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/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

@@ -1,24 +1,10 @@
 #include "instruction.hh"
 #include "util/bits.hh"
-#include <iterator>
-
-namespace matar {
-namespace thumb {
 
+namespace matar::thumb {
 Instruction::Instruction(uint16_t insn) {
-    // Format 1:  Move Shifted Register
-    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 2: Add/Subtract
-    } else if ((insn & 0xF800) == 0x1800) {
+    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);
@@ -30,6 +16,17 @@ Instruction::Instruction(uint16_t insn) {
             .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);
@@ -58,9 +55,10 @@ Instruction::Instruction(uint16_t insn) {
         HiRegisterOperations::OpCode opcode =
           static_cast<HiRegisterOperations::OpCode>(bit_range(insn, 8, 9));
 
-        data = HiRegisterOperations{
-            .rd = rd, .rs = rs, .hi_2 = hi_2, .hi_1 = hi_1, .opcode = opcode
-        };
+        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);
@@ -168,24 +166,26 @@ Instruction::Instruction(uint16_t insn) {
 
         // Format 16: Conditional branch
     } else if ((insn & 0xF000) == 0xD000) {
-        uint8_t offset      = bit_range(insn, 0, 7);
+        uint16_t offset     = bit_range(insn, 0, 7);
         Condition condition = static_cast<Condition>(bit_range(insn, 8, 11));
 
-        data = ConditionalBranch{ .offset = offset, .condition = condition };
+        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 = offset };
+        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 = offset, .high = high };
-    }
+        data = LongBranchWithLink{ .offset = static_cast<uint16_t>(offset << 1),
+                                   .high   = high };
     }
 }
 }

src/cpu/thumb/instruction.hh

@@ -1,13 +1,14 @@
 #pragma once
+
 #include "cpu/alu.hh"
 #include "cpu/psr.hh"
 #include <cstdint>
 #include <fmt/ostream.h>
 #include <variant>
 
-namespace matar {
-namespace thumb {
+namespace matar::thumb {
 
+// https://en.cppreference.com/w/cpp/utility/variant/visit
 template<class... Ts>
 struct overloaded : Ts... {
     using Ts::operator()...;
@@ -16,6 +17,7 @@ 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;
@@ -37,6 +39,21 @@ struct AddSubtract {
     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,
@@ -50,6 +67,23 @@ struct MovCmpAddSubImmediate {
     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,
@@ -75,6 +109,36 @@ struct AluOperations {
     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,
@@ -85,11 +149,26 @@ struct HiRegisterOperations {
 
     uint8_t rd;
     uint8_t rs;
-    bool hi_2;
-    bool hi_1;
     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;
@@ -156,7 +235,7 @@ struct MultipleLoad {
 };
 
 struct ConditionalBranch {
-    uint8_t offset;
+    uint16_t offset;
     Condition condition;
 };
 
@@ -196,35 +275,8 @@ struct Instruction {
 
     Instruction(uint16_t insn);
 
+#ifdef DISASSEMBLER
     std::string disassemble();
+#endif
 };
-
-std::ostream&
-operator<<(std::ostream& os, const AddSubtract::OpCode cond);
-
-std::ostream&
-operator<<(std::ostream& os, const MovCmpAddSubImmediate::OpCode cond);
-
-std::ostream&
-operator<<(std::ostream& os, const AluOperations::OpCode cond);
-
-std::ostream&
-operator<<(std::ostream& os, const HiRegisterOperations::OpCode cond);
-}
-}
-
-namespace fmt {
-template<>
-struct formatter<matar::thumb::AddSubtract::OpCode> : ostream_formatter {};
-
-template<>
-struct formatter<matar::thumb::MovCmpAddSubImmediate::OpCode>
-  : ostream_formatter {};
-
-template<>
-struct formatter<matar::thumb::AluOperations::OpCode> : ostream_formatter {};
-
-template<>
-struct formatter<matar::thumb::HiRegisterOperations::OpCode>
-  : ostream_formatter {};
 }

src/cpu/thumb/meson.build

@@ -1,3 +1,7 @@
 lib_sources += files(
   'instruction.cc'
 )
+
+if get_option('disassembler')
+  lib_sources += files('disassembler.cc')
+endif

tests/cpu/meson.build

@@ -1 +1,2 @@
 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'
+)