thumb: add disassembler

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

.clang-tidy

@@ -6,4 +6,5 @@ Checks: '
   , -cppcoreguidelines-macro-usage
   , -cppcoreguidelines-avoid-const-or-ref-data-members
   , -cppcoreguidelines-non-private-member-variables-in-classes
+  , -cppcoreguidelines-avoid-non-const-global-variables
 '

.envrc

@@ -1 +1 @@
-use flake .#matar-clang
+use flake

apps/target/main.cc

@@ -1,6 +1,7 @@
 #include "bus.hh"
 #include "cpu/cpu.hh"
 #include "memory.hh"
+#include "util/loglevel.hh"
 #include <array>
 #include <cstdlib>
 #include <fstream>
@@ -84,6 +85,8 @@ main(int argc, const char* argv[]) {
     std::flush(std::cout);
     std::flush(std::cout);
 
+    matar::set_log_level(matar::LogLevel::Debug);
+
     try {
         matar::Memory memory(std::move(bios), std::move(rom));
         matar::Bus bus(memory);

flake.nix

@@ -26,7 +26,7 @@
             ".hh"
             ".cc"
             ".build"
-            "meson_options.txt"
+            ".options"
           ];
         in
         rec {

include/memory.hh

@@ -17,12 +17,6 @@ class Memory {
     uint8_t read(size_t address) const;
     void write(size_t address, uint8_t byte);
 
-    uint16_t read_halfword(size_t address) const;
-    void write_halfword(size_t address, uint16_t halfword);
-
-    uint32_t read_word(size_t address) const;
-    void write_word(size_t address, uint32_t word);
-
   private:
 #define MEMORY_REGION(name, start, end)                                        \
     static constexpr size_t name##_START = start;                              \

include/meson.build

@@ -7,5 +7,6 @@ headers = files(
 inc = include_directories('.')
 
 subdir('cpu')
+subdir('util')
 
 install_headers(headers, subdir: meson.project_name(), preserve_path: true)

include/util/loglevel.hh

@@ -0,0 +1,14 @@
+#pragma once
+
+namespace matar {
+enum class LogLevel {
+    Off   = 1 << 0,
+    Error = 1 << 1,
+    Warn  = 1 << 2,
+    Info  = 1 << 3,
+    Debug = 1 << 4
+};
+
+void
+set_log_level(LogLevel level);
+}

include/util/meson.build

@@ -0,0 +1,3 @@
+headers += files(
+  'loglevel.hh'
+)

meson.build

@@ -4,7 +4,8 @@ project('matar', 'cpp',
   default_options : ['warning_level=3',
                      'werror=true',
                      'optimization=3',
-                     'cpp_std=c++20'])
+                     'cpp_std=c++20',
+                     'default_library=static'])
 
 compiler = meson.get_compiler('cpp')
 

meson.options

@@ -0,0 +1,2 @@
+option('tests', type : 'boolean', value : true, description: 'enable tests')
+option('disassembler', type: 'boolean', value: true, description: 'enable disassembler')

meson_options.txt

@@ -1 +0,0 @@
-option('tests', type : 'boolean', value : true, description: 'enable tests')

src/bus.cc

@@ -1,4 +1,5 @@
 #include "bus.hh"
+#include "util/log.hh"
 #include <memory>
 
 namespace matar {
@@ -17,21 +18,38 @@ Bus::write_byte(size_t address, uint8_t byte) {
 
 uint16_t
 Bus::read_halfword(size_t address) {
-    return memory->read_halfword(address);
+    if (address & 0b01)
+        glogger.warn("Reading a non aligned halfword address");
+
+    return memory->read(address) | memory->read(address + 1) << 8;
 }
 
 void
 Bus::write_halfword(size_t address, uint16_t halfword) {
-    memory->write_halfword(address, halfword);
+    if (address & 0b01)
+        glogger.warn("Writing to a non aligned halfword address");
+
+    memory->write(address, halfword & 0xFF);
+    memory->write(address + 1, halfword >> 8 & 0xFF);
 }
 
 uint32_t
 Bus::read_word(size_t address) {
-    return memory->read_word(address);
+    if (address & 0b11)
+        glogger.warn("Reading a non aligned word address");
+
+    return memory->read(address) | memory->read(address + 1) << 8 |
+           memory->read(address + 2) << 16 | memory->read(address + 3) << 24;
 }
 
 void
 Bus::write_word(size_t address, uint32_t word) {
-    memory->write_word(address, word);
+    if (address & 0b11)
+        glogger.warn("Writing to a non aligned word address");
+
+    memory->write(address, word & 0xFF);
+    memory->write(address + 1, word >> 8 & 0xFF);
+    memory->write(address + 2, word >> 16 & 0xFF);
+    memory->write(address + 3, word >> 24 & 0xFF);
 }
 }

src/cpu/alu.cc

@@ -48,24 +48,4 @@ eval_shift(ShiftType shift_type, uint32_t value, uint8_t amount, bool& carry) {
 
     return eval;
 }
-
-std::ostream&
-operator<<(std::ostream& os, const ShiftType shift_type) {
-
-#define CASE(type)                                                             \
-    case ShiftType::type:                                                      \
-        os << #type;                                                           \
-        break;
-
-    switch (shift_type) {
-        CASE(LSL)
-        CASE(LSR)
-        CASE(ASR)
-        CASE(ROR)
-    }
-
-#undef CASE
-
-    return os;
-}
 }

src/cpu/alu.hh

@@ -10,6 +10,24 @@ enum class ShiftType {
     ROR = 0b11
 };
 
+constexpr auto
+stringify(ShiftType shift_type) {
+#define CASE(type)                                                             \
+    case ShiftType::type:                                                      \
+        return #type;
+
+    switch (shift_type) {
+        CASE(LSL)
+        CASE(LSR)
+        CASE(ASR)
+        CASE(ROR)
+    }
+
+#undef CASE
+
+    return "";
+}
+
 struct ShiftData {
     ShiftType type;
     bool immediate;
@@ -23,13 +41,4 @@ struct Shift {
 
 uint32_t
 eval_shift(ShiftType shift_type, uint32_t value, uint8_t amount, bool& carry);
-
-// https://fmt.dev/dev/api.html#std-ostream-support
-std::ostream&
-operator<<(std::ostream& os, const ShiftType cond);
-}
-
-namespace fmt {
-template<>
-struct formatter<matar::ShiftType> : ostream_formatter {};
 }

