thumb: add disassembler

Signed-off-by: Amneesh Singh <natto@weirdnatto.in>

src/cpu/arm/disassembler.cc

@@ -1,8 +1,7 @@
 #include "instruction.hh"
 #include "util/bits.hh"
 
-namespace matar {
-namespace arm {
+namespace matar::arm {
 std::string
 Instruction::disassemble() {
     auto condition = stringify(this->condition);
@@ -232,4 +231,3 @@ Instruction::disassemble() {
       data);
 }
 }
-}

src/cpu/arm/instruction.cc

@@ -2,8 +2,7 @@
 #include "util/bits.hh"
 #include <iterator>
 
-namespace matar {
-namespace arm {
+namespace matar::arm {
 Instruction::Instruction(uint32_t insn)
   : condition(static_cast<Condition>(bit_range(insn, 28, 31))) {
     // Branch and exhcange
@@ -275,4 +274,3 @@ Instruction::Instruction(uint32_t insn)
     }
 }
 }
-}

src/cpu/arm/instruction.hh

@@ -5,8 +5,7 @@
 #include <fmt/ostream.h>
 #include <variant>
 
-namespace matar {
-namespace arm {
+namespace matar::arm {
 
 // https://en.cppreference.com/w/cpp/utility/variant/visit
 template<class... Ts>
@@ -223,4 +222,3 @@ struct Instruction {
 #endif
 };
 }
-}

src/cpu/cpu-impl.cc

@@ -3,6 +3,7 @@
 #include "util/log.hh"
 #include <algorithm>
 #include <cstdio>
+#include <type_traits>
 
 namespace matar {
 CpuImpl::CpuImpl(const Bus& bus) noexcept

src/cpu/cpu-impl.hh

@@ -15,6 +15,7 @@ class CpuImpl {
     void chg_mode(const Mode to);
     void exec(const arm::Instruction instruction);
 
+    // TODO: get rid of this
 #ifndef MATAR_CPU_TESTS
   private:
 #endif

src/cpu/meson.build

@@ -5,4 +5,5 @@ lib_sources += files(
   'alu.cc'
 )
 
-subdir('arm')
+subdir('arm')
+subdir('thumb')

src/cpu/thumb/disassembler.cc

@@ -0,0 +1,150 @@
+#include "instruction.hh"
+#include "util/bits.hh"
+
+namespace matar::thumb {
+std::string
+Instruction::disassemble() {
+    return std::visit(
+      overloaded{
+        [](MoveShiftedRegister& data) {
+            return fmt::format("{} R{:d},R{:d},#{:d}",
+                               stringify(data.opcode),
+                               data.rd,
+                               data.rs,
+                               data.offset);
+        },
+        [](AddSubtract& data) {
+            return fmt::format("{} R{:d},R{:d},{}{:d}",
+                               stringify(data.opcode),
+                               data.rd,
+                               data.rs,
+                               (data.imm ? '#' : 'R'),
+                               data.offset);
+        },
+        [](MovCmpAddSubImmediate& data) {
+            return fmt::format(
+              "{} R{:d},#{:d}", stringify(data.opcode), data.rd, data.offset);
+        },
+        [](AluOperations& data) {
+            return fmt::format(
+              "{} R{:d},R{:d}", stringify(data.opcode), data.rd, data.rs);
+        },
+        [](HiRegisterOperations& data) {
+            if (data.opcode == HiRegisterOperations::OpCode::BX) {
+                return fmt::format("{} R{:d}", stringify(data.opcode), data.rs);
+            }
+
+            return fmt::format(
+              "{} R{:d},R{:d}", stringify(data.opcode), data.rd, data.rs);
+        },
+
+        [](PcRelativeLoad& data) {
+            return fmt::format("LDR R{:d},[PC,#{:d}]", data.rd, data.word);
+        },
+        [](LoadStoreRegisterOffset& data) {
+            return fmt::format("{}{} R{:d},[R{:d},R{:d}]",
+                               (data.load ? "LDR" : "STR"),
+                               (data.byte ? "B" : ""),
+                               data.rd,
+                               data.rb,
+                               data.ro);
+        },
+        [](LoadStoreSignExtendedHalfword& data) {
+            if (!data.s && !data.h) {
+                return fmt::format(
+                  "STRH R{:d},[R{:d},R{:d}]", data.rd, data.rb, data.ro);
+            }
+
+            return fmt::format("{}{} R{:d},[R{:d},R{:d}]",
+                               (data.s ? "LDS" : "LDR"),
+                               (data.h ? 'H' : 'B'),
+                               data.rd,
+                               data.rb,
+                               data.ro);
+        },
+        [](LoadStoreImmediateOffset& data) {
+            return fmt::format("{}{} R{:d},[R{:d},#{:d}]",
+                               (data.load ? "LDR" : "STR"),
+                               (data.byte ? "B" : ""),
+                               data.rd,
+                               data.rb,
+                               data.offset);
+        },
+        [](LoadStoreHalfword& data) {
+            return fmt::format("{} R{:d},[R{:d},#{:d}]",
+                               (data.load ? "LDRH" : "STRH"),
+                               data.rd,
+                               data.rb,
+                               data.offset);
+        },
+        [](SpRelativeLoad& data) {
+            return fmt::format("{} R{:d},[SP,#{:d}]",
+                               (data.load ? "LDR" : "STR"),
+                               data.rd,
+                               data.word);
+        },
+        [](LoadAddress& data) {
+            return fmt::format("ADD R{:d},{},#{:d}",
+                               data.rd,
+                               (data.sp ? "SP" : "PC"),
+                               data.word);
+        },
+        [](AddOffsetStackPointer& data) {
+            return fmt::format(
+              "ADD SP,#{}{:d}", (data.sign ? '-' : '+'), data.word);
+        },
+        [](PushPopRegister& data) {
+            std::string regs;
+
+            for (uint8_t i = 0; i < 16; i++) {
+                if (get_bit(data.regs, i))
+                    fmt::format_to(std::back_inserter(regs), "R{:d},", i);
+            };
+
+            if (data.load) {
+                if (data.pclr)
+                    regs += "PC";
+                else
+                    regs.pop_back();
+
+                return fmt::format("POP {{{}}}", regs);
+            } else {
+                if (data.pclr)
+                    regs += "LR";
+                else
+                    regs.pop_back();
+
+                return fmt::format("PUSH {{{}}}", regs);
+            }
+        },
+        [](MultipleLoad& data) {
+            std::string regs;
+
+            for (uint8_t i = 0; i < 16; i++) {
+                if (get_bit(data.regs, i))
+                    fmt::format_to(std::back_inserter(regs), "R{:d},", i);
+            };
+
+            regs.pop_back();
+
+            return fmt::format(
+              "{} R{}!,{{{}}}", (data.load ? "LDMIA" : "STMIA"), data.rb, regs);
+        },
+        [](SoftwareInterrupt) { return std::string("SWI"); },
+        [](ConditionalBranch& data) {
+            return fmt::format("B{} {:d}",
+                               stringify(data.condition),
+                               data.offset);
+        },
+        [](UnconditionalBranch& data) {
+            return fmt::format("B {:d}", data.offset);
+        },
+        [](LongBranchWithLink& data) {
+            // duh this manual be empty for H = 0
+            return fmt::format(
+              "BL{} {:d}", (data.high ? "H" : ""), data.offset);
+        },
+        [](auto) { return std::string("unknown instruction"); } },
+      data);
+}
+}

