run("JR $31 emits switch for internal return targets", [](TestCase &t) {
+            Function func;
+            func.name = "jr_ra_switch";
+            func.start = 0x1300;
+            func.end = 0x1340;
+            func.isRecompiled = true;
+            func.isStub = false;
+
+            // Create an internal JAL so collectInternalBranchTargets inserts returnAddr (0x1308) as internal target.
+            Instruction jal = makeJal(0x1300, 0x1310);
+            Instruction jalDelay = makeNop(0x1304);
+            Instruction atTarget = makeNop(0x1310);
+
+            // JR $31 at 0x1314 with delay slot at 0x1318
+            Instruction jr = makeJr(0x1314, 31);
+            Instruction jrDelay = makeNop(0x1318);
+
+            CodeGenerator gen({});
+            std::string generated = gen.generateFunction(func, { jal, jalDelay, atTarget, jr, jrDelay }, false);
+            printGeneratedCode("JR $31 emits switch for internal return targets", generated);
+
+            t.IsTrue(generated.find("switch (jumpTarget)") != std::string::npos, "JR $31 should emit switch for internal targets");
+            t.IsTrue(generated.find("case 0x1308u: goto label_1308;") != std::string::npos, "switch should include return address from internal JAL");
+        });
+    
+    
+    });
 }