src/cpu/arm/disassembler.cc

@@ -0,0 +1,233 @@
+#include "instruction.hh"
+#include "util/bits.hh"
+
+namespace matar::arm {
+std::string
+Instruction::disassemble() {
+    auto condition = stringify(this->condition);
+
+    return std::visit(
+      overloaded{
+        [condition](BranchAndExchange& data) {
+            return fmt::format("BX{} R{:d}", condition, data.rn);
+        },
+        [condition](Branch& data) {
+            return fmt::format(
+              "B{}{} 0x{:06X}", (data.link ? "L" : ""), condition, data.offset);
+        },
+        [condition](Multiply& data) {
+            if (data.acc) {
+                return fmt::format("MLA{}{} R{:d},R{:d},R{:d},R{:d}",
+                                   condition,
+                                   (data.set ? "S" : ""),
+                                   data.rd,
+                                   data.rm,
+                                   data.rs,
+                                   data.rn);
+            } else {
+                return fmt::format("MUL{}{} R{:d},R{:d},R{:d}",
+                                   condition,
+                                   (data.set ? "S" : ""),
+                                   data.rd,
+                                   data.rm,
+                                   data.rs);
+            }
+        },
+        [condition](MultiplyLong& data) {
+            return fmt::format("{}{}{}{} R{:d},R{:d},R{:d},R{:d}",
+                               (data.uns ? 'U' : 'S'),
+                               (data.acc ? "MLAL" : "MULL"),
+                               condition,
+                               (data.set ? "S" : ""),
+                               data.rdlo,
+                               data.rdhi,
+                               data.rm,
+                               data.rs);
+        },
+        [](Undefined) { return std::string("UND"); },
+        [condition](SingleDataSwap& data) {
+            return fmt::format("SWP{}{} R{:d},R{:d},[R{:d}]",
+                               condition,
+                               (data.byte ? "B" : ""),
+                               data.rd,
+                               data.rm,
+                               data.rn);
+        },
+        [condition](SingleDataTransfer& data) {
+            std::string expression;
+            std::string address;
+
+            if (const uint16_t* offset = std::get_if<uint16_t>(&data.offset)) {
+                if (*offset == 0) {
+                    expression = "";
+                } else {
+                    expression =
+                      fmt::format(",{}#{:d}", (data.up ? '+' : '-'), *offset);
+                }
+            } else if (const Shift* shift = std::get_if<Shift>(&data.offset)) {
+                // Shifts are always immediate in single data transfer
+                expression = fmt::format(",{}R{:d},{} #{:d}",
+                                         (data.up ? '+' : '-'),
+                                         shift->rm,
+                                         stringify(shift->data.type),
+                                         shift->data.operand);
+            }
+
+            return fmt::format(
+              "{}{}{}{} R{:d},[R{:d}{}]{}",
+              (data.load ? "LDR" : "STR"),
+              condition,
+              (data.byte ? "B" : ""),
+              (!data.pre && data.write ? "T" : ""),
+              data.rd,
+              data.rn,
+              (data.pre ? expression : ""),
+              (data.pre ? (data.write ? "!" : "") : expression));
+        },
+        [condition](HalfwordTransfer& data) {
+            std::string expression;
+
+            if (data.imm) {
+                if (data.offset == 0) {
+                    expression = "";
+                } else {
+                    expression = fmt::format(
+                      ",{}#{:d}", (data.up ? '+' : '-'), data.offset);
+                }
+            } else {
+                expression =
+                  fmt::format(",{}R{:d}", (data.up ? '+' : '-'), data.offset);
+            }
+
+            return fmt::format(
+              "{}{}{}{} R{:d},[R{:d}{}]{}",
+              (data.load ? "LDR" : "STR"),
+              condition,
+              (data.sign ? "S" : ""),
+              (data.half ? 'H' : 'B'),
+              data.rd,
+              data.rn,
+              (data.pre ? expression : ""),
+              (data.pre ? (data.write ? "!" : "") : expression));
+        },
+        [condition](BlockDataTransfer& 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{:d}{},{{{}}}{}",
+                               (data.load ? "LDM" : "STM"),
+                               condition,
+                               (data.up ? 'I' : 'D'),
+                               (data.pre ? 'B' : 'A'),
+                               data.rn,
+                               (data.write ? "!" : ""),
+                               regs,
+                               (data.s ? "^" : ""));
+        },
+        [condition](PsrTransfer& data) {
+            if (data.type == PsrTransfer::Type::Mrs) {
+                return fmt::format("MRS{} R{:d},{}",
+                                   condition,
+                                   data.operand,
+                                   (data.spsr ? "SPSR_all" : "CPSR_all"));
+            } else {
+                return fmt::format(
+                  "MSR{} {}_{},{}{}",
+                  condition,
+                  (data.spsr ? "SPSR" : "CPSR"),
+                  (data.type == PsrTransfer::Type::Msr_flg ? "flg" : "all"),
+                  (data.imm ? '#' : 'R'),
+                  data.operand);
+            }
+        },
+        [condition](DataProcessing& data) {
+            using OpCode = DataProcessing::OpCode;
+
+            std::string op_2;
+
+            if (const uint32_t* operand =
+                  std::get_if<uint32_t>(&data.operand)) {
+                op_2 = fmt::format("#{:d}", *operand);
+            } else if (const Shift* shift = std::get_if<Shift>(&data.operand)) {
+                op_2 = fmt::format("R{:d},{} {}{:d}",
+                                   shift->rm,
+                                   stringify(shift->data.type),
+                                   (shift->data.immediate ? '#' : 'R'),
+                                   shift->data.operand);
+            }
+
+            switch (data.opcode) {
+                case OpCode::MOV:
+                case OpCode::MVN:
+                    return fmt::format("{}{}{} R{:d},{}",
+                                       stringify(data.opcode),
+                                       condition,
+                                       (data.set ? "S" : ""),
+                                       data.rd,
+                                       op_2);
+                case OpCode::TST:
+                case OpCode::TEQ:
+                case OpCode::CMP:
+                case OpCode::CMN:
+                    return fmt::format("{}{} R{:d},{}",
+                                       stringify(data.opcode),
+                                       condition,
+                                       data.rn,
+                                       op_2);
+                default:
+                    return fmt::format("{}{}{} R{:d},R{:d},{}",
+                                       stringify(data.opcode),
+                                       condition,
+                                       (data.set ? "S" : ""),
+                                       data.rd,
+                                       data.rn,
+                                       op_2);
+            }
+        },
+        [condition](SoftwareInterrupt) {
+            return fmt::format("SWI{}", condition);
+        },
+        [condition](CoprocessorDataTransfer& data) {
+            std::string expression = fmt::format(",#{:d}", data.offset);
+            return fmt::format(
+              "{}{}{} p{:d},c{:d},[R{:d}{}]{}",
+              (data.load ? "LDC" : "STC"),
+              condition,
+              (data.len ? "L" : ""),
+              data.cpn,
+              data.crd,
+              data.rn,
+              (data.pre ? expression : ""),
+              (data.pre ? (data.write ? "!" : "") : expression));
+        },
+        [condition](CoprocessorDataOperation& data) {
+            return fmt::format("CDP{} p{},{},c{},c{},c{},{}",
+                               condition,
+                               data.cpn,
+                               data.cp_opc,
+                               data.crd,
+                               data.crn,
+                               data.crm,
+                               data.cp);
+        },
+        [condition](CoprocessorRegisterTransfer& data) {
+            return fmt::format("{}{} p{},{},R{},c{},c{},{}",
+                               (data.load ? "MRC" : "MCR"),
+                               condition,
+                               data.cpn,
+                               data.cp_opc,
+                               data.rd,
+                               data.crn,
+                               data.crm,
+                               data.cp);
+        },
+        [](auto) { return std::string("unknown instruction"); } },
+      data);
+}
+}