src/cpu/thumb/instruction.cc

@@ -0,0 +1,191 @@
+#include "instruction.hh"
+#include "util/bits.hh"
+
+namespace matar::thumb {
+Instruction::Instruction(uint16_t insn) {
+    // Format 2: Add/Subtract
+    if ((insn & 0xF800) == 0x1800) {
+        uint8_t rd     = bit_range(insn, 0, 2);
+        uint8_t rs     = bit_range(insn, 3, 5);
+        uint8_t offset = bit_range(insn, 6, 8);
+        AddSubtract::OpCode opcode =
+          static_cast<AddSubtract::OpCode>(get_bit(insn, 9));
+        bool imm = get_bit(insn, 10);
+
+        data = AddSubtract{
+            .rd = rd, .rs = rs, .offset = offset, .opcode = opcode, .imm = imm
+        };
+
+        // Format 1:  Move Shifted Register
+    } else if ((insn & 0xE000) == 0x0000) {
+        uint8_t rd       = bit_range(insn, 0, 2);
+        uint8_t rs       = bit_range(insn, 3, 5);
+        uint8_t offset   = bit_range(insn, 6, 10);
+        ShiftType opcode = static_cast<ShiftType>(bit_range(insn, 11, 12));
+
+        data = MoveShiftedRegister{
+            .rd = rd, .rs = rs, .offset = offset, .opcode = opcode
+        };
+
+        // Format 3: Move/compare/add/subtract immediate
+    } else if ((insn & 0xE000) == 0x2000) {
+        uint8_t offset = bit_range(insn, 0, 7);
+        uint8_t rd     = bit_range(insn, 8, 10);
+        MovCmpAddSubImmediate::OpCode opcode =
+          static_cast<MovCmpAddSubImmediate::OpCode>(bit_range(insn, 11, 12));
+
+        data =
+          MovCmpAddSubImmediate{ .offset = offset, .rd = rd, .opcode = opcode };
+
+        // Format 4: ALU operations
+    } else if ((insn & 0xFC00) == 0x4000) {
+        uint8_t rd = bit_range(insn, 0, 2);
+        uint8_t rs = bit_range(insn, 3, 5);
+        AluOperations::OpCode opcode =
+          static_cast<AluOperations::OpCode>(bit_range(insn, 6, 9));
+
+        data = AluOperations{ .rd = rd, .rs = rs, .opcode = opcode };
+
+        // Format 5: Hi register operations/branch exchange
+    } else if ((insn & 0xFC00) == 0x4400) {
+        uint8_t rd = bit_range(insn, 0, 2);
+        uint8_t rs = bit_range(insn, 3, 5);
+        bool hi_2  = get_bit(insn, 6);
+        bool hi_1  = get_bit(insn, 7);
+        HiRegisterOperations::OpCode opcode =
+          static_cast<HiRegisterOperations::OpCode>(bit_range(insn, 8, 9));
+
+        rd += (hi_1 ? LO_GPR_COUNT : 0);
+        rs += (hi_2 ? LO_GPR_COUNT : 0);
+
+        data = HiRegisterOperations{ .rd = rd, .rs = rs, .opcode = opcode };
+        // Format 6: PC-relative load
+    } else if ((insn & 0xF800) == 0x4800) {
+        uint8_t word = bit_range(insn, 0, 7);
+        uint8_t rd   = bit_range(insn, 8, 10);
+
+        data = PcRelativeLoad{ .word = word, .rd = rd };
+
+        // Format 7: Load/store with register offset
+    } else if ((insn & 0xF200) == 0x5000) {
+        uint8_t rd = bit_range(insn, 0, 2);
+        uint8_t rb = bit_range(insn, 3, 5);
+        uint8_t ro = bit_range(insn, 6, 8);
+        bool byte  = get_bit(insn, 10);
+        bool load  = get_bit(insn, 11);
+
+        data = LoadStoreRegisterOffset{
+            .rd = rd, .rb = rb, .ro = ro, .byte = byte, .load = load
+        };
+
+        // Format 8: Load/store sign-extended byte/halfword
+    } else if ((insn & 0xF200) == 0x5200) {
+        uint8_t rd = bit_range(insn, 0, 2);
+        uint8_t rb = bit_range(insn, 3, 5);
+        uint8_t ro = bit_range(insn, 6, 8);
+        bool s     = get_bit(insn, 10);
+        bool h     = get_bit(insn, 11);
+
+        data = LoadStoreSignExtendedHalfword{
+            .rd = rd, .rb = rb, .ro = ro, .s = s, .h = h
+        };
+
+        // Format 9: Load/store with immediate offset
+    } else if ((insn & 0xF000) == 0x6000) {
+        uint8_t rd     = bit_range(insn, 0, 2);
+        uint8_t rb     = bit_range(insn, 3, 5);
+        uint8_t offset = bit_range(insn, 6, 10);
+        bool load      = get_bit(insn, 11);
+        bool byte      = get_bit(insn, 12);
+
+        data = LoadStoreImmediateOffset{
+            .rd = rd, .rb = rb, .offset = offset, .load = load, .byte = byte
+        };
+
+        // Format 10: Load/store halfword
+    } else if ((insn & 0xF000) == 0x8000) {
+        uint8_t rd     = bit_range(insn, 0, 2);
+        uint8_t rb     = bit_range(insn, 3, 5);
+        uint8_t offset = bit_range(insn, 6, 10);
+        bool load      = get_bit(insn, 11);
+
+        data = LoadStoreHalfword{
+            .rd = rd, .rb = rb, .offset = offset, .load = load
+        };
+
+        // Format 11: SP-relative load/store
+    } else if ((insn & 0xF000) == 0x9000) {
+        uint8_t word = bit_range(insn, 0, 7);
+        uint8_t rd   = bit_range(insn, 8, 10);
+        bool load    = get_bit(insn, 11);
+
+        data = SpRelativeLoad{ .word = word, .rd = rd, .load = load };
+
+        // Format 12: Load address
+    } else if ((insn & 0xF000) == 0xA000) {
+        uint8_t word = bit_range(insn, 0, 7);
+        uint8_t rd   = bit_range(insn, 8, 10);
+        bool sp      = get_bit(insn, 11);
+
+        data = LoadAddress{ .word = word, .rd = rd, .sp = sp };
+
+        // Format 12: Load address
+    } else if ((insn & 0xF000) == 0xA000) {
+        uint8_t word = bit_range(insn, 0, 7);
+        uint8_t rd   = bit_range(insn, 8, 10);
+        bool sp      = get_bit(insn, 11);
+
+        data = LoadAddress{ .word = word, .rd = rd, .sp = sp };
+
+        // Format 13: Add offset to stack pointer
+    } else if ((insn & 0xFF00) == 0xB000) {
+        uint8_t word = bit_range(insn, 0, 6);
+        bool sign    = get_bit(insn, 7);
+
+        data = AddOffsetStackPointer{ .word = word, .sign = sign };
+
+        // Format 14: Push/pop registers
+    } else if ((insn & 0xF600) == 0xB400) {
+        uint8_t regs = bit_range(insn, 0, 7);
+        bool pclr    = get_bit(insn, 8);
+        bool load    = get_bit(insn, 11);
+
+        data = PushPopRegister{ .regs = regs, .pclr = pclr, .load = load };
+
+        // Format 15: Multiple load/store
+    } else if ((insn & 0xF000) == 0xC000) {
+        uint8_t regs = bit_range(insn, 0, 7);
+        uint8_t rb   = bit_range(insn, 8, 10);
+        bool load    = get_bit(insn, 11);
+
+        data = MultipleLoad{ .regs = regs, .rb = rb, .load = load };
+
+        // Format 17: Software interrupt
+    } else if ((insn & 0xFF00) == 0xDF00) {
+        data = SoftwareInterrupt{};
+
+        // Format 16: Conditional branch
+    } else if ((insn & 0xF000) == 0xD000) {
+        uint16_t offset     = bit_range(insn, 0, 7);
+        Condition condition = static_cast<Condition>(bit_range(insn, 8, 11));
+
+        data = ConditionalBranch{ .offset = static_cast<uint16_t>(offset << 1),
+                                  .condition = condition };
+
+        // Format 18: Unconditional branch
+    } else if ((insn & 0xF800) == 0xE000) {
+        uint16_t offset = bit_range(insn, 0, 10);
+
+        data =
+          UnconditionalBranch{ .offset = static_cast<uint16_t>(offset << 1) };
+
+        // Format 19: Long branch with link
+    } else if ((insn & 0xF000) == 0xF000) {
+        uint16_t offset = bit_range(insn, 0, 10);
+        bool high       = get_bit(insn, 11);
+
+        data = LongBranchWithLink{ .offset = static_cast<uint16_t>(offset << 1),
+                                   .high   = high };
+    }
+}
+}