src/cpu/arm/exec.cc

@@ -2,27 +2,24 @@
 #include "util/bits.hh"
 #include "util/log.hh"
 
-using namespace logger;
-
 namespace matar {
 void
-CpuImpl::exec_arm(const arm::Instruction instruction) {
+CpuImpl::exec(const arm::Instruction instruction) {
     Condition cond            = instruction.condition;
     arm::InstructionData data = instruction.data;
 
-    debug(cpsr.condition(cond));
     if (!cpsr.condition(cond)) {
         return;
     }
 
     auto pc_error = [](uint8_t r) {
         if (r == PC_INDEX)
-            log_error("Using PC (R15) as operand register");
+            glogger.error("Using PC (R15) as operand register");
     };
 
     auto pc_warn = [](uint8_t r) {
         if (r == PC_INDEX)
-            log_warn("Using PC (R15) as operand register");
+            glogger.warn("Using PC (R15) as operand register");
     };
 
     using namespace arm;
@@ -62,8 +59,8 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
         },
         [this, pc_error](Multiply& data) {
             if (data.rd == data.rm)
-                log_error("rd and rm are not distinct in {}",
-                          typeid(data).name());
+                glogger.error("rd and rm are not distinct in {}",
+                              typeid(data).name());
 
             pc_error(data.rd);
             pc_error(data.rd);
@@ -81,8 +78,8 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
         [this, pc_error](MultiplyLong& data) {
             if (data.rdhi == data.rdlo || data.rdhi == data.rm ||
                 data.rdlo == data.rm)
-                log_error("rdhi, rdlo and rm are not distinct in {}",
-                          typeid(data).name());
+                glogger.error("rdhi, rdlo and rm are not distinct in {}",
+                              typeid(data).name());
 
             pc_error(data.rdhi);
             pc_error(data.rdlo);
@@ -123,7 +120,7 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
                 cpsr.set_v(0);
             }
         },
-        [](Undefined) { log_warn("Undefined instruction"); },
+        [](Undefined) { glogger.warn("Undefined instruction"); },
         [this, pc_error](SingleDataSwap& data) {
             pc_error(data.rm);
             pc_error(data.rn);
@@ -142,12 +139,12 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
             uint32_t address = gpr[data.rn];
 
             if (!data.pre && data.write)
-                log_warn("Write-back enabled with post-indexing in {}",
-                         typeid(data).name());
+                glogger.warn("Write-back enabled with post-indexing in {}",
+                             typeid(data).name());
 
             if (data.rn == PC_INDEX && data.write)
-                log_warn("Write-back enabled with base register as PC {}",
-                         typeid(data).name());
+                glogger.warn("Write-back enabled with base register as PC {}",
+                             typeid(data).name());
 
             if (data.write)
                 pc_warn(data.rn);
@@ -216,11 +213,11 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
             uint32_t offset  = 0;
 
             if (!data.pre && data.write)
-                log_error("Write-back enabled with post-indexing in {}",
-                          typeid(data).name());
+                glogger.error("Write-back enabled with post-indexing in {}",
+                              typeid(data).name());
 
             if (data.sign && !data.load)
-                log_error("Signed data found in {}", typeid(data).name());
+                glogger.error("Signed data found in {}", typeid(data).name());
 
             if (data.write)
                 pc_warn(data.rn);
@@ -294,8 +291,8 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
             pc_error(data.rn);
 
             if (cpsr.mode() == Mode::User && data.s) {
-                log_error("Bit S is set outside priviliged modes in {}",
-                          typeid(data).name());
+                glogger.error("Bit S is set outside priviliged modes in {}",
+                              typeid(data).name());
             }
 
             // we just change modes to load user registers
@@ -304,8 +301,9 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
                 chg_mode(Mode::User);
 
                 if (data.write) {
-                    log_error("Write-back enable for user bank registers in {}",
-                              typeid(data).name());
+                    glogger.error(
+                      "Write-back enable for user bank registers in {}",
+                      typeid(data).name());
                 }
             }
 
@@ -358,8 +356,8 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
         },
         [this, pc_error](PsrTransfer& data) {
             if (data.spsr && cpsr.mode() == Mode::User) {
-                log_error("Accessing SPSR in User mode in {}",
-                          typeid(data).name());
+                glogger.error("Accessing SPSR in User mode in {}",
+                              typeid(data).name());
             }
 
             Psr& psr = data.spsr ? spsr : cpsr;
@@ -513,8 +511,8 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
             if (data.set) {
                 if (data.rd == PC_INDEX) {
                     if (cpsr.mode() == Mode::User)
-                        log_error("Running {} in User mode",
-                                  typeid(data).name());
+                        glogger.error("Running {} in User mode",
+                                      typeid(data).name());
                     spsr = cpsr;
                 } else {
                     set_conditions();
@@ -536,7 +534,7 @@ CpuImpl::exec_arm(const arm::Instruction instruction) {
             spsr = cpsr;
         },
         [](auto& data) {
-            log_error("Unimplemented {} instruction", typeid(data).name());
+            glogger.error("Unimplemented {} instruction", typeid(data).name());
         } },
       data);
 }