src/cpu/thumb/instruction.hh

@@ -0,0 +1,282 @@
+#pragma once
+
+#include "cpu/alu.hh"
+#include "cpu/psr.hh"
+#include <cstdint>
+#include <fmt/ostream.h>
+#include <variant>
+
+namespace matar::thumb {
+
+// https://en.cppreference.com/w/cpp/utility/variant/visit
+template<class... Ts>
+struct overloaded : Ts... {
+    using Ts::operator()...;
+};
+template<class... Ts>
+overloaded(Ts...) -> overloaded<Ts...>;
+
+static constexpr size_t INSTRUCTION_SIZE = 2;
+static constexpr uint8_t LO_GPR_COUNT    = 8;
+
+struct MoveShiftedRegister {
+    uint8_t rd;
+    uint8_t rs;
+    uint8_t offset;
+    ShiftType opcode;
+};
+
+struct AddSubtract {
+    enum class OpCode {
+        ADD = 0,
+        SUB = 1
+    };
+
+    uint8_t rd;
+    uint8_t rs;
+    uint8_t offset;
+    OpCode opcode;
+    bool imm;
+};
+
+constexpr auto
+stringify(AddSubtract::OpCode opcode) {
+#define CASE(opcode)                                                           \
+    case AddSubtract::OpCode::opcode:                                          \
+        return #opcode;
+
+    switch (opcode) {
+        CASE(ADD)
+        CASE(SUB)
+    }
+
+#undef CASE
+    return "";
+}
+
+struct MovCmpAddSubImmediate {
+    enum class OpCode {
+        MOV = 0b00,
+        CMP = 0b01,
+        ADD = 0b10,
+        SUB = 0b11
+    };
+
+    uint8_t offset;
+    uint8_t rd;
+    OpCode opcode;
+};
+
+constexpr auto
+stringify(MovCmpAddSubImmediate::OpCode opcode) {
+#define CASE(opcode)                                                           \
+    case MovCmpAddSubImmediate::OpCode::opcode:                                \
+        return #opcode;
+
+    switch (opcode) {
+        CASE(MOV)
+        CASE(CMP)
+        CASE(ADD)
+        CASE(SUB)
+    }
+
+#undef CASE
+    return "";
+}
+
+struct AluOperations {
+    enum class OpCode {
+        AND = 0b0000,
+        EOR = 0b0001,
+        LSL = 0b0010,
+        LSR = 0b0011,
+        ASR = 0b0100,
+        ADC = 0b0101,
+        SBC = 0b0110,
+        ROR = 0b0111,
+        TST = 0b1000,
+        NEG = 0b1001,
+        CMP = 0b1010,
+        CMN = 0b1011,
+        ORR = 0b1100,
+        MUL = 0b1101,
+        BIC = 0b1110,
+        MVN = 0b1111
+    };
+
+    uint8_t rd;
+    uint8_t rs;
+    OpCode opcode;
+};
+
+constexpr auto
+stringify(AluOperations::OpCode opcode) {
+
+#define CASE(opcode)                                                           \
+    case AluOperations::OpCode::opcode:                                        \
+        return #opcode;
+
+    switch (opcode) {
+        CASE(AND)
+        CASE(EOR)
+        CASE(LSL)
+        CASE(LSR)
+        CASE(ASR)
+        CASE(ADC)
+        CASE(SBC)
+        CASE(ROR)
+        CASE(TST)
+        CASE(NEG)
+        CASE(CMP)
+        CASE(CMN)
+        CASE(ORR)
+        CASE(MUL)
+        CASE(BIC)
+        CASE(MVN)
+    }
+
+#undef CASE
+    return "";
+}
+
+struct HiRegisterOperations {
+    enum class OpCode {
+        ADD = 0b00,
+        CMP = 0b01,
+        MOV = 0b10,
+        BX  = 0b11
+    };
+
+    uint8_t rd;
+    uint8_t rs;
+    OpCode opcode;
+};
+
+constexpr auto
+stringify(HiRegisterOperations::OpCode opcode) {