src/cpu/arm/instruction.cc

@@ -2,9 +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
@@ -275,261 +273,4 @@ Instruction::Instruction(uint32_t insn)
         data = Undefined{};
     }
 }
-
-std::string
-Instruction::disassemble() {
-    // goddamn this is gore
-    // TODO: make this less ugly
-    return std::visit(
-      overloaded{
-        [this](BranchAndExchange& data) {
-            return fmt::format("BX{} R{:d}", condition, data.rn);
-        },
-        [this](Branch& data) {
-            return fmt::format(
-              "B{}{} 0x{:06X}", (data.link ? "L" : ""), condition, data.offset);
-        },
-        [this](Multiply& data) {
-            if (data.acc) {
-                return fmt::format("MLA{}{} R{:d},R{:d},R{:d},R{:d}",
-                                   condition,
-                                   (data.set ? "S" : ""),
-                                   data.rd,
-                                   data.rm,
-                                   data.rs,
-                                   data.rn);
-            } else {
-                return fmt::format("MUL{}{} R{:d},R{:d},R{:d}",
-                                   condition,
-                                   (data.set ? "S" : ""),
-                                   data.rd,
-                                   data.rm,
-                                   data.rs);
-            }
-        },
-        [this](MultiplyLong& data) {
-            return fmt::format("{}{}{}{} R{:d},R{:d},R{:d},R{:d}",
-                               (data.uns ? 'U' : 'S'),
-                               (data.acc ? "MLAL" : "MULL"),
-                               condition,
-                               (data.set ? "S" : ""),
-                               data.rdlo,
-                               data.rdhi,
-                               data.rm,
-                               data.rs);
-        },
-        [](Undefined) { return std::string("UND"); },
-        [this](SingleDataSwap& data) {
-            return fmt::format("SWP{}{} R{:d},R{:d},[R{:d}]",
-                               condition,
-                               (data.byte ? "B" : ""),
-                               data.rd,
-                               data.rm,
-                               data.rn);
-        },
-        [this](SingleDataTransfer& data) {
-            std::string expression;
-            std::string address;
-
-            if (const uint16_t* offset = std::get_if<uint16_t>(&data.offset)) {
-                if (*offset == 0) {
-                    expression = "";
-                } else {
-                    expression =
-                      fmt::format(",{}#{:d}", (data.up ? '+' : '-'), *offset);
-                }
-            } else if (const Shift* shift = std::get_if<Shift>(&data.offset)) {
-                // Shifts are always immediate in single data transfer
-                expression = fmt::format(",{}R{:d},{} #{:d}",
-                                         (data.up ? '+' : '-'),
-                                         shift->rm,
-                                         shift->data.type,
-                                         shift->data.operand);
-            }
-
-            return fmt::format(
-              "{}{}{}{} R{:d},[R{:d}{}]{}",
-              (data.load ? "LDR" : "STR"),
-              condition,
-              (data.byte ? "B" : ""),
-              (!data.pre && data.write ? "T" : ""),
-              data.rd,
-              data.rn,
-              (data.pre ? expression : ""),
-              (data.pre ? (data.write ? "!" : "") : expression));
-        },
-        [this](HalfwordTransfer& data) {
-            std::string expression;
-
-            if (data.imm) {
-                if (data.offset == 0) {
-                    expression = "";
-                } else {
-                    expression = fmt::format(
-                      ",{}#{:d}", (data.up ? '+' : '-'), data.offset);
-                }
-            } else {
-                expression =
-                  fmt::format(",{}R{:d}", (data.up ? '+' : '-'), data.offset);
-            }
-
-            return fmt::format(
-              "{}{}{}{} R{:d},[R{:d}{}]{}",
-              (data.load ? "LDR" : "STR"),
-              condition,
-              (data.sign ? "S" : ""),
-              (data.half ? 'H' : 'B'),
-              data.rd,
-              data.rn,
-              (data.pre ? expression : ""),
-              (data.pre ? (data.write ? "!" : "") : expression));
-        },
-        [this](BlockDataTransfer& 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{:d}{},{{{}}}{}",
-                               (data.load ? "LDM" : "STM"),
-                               condition,
-                               (data.up ? 'I' : 'D'),
-                               (data.pre ? 'B' : 'A'),
-                               data.rn,
-                               (data.write ? "!" : ""),
-                               regs,
-                               (data.s ? "^" : ""));
-        },
-        [this](PsrTransfer& data) {
-            if (data.type == PsrTransfer::Type::Mrs) {
-                return fmt::format("MRS{} R{:d},{}",
-                                   condition,
-                                   data.operand,
-                                   (data.spsr ? "SPSR_all" : "CPSR_all"));
-            } else {
-                return fmt::format(
-                  "MSR{} {}_{},{}{}",
-                  condition,
-                  (data.spsr ? "SPSR" : "CPSR"),
-                  (data.type == PsrTransfer::Type::Msr_flg ? "flg" : "all"),
-                  (data.imm ? '#' : 'R'),
-                  data.operand);
-            }
-        },
-        [this](DataProcessing& data) {
-            using OpCode = DataProcessing::OpCode;
-
-            std::string op_2;
-
-            if (const uint32_t* operand =
-                  std::get_if<uint32_t>(&data.operand)) {
-                op_2 = fmt::format("#{:d}", *operand);
-            } else if (const Shift* shift = std::get_if<Shift>(&data.operand)) {
-                op_2 = fmt::format("R{:d},{} {}{:d}",
-                                   shift->rm,
-                                   shift->data.type,
-                                   (shift->data.immediate ? '#' : 'R'),
-                                   shift->data.operand);
-            }
-
-            switch (data.opcode) {
-                case OpCode::MOV:
-                case OpCode::MVN:
-                    return fmt::format("{}{}{} R{:d},{}",
-                                       data.opcode,
-                                       condition,
-                                       (data.set ? "S" : ""),
-                                       data.rd,
-                                       op_2);
-                case OpCode::TST:
-                case OpCode::TEQ:
-                case OpCode::CMP:
-                case OpCode::CMN:
-                    return fmt::format(
-                      "{}{} R{:d},{}", data.opcode, condition, data.rn, op_2);
-                default:
-                    return fmt::format("{}{}{} R{:d},R{:d},{}",
-                                       data.opcode,
-                                       condition,
-                                       (data.set ? "S" : ""),
-                                       data.rd,
-                                       data.rn,
-                                       op_2);
-            }
-        },
-        [this](SoftwareInterrupt) { return fmt::format("SWI{}", condition); },
-        [this](CoprocessorDataTransfer& data) {
-            std::string expression = fmt::format(",#{:d}", data.offset);
-            return fmt::format(
-              "{}{}{} p{:d},c{:d},[R{:d}{}]{}",
-              (data.load ? "LDC" : "STC"),
-              condition,
-              (data.len ? "L" : ""),
-              data.cpn,
-              data.crd,
-              data.rn,
-              (data.pre ? expression : ""),
-              (data.pre ? (data.write ? "!" : "") : expression));
-        },
-        [this](CoprocessorDataOperation& data) {
-            return fmt::format("CDP{} p{},{},c{},c{},c{},{}",
-                               condition,
-                               data.cpn,
-                               data.cp_opc,
-                               data.crd,
-                               data.crn,
-                               data.crm,
-                               data.cp);
-        },
-        [this](CoprocessorRegisterTransfer& data) {
-            return fmt::format("{}{} p{},{},R{},c{},c{},{}",
-                               (data.load ? "MRC" : "MCR"),
-                               condition,
-                               data.cpn,
-                               data.cp_opc,
-                               data.rd,
-                               data.crn,
-                               data.crm,
-                               data.cp);
-        },
-        [](auto) { return std::string("unknown instruction"); } },
-      data);
-}
-
-std::ostream&
-operator<<(std::ostream& os, const DataProcessing::OpCode opcode) {
-
-#define CASE(opcode)                                                           \
-    case DataProcessing::OpCode::opcode:                                       \
-        os << #opcode;                                                         \
-        break;
-
-    switch (opcode) {
-        CASE(AND)
-        CASE(EOR)
-        CASE(SUB)
-        CASE(RSB)
-        CASE(ADD)
-        CASE(ADC)
-        CASE(SBC)
-        CASE(RSC)
-        CASE(TST)
-        CASE(TEQ)
-        CASE(CMP)
-        CASE(CMN)
-        CASE(ORR)
-        CASE(MOV)
-        CASE(BIC)
-        CASE(MVN)
-    }
-
-#undef CASE
-
-    return os;
-}
-}
 }