+#define CASE(opcode)                                                           \
+    case HiRegisterOperations::OpCode::opcode:                                 \
+        return #opcode;
+
+    switch (opcode) {
+        CASE(ADD)
+        CASE(CMP)
+        CASE(MOV)
+        CASE(BX)
+    }
+
+#undef CASE
+    return "";
+}
+
+struct PcRelativeLoad {
+    uint8_t word;
+    uint8_t rd;
+};
+
+struct LoadStoreRegisterOffset {
+    uint8_t rd;
+    uint8_t rb;
+    uint8_t ro;
+    bool byte;
+    bool load;
+};
+
+struct LoadStoreSignExtendedHalfword {
+    uint8_t rd;
+    uint8_t rb;
+    uint8_t ro;
+    bool s;
+    bool h;
+};
+
+struct LoadStoreImmediateOffset {
+    uint8_t rd;
+    uint8_t rb;
+    uint8_t offset;
+    bool load;
+    bool byte;
+};
+
+struct LoadStoreHalfword {
+    uint8_t rd;
+    uint8_t rb;
+    uint8_t offset;
+    bool load;
+};
+
+struct SpRelativeLoad {
+    uint8_t word;
+    uint8_t rd;
+    bool load;
+};
+
+struct LoadAddress {
+    uint8_t word;
+    uint8_t rd;
+    bool sp;
+};
+
+struct AddOffsetStackPointer {
+    uint8_t word;
+    bool sign;
+};
+
+struct PushPopRegister {
+    uint8_t regs;
+    bool pclr;
+    bool load;
+};
+
+struct MultipleLoad {
+    uint8_t regs;
+    uint8_t rb;
+    bool load;
+};
+
+struct ConditionalBranch {
+    uint16_t offset;
+    Condition condition;
+};
+
+struct SoftwareInterrupt {};
+
+struct UnconditionalBranch {
+    uint16_t offset;
+};
+
+struct LongBranchWithLink {
+    uint16_t offset;
+    bool high;
+};
+
+using InstructionData = std::variant<MoveShiftedRegister,
+                                     AddSubtract,
+                                     MovCmpAddSubImmediate,
+                                     AluOperations,
+                                     HiRegisterOperations,
+                                     PcRelativeLoad,
+                                     LoadStoreRegisterOffset,
+                                     LoadStoreSignExtendedHalfword,
+                                     LoadStoreImmediateOffset,
+                                     LoadStoreHalfword,
+                                     SpRelativeLoad,
+                                     LoadAddress,
+                                     AddOffsetStackPointer,
+                                     PushPopRegister,
+                                     MultipleLoad,
+                                     ConditionalBranch,
+                                     SoftwareInterrupt,
+                                     UnconditionalBranch,
+                                     LongBranchWithLink>;
+
+struct Instruction {
+    InstructionData data;
+
+    Instruction(uint16_t insn);
+
+#ifdef DISASSEMBLER
+    std::string disassemble();
+#endif
+};
+}

src/cpu/thumb/meson.build

@@ -0,0 +1,7 @@
+lib_sources += files(
+  'instruction.cc'
+)
+
+if get_option('disassembler')
+  lib_sources += files('disassembler.cc')
+endif

src/meson.build

@@ -6,7 +6,7 @@ lib_sources = files(
 subdir('util')
 subdir('cpu')
 
-lib_cpp_args = [ ]
+lib_cpp_args = []
 
 fmt = dependency('fmt', version : '>=10.1.0', static: true)
 if not fmt.found()

tests/cpu/meson.build

@@ -1 +1,2 @@
-subdir('arm')
+subdir('arm')
+subdir('thumb')