src/cpu/arm/instruction.hh

@@ -5,9 +5,9 @@
 #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>
 struct overloaded : Ts... {
     using Ts::operator()...;
@@ -113,6 +113,37 @@ struct DataProcessing {
     OpCode opcode;
 };
 
+constexpr auto
+stringify(DataProcessing::OpCode opcode) {
+
+#define CASE(opcode)                                                           \
+    case DataProcessing::OpCode::opcode:                                       \
+        return #opcode;
+
+    switch (opcode) {
+        CASE(AND)
+        CASE(EOR)
+        CASE(SUB)
+        CASE(RSB)
+        CASE(ADD)
+        CASE(ADC)
+        CASE(SBC)
+        CASE(RSC)
+        CASE(TST)
+        CASE(TEQ)
+        CASE(CMP)
+        CASE(CMN)
+        CASE(ORR)
+        CASE(MOV)
+        CASE(BIC)
+        CASE(MVN)
+    }
+
+#undef CASE
+
+    return "";
+}
+
 struct PsrTransfer {
     enum class Type {
         Mrs,
@@ -186,15 +217,8 @@ struct Instruction {
       : condition(condition)
       , data(data){};
 
+#ifdef DISASSEMBLER
     std::string disassemble();
+#endif
 };
-
-std::ostream&
-operator<<(std::ostream& os, const DataProcessing::OpCode cond);
-}
-}
-
-namespace fmt {
-template<>
-struct formatter<matar::arm::DataProcessing::OpCode> : ostream_formatter {};
 }

src/cpu/arm/meson.build

@@ -1,4 +1,8 @@
 lib_sources += files(
   'instruction.cc',
   'exec.cc'
-)
+)
+
+if get_option('disassembler')
+  lib_sources += files('disassembler.cc')
+endif

src/cpu/cpu-impl.cc

@@ -3,8 +3,7 @@
 #include "util/log.hh"
 #include <algorithm>
 #include <cstdio>
-
-using namespace logger;
+#include <type_traits>
 
 namespace matar {
 CpuImpl::CpuImpl(const Bus& bus) noexcept
@@ -19,7 +18,7 @@ CpuImpl::CpuImpl(const Bus& bus) noexcept
     cpsr.set_irq_disabled(true);
     cpsr.set_fiq_disabled(true);
     cpsr.set_state(State::Arm);
-    log_info("CPU successfully initialised");
+    glogger.info("CPU successfully initialised");
 
     // PC always points to two instructions ahead
     // PC - 2 is the instruction being executed
@@ -121,14 +120,15 @@ CpuImpl::step() {
     uint32_t cur_pc = pc - 2 * arm::INSTRUCTION_SIZE;
 
     if (cpsr.state() == State::Arm) {
-        debug(cur_pc);
         uint32_t x = bus->read_word(cur_pc);
         arm::Instruction instruction(x);
-        log_info("{:#034b}", x);
 
-        exec_arm(instruction);
+        exec(instruction);
 
-        log_info("0x{:08X} : {}", cur_pc, instruction.disassemble());
+#ifdef DISASSEMBLER
+        glogger.info("{:#034b}", x);
+        glogger.info("0x{:08X} : {}", cur_pc, instruction.disassemble());
+#endif
 
         if (is_flushed) {
             // if flushed, do not increment the PC, instead set it to two

src/cpu/cpu-impl.hh

@@ -13,7 +13,12 @@ class CpuImpl {
 
     void step();
     void chg_mode(const Mode to);
-    void exec_arm(const arm::Instruction instruction);
+    void exec(const arm::Instruction instruction);
+
+    // TODO: get rid of this
+#ifndef MATAR_CPU_TESTS
+  private:
+#endif
 
     static constexpr uint8_t GPR_COUNT = 16;
 

src/cpu/meson.build

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

src/cpu/psr.cc

@@ -96,37 +96,4 @@ Psr::condition(Condition cond) const {
 
     return false;
 }
-
-std::ostream&
-operator<<(std::ostream& os, const Condition cond) {
-
-#define CASE(cond)                                                             \
-    case Condition::cond:                                                      \
-        os << #cond;                                                           \
-        break;
-
-    switch (cond) {
-        CASE(EQ)
-        CASE(NE)
-        CASE(CS)
-        CASE(CC)
-        CASE(MI)
-        CASE(PL)
-        CASE(VS)
-        CASE(VC)
-        CASE(HI)
-        CASE(LS)
-        CASE(GE)
-        CASE(LT)
-        CASE(GT)
-        CASE(LE)
-        case Condition::AL: {
-            // empty
-        }
-    }
-
-#undef CASE
-
-    return os;
-}
 }

src/cpu/psr.hh

@@ -38,6 +38,38 @@ enum class Condition {
     AL = 0b1110
 };
 
+constexpr auto
+stringify(Condition cond) {
+
+#define CASE(cond)                                                             \
+    case Condition::cond:                                                      \
+        return #cond;
+
+    switch (cond) {
+        CASE(EQ)
+        CASE(NE)
+        CASE(CS)
+        CASE(CC)
+        CASE(MI)
+        CASE(PL)
+        CASE(VS)
+        CASE(VC)
+        CASE(HI)
+        CASE(LS)
+        CASE(GE)
+        CASE(LT)
+        CASE(GT)
+        CASE(LE)
+        case Condition::AL: {
+            // empty
+        }
+    }
+
+#undef CASE
+
+    return "";
+}
+
 class Psr {
   public:
     // clear the reserved bits i.e, [8:27]
@@ -88,13 +120,4 @@ class Psr {
 
     uint32_t psr;
 };
-
-// https://fmt.dev/dev/api.html#std-ostream-support
-std::ostream&
-operator<<(std::ostream& os, const Condition cond);
-}
-
-namespace fmt {
-template<>
-struct formatter<matar::Condition> : ostream_formatter {};
 }

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/memory.cc

@@ -1,13 +1,11 @@
 #include "memory.hh"
 #include "header.hh"
 #include "util/bits.hh"
+#include "util/crypto.hh"
 #include "util/log.hh"
-#include "util/utils.hh"
 #include <bitset>
 #include <stdexcept>
 
-using namespace logger;
-
 namespace matar {
 Memory::Memory(std::array<uint8_t, BIOS_SIZE>&& bios,
                std::vector<uint8_t>&& rom)
@@ -23,17 +21,17 @@ Memory::Memory(std::array<uint8_t, BIOS_SIZE>&& bios,
       "fd2547724b505f487e6dcb29ec2ecff3af35a841a77ab2e85fd87350abd36570";
 
     if (bios_hash != expected_hash) {
-        log_warn("BIOS hash failed to match, run at your own risk"
-                 "\nExpected : {} "
-                 "\nGot      : {}",
-                 expected_hash,
-                 bios_hash);
+        glogger.warn("BIOS hash failed to match, run at your own risk"
+                     "\nExpected : {} "
+                     "\nGot      : {}",
+                     expected_hash,
+                     bios_hash);
     }
 
     parse_header();
 
-    log_info("Memory successfully initialised");
-    log_info("Cartridge Title: {}", header.title);
+    glogger.info("Memory successfully initialised");
+    glogger.info("Cartridge Title: {}", header.title);
 };
 
 #define MATCHES(area) address >= area##_START&& address <= area##_END
@@ -59,7 +57,7 @@ Memory::read(size_t address) const {
     } else if (MATCHES(ROM_2)) {
         return rom[address - ROM_2_START];
     } else {
-        log_error("Invalid memory region accessed");
+        glogger.error("Invalid memory region accessed");
         return 0xFF;
     }
 }
@@ -85,49 +83,12 @@ Memory::write(size_t address, uint8_t byte) {
     } else if (MATCHES(ROM_2)) {
         rom[address - ROM_2_START] = byte;
     } else {
-        log_error("Invalid memory region accessed");
+        glogger.error("Invalid memory region accessed");
     }
 }
 
 #undef MATCHES
 
-uint16_t
-Memory::read_halfword(size_t address) const {
-    if (address & 0b01)
-        log_warn("Reading a non aligned halfword address");
-
-    return read(address) | read(address + 1) << 8;
-}
-
-void
-Memory::write_halfword(size_t address, uint16_t halfword) {
-    if (address & 0b01)
-        log_warn("Writing to a non aligned halfword address");
-
-    write(address, halfword & 0xFF);
-    write(address + 1, halfword >> 8 & 0xFF);
-}
-
-uint32_t
-Memory::read_word(size_t address) const {
-    if (address & 0b11)
-        log_warn("Reading a non aligned word address");
-
-    return read(address) | read(address + 1) << 8 | read(address + 2) << 16 |
-           read(address + 3) << 24;
-}
-
-void
-Memory::write_word(size_t address, uint32_t word) {
-    if (address & 0b11)
-        log_warn("Writing to a non aligned word address");
-
-    write(address, word & 0xFF);
-    write(address + 1, word >> 8 & 0xFF);
-    write(address + 2, word >> 16 & 0xFF);
-    write(address + 3, word >> 24 & 0xFF);
-}
-
 void
 Memory::parse_header() {
 
@@ -142,7 +103,7 @@ Memory::parse_header() {
 
     // nintendo logo
     if (rom[0x9C] != 0x21)
-        log_info("HEADER: BIOS debugger bits not set to 0");
+        glogger.info("HEADER: BIOS debugger bits not set to 0");
 
     // game info
     header.title = std::string(&rom[0xA0], &rom[0xA0 + 12]);
@@ -177,7 +138,7 @@ Memory::parse_header() {
             break;
 
         default:
-            log_error("HEADER: invalid unique code: {}", rom[0xAC]);
+            glogger.error("HEADER: invalid unique code: {}", rom[0xAC]);
     }
 
     header.title_code = std::string(&rom[0xAD], &rom[0xAE]);
@@ -206,15 +167,16 @@ Memory::parse_header() {
             break;
 
         default:
-            log_error("HEADER: invalid destination/language: {}", rom[0xAF]);
+            glogger.error("HEADER: invalid destination/language: {}",
+                          rom[0xAF]);
     }
 
     if (rom[0xB2] != 0x96)
-        log_error("HEADER: invalid fixed byte at 0xB2");
+        glogger.error("HEADER: invalid fixed byte at 0xB2");
 
     for (size_t i = 0xB5; i < 0xBC; i++) {
         if (rom[i] != 0x00)
-            log_error("HEADER: invalid fixed bytes at 0xB5");
+            glogger.error("HEADER: invalid fixed bytes at 0xB5");
     }
 
     header.version = rom[0xBC];
@@ -228,7 +190,7 @@ Memory::parse_header() {
         chk &= 0xFF;
 
         if (chk != rom[0xBD])
-            log_error("HEADER: checksum does not match");
+            glogger.error("HEADER: checksum does not match");
     }
 
     // multiboot not required right now

src/meson.build

@@ -3,15 +3,19 @@ lib_sources = files(
   'bus.cc'
 )
 
+subdir('util')
 subdir('cpu')
 
-
-lib_cpp_args = [ ]
+lib_cpp_args = []
 
 fmt = dependency('fmt', version : '>=10.1.0', static: true)
 if not fmt.found()
   fmt = dependency('fmt', version : '>=10.1.0', static: false)
-  lib_cpp_args += 'DFMT_HEADER_ONLY'
+  lib_cpp_args += '-DFMT_HEADER_ONLY'
+endif
+
+if get_option('disassembler')
+  lib_cpp_args += '-DDISASSEMBLER'
 endif
 
 lib = library(

src/util/bits.hh

@@ -14,19 +14,19 @@ get_bit(Int num, size_t n) {
 template<std::integral Int>
 inline void
 set_bit(Int& num, size_t n) {
-    num |= (1 << n);
+    num |= (static_cast<Int>(1) << n);
 }
 
 template<std::integral Int>
 inline void
 rst_bit(Int& num, size_t n) {
-    num &= ~(1 << n);
+    num &= ~(static_cast<Int>(1) << n);
 }
 
 template<std::integral Int>
 inline void
 chg_bit(Int& num, size_t n, bool x) {
-    num = (num & ~(1 << n)) | (x << n);
+    num = (num & ~(static_cast<Int>(1) << n)) | (static_cast<Int>(x) << n);
 }
 