tests/cpu/thumb/instruction.cc

@@ -0,0 +1,439 @@
+#include "cpu/thumb/instruction.hh"
+#include <catch2/catch_test_macros.hpp>
+
+static constexpr auto TAG = "[thumb][disassembly]";
+
+using namespace matar;
+using namespace thumb;
+
+TEST_CASE("Move Shifted Register", TAG) {
+    uint16_t raw = 0b0001001101100011;
+    Instruction instruction(raw);
+    MoveShiftedRegister* lsl = nullptr;
+
+    REQUIRE((lsl = std::get_if<MoveShiftedRegister>(&instruction.data)));
+    CHECK(lsl->rd == 3);
+    CHECK(lsl->rs == 4);
+    CHECK(lsl->offset == 13);
+    CHECK(lsl->opcode == ShiftType::ASR);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "ASR R3,R4,#13");
+
+    lsl->opcode = ShiftType::LSR;
+    CHECK(instruction.disassemble() == "LSR R3,R4,#13");
+
+    lsl->opcode = ShiftType::LSL;
+    CHECK(instruction.disassemble() == "LSL R3,R4,#13");
+#endif
+}
+
+TEST_CASE("Add/Subtract", TAG) {
+    uint16_t raw = 0b0001111101001111;
+    Instruction instruction(raw);
+    AddSubtract* add = nullptr;
+
+    REQUIRE((add = std::get_if<AddSubtract>(&instruction.data)));
+    CHECK(add->rd == 7);
+    CHECK(add->rs == 1);
+    CHECK(add->offset == 5);
+    CHECK(add->opcode == AddSubtract::OpCode::SUB);
+    CHECK(add->imm == true);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "SUB R7,R1,#5");
+
+    add->imm = false;
+    CHECK(instruction.disassemble() == "SUB R7,R1,R5");
+
+    add->opcode = AddSubtract::OpCode::ADD;
+    CHECK(instruction.disassemble() == "ADD R7,R1,R5");
+#endif
+}
+
+TEST_CASE("Move/Compare/Add/Subtract Immediate", TAG) {
+    uint16_t raw = 0b0010111001011011;
+    Instruction instruction(raw);
+    MovCmpAddSubImmediate* mov = nullptr;
+
+    REQUIRE((mov = std::get_if<MovCmpAddSubImmediate>(&instruction.data)));
+    CHECK(mov->offset == 91);
+    CHECK(mov->rd == 6);
+    CHECK(mov->opcode == MovCmpAddSubImmediate::OpCode::CMP);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "CMP R6,#91");
+
+    mov->opcode = MovCmpAddSubImmediate::OpCode::ADD;
+    CHECK(instruction.disassemble() == "ADD R6,#91");
+
+    mov->opcode = MovCmpAddSubImmediate::OpCode::SUB;
+    CHECK(instruction.disassemble() == "SUB R6,#91");
+
+    mov->opcode = MovCmpAddSubImmediate::OpCode::MOV;
+    CHECK(instruction.disassemble() == "MOV R6,#91");
+#endif
+}
+
+TEST_CASE("ALU Operations", TAG) {
+    uint16_t raw = 0b0100000110011111;
+    Instruction instruction(raw);
+    AluOperations* alu = nullptr;
+
+    REQUIRE((alu = std::get_if<AluOperations>(&instruction.data)));
+    CHECK(alu->rd == 7);
+    CHECK(alu->rs == 3);
+    CHECK(alu->opcode == AluOperations::OpCode::SBC);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "SBC R7,R3");
+
+#define OPCODE(op)                                                             \
+    alu->opcode = AluOperations::OpCode::op;                                   \
+    CHECK(instruction.disassemble() == #op " R7,R3");
+
+    OPCODE(AND)
+    OPCODE(EOR)
+    OPCODE(LSL)
+    OPCODE(LSR)
+    OPCODE(ASR)
+    OPCODE(ADC)
+    OPCODE(SBC)
+    OPCODE(ROR)
+    OPCODE(TST)
+    OPCODE(NEG)
+    OPCODE(CMP)
+    OPCODE(CMN)
+    OPCODE(ORR)
+    OPCODE(MUL)
+    OPCODE(BIC)
+    OPCODE(MVN)
+
+#undef OPCODE
+#endif
+}
+
+TEST_CASE("Hi Register Operations/Branch Exchange", TAG) {
+    HiRegisterOperations* hi = nullptr;
+
+    uint16_t raw = 0b0100011000011010;
+
+    SECTION("both lo") {
+        Instruction instruction(raw);
+        REQUIRE((hi = std::get_if<HiRegisterOperations>(&instruction.data)));
+
+        CHECK(hi->rd == 2);
+        CHECK(hi->rs == 3);
+    }
+
+    SECTION("hi rd") {
+        raw |= 1 << 7;
+        Instruction instruction(raw);
+        REQUIRE((hi = std::get_if<HiRegisterOperations>(&instruction.data)));
+
+        CHECK(hi->rd == 10);
+        CHECK(hi->rs == 3);
+    }
+
+    SECTION("hi rs") {
+        raw |= 1 << 6;
+        Instruction instruction(raw);
+        REQUIRE((hi = std::get_if<HiRegisterOperations>(&instruction.data)));
+
+        CHECK(hi->rd == 2);
+        CHECK(hi->rs == 11);
+    }
+
+    if (hi)
+        CHECK(hi->opcode == HiRegisterOperations::OpCode::MOV);
+
+    SECTION("both hi") {
+        raw |= 1 << 6;
+        raw |= 1 << 7;
+        Instruction instruction(raw);
+        REQUIRE((hi = std::get_if<HiRegisterOperations>(&instruction.data)));
+
+        CHECK(hi->rd == 10);
+        CHECK(hi->rs == 11);
+        CHECK(hi->opcode == HiRegisterOperations::OpCode::MOV);
+
+#ifdef DISASSEMBLER
+        CHECK(instruction.disassemble() == "MOV R10,R11");
+
+        hi->opcode = HiRegisterOperations::OpCode::ADD;
+        CHECK(instruction.disassemble() == "ADD R10,R11");
+
+        hi->opcode = HiRegisterOperations::OpCode::CMP;
+        CHECK(instruction.disassemble() == "CMP R10,R11");
+
+        hi->opcode = HiRegisterOperations::OpCode::BX;
+        CHECK(instruction.disassemble() == "BX R11");
+#endif
+    }
+}
+
+TEST_CASE("PC Relative Load", TAG) {
+    uint16_t raw = 0b0100101011100110;
+    Instruction instruction(raw);
+    PcRelativeLoad* ldr = nullptr;
+
+    REQUIRE((ldr = std::get_if<PcRelativeLoad>(&instruction.data)));
+    CHECK(ldr->word == 230);
+    CHECK(ldr->rd == 2);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "LDR R2,[PC,#230]");
+#endif
+}
+
+TEST_CASE("Load/Store with Register Offset", TAG) {
+    uint16_t raw = 0b0101000110011101;
+    Instruction instruction(raw);
+    LoadStoreRegisterOffset* ldr = nullptr;
+
+    REQUIRE((ldr = std::get_if<LoadStoreRegisterOffset>(&instruction.data)));
+    CHECK(ldr->rd == 5);
+    CHECK(ldr->rb == 3);
+    CHECK(ldr->ro == 6);
+    CHECK(ldr->byte == false);
+    CHECK(ldr->load == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "STR R5,[R3,R6]");
+
+    ldr->byte = true;
+    CHECK(instruction.disassemble() == "STRB R5,[R3,R6]");
+
+    ldr->load = true;
+    CHECK(instruction.disassemble() == "LDRB R5,[R3,R6]");
+
+    ldr->byte = false;
+    CHECK(instruction.disassemble() == "LDR R5,[R3,R6]");
+#endif
+}
+
+TEST_CASE("Load/Store Sign-Extended Byte/Halfword", TAG) {
+    uint16_t raw = 0b0101001110011101;
+    Instruction instruction(raw);
+    LoadStoreSignExtendedHalfword* ldr = nullptr;
+
+    REQUIRE(
+      (ldr = std::get_if<LoadStoreSignExtendedHalfword>(&instruction.data)));
+    CHECK(ldr->rd == 5);
+    CHECK(ldr->rb == 3);
+    CHECK(ldr->ro == 6);
+    CHECK(ldr->s == false);
+    CHECK(ldr->h == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "STRH R5,[R3,R6]");
+
+    ldr->h = true;
+    CHECK(instruction.disassemble() == "LDRH R5,[R3,R6]");
+
+    ldr->s = true;
+    CHECK(instruction.disassemble() == "LDSH R5,[R3,R6]");
+
+    ldr->h = false;
+    CHECK(instruction.disassemble() == "LDSB R5,[R3,R6]");
+#endif
+}
+
+TEST_CASE("Load/Store with Immediate Offset", TAG) {
+    uint16_t raw = 0b0110010110011101;
+    Instruction instruction(raw);
+    LoadStoreImmediateOffset* ldr = nullptr;
+
+    REQUIRE((ldr = std::get_if<LoadStoreImmediateOffset>(&instruction.data)));
+    CHECK(ldr->rd == 5);
+    CHECK(ldr->rb == 3);
+    CHECK(ldr->offset == 22);
+    CHECK(ldr->byte == false);
+    CHECK(ldr->load == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "STR R5,[R3,#22]");
+
+    ldr->byte = true;
+    CHECK(instruction.disassemble() == "STRB R5,[R3,#22]");
+
+    ldr->load = true;
+    CHECK(instruction.disassemble() == "LDRB R5,[R3,#22]");
+
+    ldr->byte = false;
+    CHECK(instruction.disassemble() == "LDR R5,[R3,#22]");
+#endif
+}
+
+TEST_CASE("Load/Store Halfword", TAG) {
+    uint16_t raw = 0b1000011010011101;
+    Instruction instruction(raw);
+    LoadStoreHalfword* ldr = nullptr;
+
+    REQUIRE((ldr = std::get_if<LoadStoreHalfword>(&instruction.data)));
+    CHECK(ldr->rd == 5);
+    CHECK(ldr->rb == 3);
+    CHECK(ldr->offset == 26);
+    CHECK(ldr->load == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "STRH R5,[R3,#26]");
+
+    ldr->load = true;
+    CHECK(instruction.disassemble() == "LDRH R5,[R3,#26]");
+#endif
+}
+
+TEST_CASE("SP-Relative Load/Store", TAG) {
+    uint16_t raw = 0b1001010010011101;
+    Instruction instruction(raw);