 /// read range of bits from start to end inclusive
@@ -36,5 +36,5 @@ bit_range(Int num, size_t start, size_t end) {
     // NOTE: we do not require -1 if it is a signed integral
     Int left =
       std::numeric_limits<Int>::digits - (std::is_unsigned<Int>::value) - end;
-    return num << left >> (left + start);
+    return static_cast<Int>(num << left) >> (left + start);
 }

src/util/log.cc

@@ -0,0 +1,8 @@
+#include "log.hh"
+
+logging::Logger glogger = logging::Logger();
+
+void
+matar::set_log_level(LogLevel level) {
+    glogger.set_level(level);
+}

src/util/log.hh

@@ -1,58 +1,83 @@
 #pragma once
 
+#include "util/loglevel.hh"
 #include <fmt/ostream.h>
 #include <iostream>
 
-using fmt::print;
-using std::clog;
-
-namespace logger {
+namespace logging {
 namespace ansi {
-static constexpr std::string_view RED     = "\033[31m";
-static constexpr std::string_view YELLOW  = "\033[33m";
-static constexpr std::string_view MAGENTA = "\033[35m";
-static constexpr std::string_view WHITE   = "\033[37m";
-static constexpr std::string_view BOLD    = "\033[1m";
-static constexpr std::string_view RESET   = "\033[0m";
+static constexpr auto RED     = "\033[31m";
+static constexpr auto YELLOW  = "\033[33m";
+static constexpr auto MAGENTA = "\033[35m";
+static constexpr auto WHITE   = "\033[37m";
+static constexpr auto BOLD    = "\033[1m";
+static constexpr auto RESET   = "\033[0m";
 }
 
-template<typename... Args>
-inline void
-log_raw(const fmt::format_string<Args...>& fmt, Args&&... args) {
-    fmt::println(clog, fmt, std::forward<Args>(args)...);
+using fmt::print;
+
+class Logger {
+    using LogLevel = matar::LogLevel;
+
+  public:
+    Logger(LogLevel level = LogLevel::Debug, FILE* stream = stderr)
+      : level(0)
+      , stream(stream) {
+        set_level(level);
+    }
+
+    template<typename... Args>
+    void log(const fmt::format_string<Args...>& fmt, Args&&... args) {
+        fmt::println(stream, fmt, std::forward<Args>(args)...);
+    }
+
+    template<typename... Args>
+    void debug(const fmt::format_string<Args...>& fmt, Args&&... args) {
+        if (level & static_cast<uint8_t>(LogLevel::Debug)) {
+            print(stream, "{}{}[DEBUG] ", ansi::MAGENTA, ansi::BOLD);
+            log(fmt, std::forward<Args>(args)...);
+            print(stream, ansi::RESET);
+        }
+    }
+
+    template<typename... Args>
+    void info(const fmt::format_string<Args...>& fmt, Args&&... args) {
+        if (level & static_cast<uint8_t>(LogLevel::Info)) {
+            print(stream, "{}[INFO] ", ansi::WHITE);
+            log(fmt, std::forward<Args>(args)...);
+            print(stream, ansi::RESET);
+        }
+    }
+
+    template<typename... Args>
+    void warn(const fmt::format_string<Args...>& fmt, Args&&... args) {
+        if (level & static_cast<uint8_t>(LogLevel::Warn)) {
+            print(stream, "{}[WARN] ", ansi::YELLOW);
+            log(fmt, std::forward<Args>(args)...);
+            print(stream, ansi::RESET);
+        }
+    }
+
+    template<typename... Args>
+    void error(const fmt::format_string<Args...>& fmt, Args&&... args) {
+        if (level & static_cast<uint8_t>(LogLevel::Error)) {
+            print(stream, "{}{}[ERROR] ", ansi::RED, ansi::BOLD);
+            log(fmt, std::forward<Args>(args)...);
+            print(stream, ansi::RESET);
+        }
+    }
+
+    void set_level(LogLevel level) {
+        this->level = (static_cast<uint8_t>(level) << 1) - 1;
+    }
+    void set_stream(FILE* stream) { this->stream = stream; }
+
+  private:
+    uint8_t level;
+    FILE* stream;
+};
 }
 
-template<typename... Args>
-inline void
-log_debug(const fmt::format_string<Args...>& fmt, Args&&... args) {
-    print(clog, "{}{}[DEBUG] ", ansi::MAGENTA, ansi::BOLD);
-    log_raw(fmt, std::forward<Args>(args)...);
-    print(clog, ansi::RESET);
-}
+extern logging::Logger glogger;
 
-template<typename... Args>
-inline void
-log_info(const fmt::format_string<Args...>& fmt, Args&&... args) {
-    print(clog, "{}[INFO] ", ansi::WHITE);
-    log_raw(fmt, std::forward<Args>(args)...);
-    print(clog, ansi::RESET);
-}
-
-template<typename... Args>
-inline void
-log_warn(const fmt::format_string<Args...>& fmt, Args&&... args) {
-    print(clog, "{}[WARN] ", ansi::YELLOW);
-    log_raw(fmt, std::forward<Args>(args)...);
-    print(clog, ansi::RESET);
-}
-
-template<typename... Args>
-inline void
-log_error(const fmt::format_string<Args...>& fmt, Args&&... args) {
-    print(clog, "{}{}[ERROR] ", ansi::RED, ansi::BOLD);
-    log_raw(fmt, std::forward<Args>(args)...);
-    print(clog, ansi::RESET);
-}
-}
-
-#define debug(value) logger::log_debug("{} = {}", #value, value)
+#define dbg(x) glogger.debug("{} = {}", #x, x);

src/util/meson.build

@@ -0,0 +1,3 @@
+lib_sources += files(
+  'log.cc'
+)