+    SpRelativeLoad* ldr = nullptr;
+
+    REQUIRE((ldr = std::get_if<SpRelativeLoad>(&instruction.data)));
+    CHECK(ldr->rd == 4);
+    CHECK(ldr->word == 157);
+    CHECK(ldr->load == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "STR R4,[SP,#157]");
+
+    ldr->load = true;
+    CHECK(instruction.disassemble() == "LDR R4,[SP,#157]");
+#endif
+}
+
+TEST_CASE("Load Adress", TAG) {
+    uint16_t raw = 0b1010000110001111;
+    Instruction instruction(raw);
+    LoadAddress* add = nullptr;
+
+    REQUIRE((add = std::get_if<LoadAddress>(&instruction.data)));
+    CHECK(add->word == 143);
+    CHECK(add->rd == 1);
+    CHECK(add->sp == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "ADD R1,PC,#143");
+
+    add->sp = true;
+    CHECK(instruction.disassemble() == "ADD R1,SP,#143");
+#endif
+}
+
+TEST_CASE("Add Offset to Stack Pointer", TAG) {
+    uint16_t raw = 0b1011000000100101;
+    Instruction instruction(raw);
+    AddOffsetStackPointer* add = nullptr;
+
+    REQUIRE((add = std::get_if<AddOffsetStackPointer>(&instruction.data)));
+    CHECK(add->word == 37);
+    CHECK(add->sign == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "ADD SP,#+37");
+
+    add->sign = true;
+    CHECK(instruction.disassemble() == "ADD SP,#-37");
+#endif
+}
+
+TEST_CASE("Push/Pop Registers", TAG) {
+    uint16_t raw = 0b1011010000110101;
+    Instruction instruction(raw);
+    PushPopRegister* push = nullptr;
+
+    REQUIRE((push = std::get_if<PushPopRegister>(&instruction.data)));
+    CHECK(push->regs == 53);
+    CHECK(push->pclr == false);
+    CHECK(push->load == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "PUSH {R0,R2,R4,R5}");
+
+    push->pclr = true;
+    CHECK(instruction.disassemble() == "PUSH {R0,R2,R4,R5,LR}");
+
+    push->load = true;
+    CHECK(instruction.disassemble() == "POP {R0,R2,R4,R5,PC}");
+
+    push->pclr = false;
+    CHECK(instruction.disassemble() == "POP {R0,R2,R4,R5}");
+#endif
+}
+
+TEST_CASE("Multiple Load/Store", TAG) {
+    uint16_t raw = 0b1100011001100101;
+    Instruction instruction(raw);
+    MultipleLoad* ldm = nullptr;
+
+    REQUIRE((ldm = std::get_if<MultipleLoad>(&instruction.data)));
+    CHECK(ldm->regs == 101);
+    CHECK(ldm->rb == 6);
+    CHECK(ldm->load == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "STMIA R6!,{R0,R2,R5,R6}");
+
+    ldm->load = true;
+    CHECK(instruction.disassemble() == "LDMIA R6!,{R0,R2,R5,R6}");
+#endif
+}
+
+TEST_CASE("Conditional Branch", TAG) {
+    uint16_t raw = 0b1101100101110100;
+    Instruction instruction(raw);
+    ConditionalBranch* b = nullptr;
+
+    REQUIRE((b = std::get_if<ConditionalBranch>(&instruction.data)));
+    // 116 << 2
+    CHECK(b->offset == 232);
+    CHECK(b->condition == Condition::LS);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "BLS 232");
+#endif
+}
+
+TEST_CASE("SoftwareInterrupt") {
+    uint16_t raw = 0b1101111100110011;
+    Instruction instruction(raw);
+    SoftwareInterrupt* swi = nullptr;
+
+    REQUIRE((swi = std::get_if<SoftwareInterrupt>(&instruction.data)));
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "SWI");
+#endif
+}
+
+TEST_CASE("Unconditional Branch") {
+    uint16_t raw = 0b1110011100110011;
+    Instruction instruction(raw);
+    UnconditionalBranch* b = nullptr;
+
+    REQUIRE((b = std::get_if<UnconditionalBranch>(&instruction.data)));
+    // 1843 << 2
+    REQUIRE(b->offset == 3686);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "B 3686");
+#endif
+}
+
+TEST_CASE("Long Branch with link") {
+    uint16_t raw = 0b1111010011101100;
+    Instruction instruction(raw);
+    LongBranchWithLink* bl = nullptr;
+
+    REQUIRE((bl = std::get_if<LongBranchWithLink>(&instruction.data)));
+    // 1260 << 1
+    CHECK(bl->offset == 2520);
+    CHECK(bl->high == false);
+
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "BL 2520");
+
+    bl->high = true;
+    CHECK(instruction.disassemble() == "BLH 2520");
+#endif
+}

tests/cpu/thumb/meson.build

@@ -0,0 +1,3 @@
+tests_sources += files(
+  'instruction.cc'
+)