tests/bus.cc

@@ -0,0 +1,43 @@
+#include "bus.hh"
+#include <catch2/catch_test_macros.hpp>
+
+static constexpr auto TAG = "[bus]";
+
+using namespace matar;
+
+class BusFixture {
+  public:
+    BusFixture()
+      : bus(Memory(std::array<uint8_t, Memory::BIOS_SIZE>(),
+                   std::vector<uint8_t>(Header::HEADER_SIZE))) {}
+
+  protected:
+    Bus bus;
+};
+
+TEST_CASE_METHOD(BusFixture, "Byte", TAG) {
+    CHECK(bus.read_byte(3349) == 0);
+
+    bus.write_byte(3349, 0xEC);
+    CHECK(bus.read_byte(3349) == 0xEC);
+    CHECK(bus.read_word(3349) == 0xEC);
+    CHECK(bus.read_halfword(3349) == 0xEC);
+}
+
+TEST_CASE_METHOD(BusFixture, "Halfword", TAG) {
+    CHECK(bus.read_halfword(33750745) == 0);
+
+    bus.write_halfword(33750745, 0x1A4A);
+    CHECK(bus.read_halfword(33750745) == 0x1A4A);
+    CHECK(bus.read_word(33750745) == 0x1A4A);
+    CHECK(bus.read_byte(33750745) == 0x4A);
+}
+
+TEST_CASE_METHOD(BusFixture, "Word", TAG) {
+    CHECK(bus.read_word(100724276) == 0);
+
+    bus.write_word(100724276, 0x3ACC491D);
+    CHECK(bus.read_word(100724276) == 0x3ACC491D);
+    CHECK(bus.read_halfword(100724276) == 0x491D);
+    CHECK(bus.read_byte(100724276) == 0x1D);
+}

tests/cpu/arm/exec.cc

@@ -1,4 +1,7 @@
+#define MATAR_CPU_TESTS
 #include "cpu/cpu-impl.hh"
+#undef MATAR_CPU_TESTS
+
 #include "util/bits.hh"
 #include <catch2/catch_test_macros.hpp>
 #include <limits>
@@ -13,10 +16,9 @@ class CpuFixture {
                        std::vector<uint8_t>(Header::HEADER_SIZE)))) {}
 
   protected:
-    // TODO: test with other conditions
     void exec(arm::InstructionData data, Condition condition = Condition::AL) {
         arm::Instruction instruction(condition, data);
-        cpu.exec_arm(instruction);
+        cpu.exec(instruction);
     }
 
     void reset(uint32_t value = 0) {
@@ -32,7 +34,7 @@ class CpuFixture {
     };
 };
 
-#define TAG "arm execution"
+static constexpr auto TAG = "[arm][execution]";
 
 using namespace arm;
 
@@ -333,7 +335,7 @@ TEST_CASE_METHOD(CpuFixture, "Single Data Transfer", TAG) {
 
     // r15 as rn
     {
-        data_transfer->rn = 15;
+        data_transfer->rn = cpu.PC_INDEX;
         cpu.gpr[15]       = 7577;
 
         exec(data);
@@ -349,7 +351,7 @@ TEST_CASE_METHOD(CpuFixture, "Single Data Transfer", TAG) {
     // r15 as rd
     {
         // 4088
-        data_transfer->rd = 15;
+        data_transfer->rd = cpu.PC_INDEX;
         cpu.gpr[15]       = 444444;
 
         exec(data);
@@ -466,7 +468,7 @@ TEST_CASE_METHOD(CpuFixture, "Halfword Transfer", TAG) {
 
     // r15 as rn
     {
-        hw_transfer->rn = 15;
+        hw_transfer->rn = cpu.PC_INDEX;
         cpu.gpr[15]     = 399;
 
         exec(data);
@@ -482,7 +484,7 @@ TEST_CASE_METHOD(CpuFixture, "Halfword Transfer", TAG) {
 
     // r15 as rd
     {
-        hw_transfer->rd = 15;
+        hw_transfer->rd = cpu.PC_INDEX;
         cpu.gpr[15]     = 224;
 
         exec(data);
@@ -793,7 +795,7 @@ TEST_CASE_METHOD(CpuFixture, "Data Processing", TAG) {
 
     // same as above but with rn (oprerand 1) = 15
     {
-        processing->rn = 15;
+        processing->rn = cpu.PC_INDEX;
         cpu.gpr[15]    = -2871;
         exec(data);
 
@@ -804,29 +806,41 @@ TEST_CASE_METHOD(CpuFixture, "Data Processing", TAG) {
         processing->rn = 7;
     }
 
-    auto flags = [this](bool n, bool z, bool v, bool c) {
-        CHECK(cpu.cpsr.n() == n);
-        CHECK(cpu.cpsr.z() == z);
-        CHECK(cpu.cpsr.v() == v);
-        CHECK(cpu.cpsr.c() == c);
-
+    auto reset_flags = [this]() {
         cpu.cpsr.set_n(false);
         cpu.cpsr.set_z(false);
         cpu.cpsr.set_v(false);
         cpu.cpsr.set_c(false);
     };
 
+    auto flags = [this, reset_flags](bool n, bool z, bool v, bool c) {
+        CHECK(cpu.cpsr.n() == n);
+        CHECK(cpu.cpsr.z() == z);
+        CHECK(cpu.cpsr.v() == v);
+        CHECK(cpu.cpsr.c() == c);
+        reset_flags();
+    };
+
     // immediate operand
     processing->operand = static_cast<uint32_t>(54924809);
     // rs
     cpu.gpr[12] = 2;
     cpu.gpr[5]  = 0;
+    reset_flags();
 
-    SECTION("AND") {
+    SECTION("AND (with condition check)") {
         processing->opcode = OpCode::AND;
-        exec(data);
+        cpu.cpsr.set_z(false);
+        exec(data, Condition::EQ);
+
+        // condition is false
+        CHECK(cpu.gpr[5] == 0);
+
+        cpu.cpsr.set_z(true);
+        exec(data, Condition::EQ);
 
         // -28717 & 54924809
+        // condition is true now
         CHECK(cpu.gpr[5] == 54920705);
 
         // check set flags
@@ -846,11 +860,19 @@ TEST_CASE_METHOD(CpuFixture, "Data Processing", TAG) {
         flags(false, false, false, false);
     }
 
-    SECTION("EOR") {
+    SECTION("EOR (with condition check)") {
         processing->opcode = OpCode::EOR;
-        exec(data);
+        cpu.cpsr.set_c(true);
+        exec(data, Condition::CC);
+
+        // condition fails
+        CHECK(cpu.gpr[5] == 0);
+
+        cpu.cpsr.set_c(false);
+        exec(data, Condition::CC);
 
         // -28717 ^ 54924809
+        // condition is true now
         CHECK(cpu.gpr[5] == 4240021978);
 
         // check set flags
@@ -1038,7 +1060,7 @@ TEST_CASE_METHOD(CpuFixture, "Data Processing", TAG) {
 
     SECTION("R15 as destination") {
         processing->opcode = OpCode::MVN;
-        processing->rd     = 15;
+        processing->rd     = cpu.PC_INDEX;
         cpu.gpr[15]        = 0;
         CHECK(cpu.spsr.raw() != cpu.cpsr.raw());
         exec(data);
@@ -1051,5 +1073,3 @@ TEST_CASE_METHOD(CpuFixture, "Data Processing", TAG) {
         CHECK(cpu.spsr.raw() == cpu.cpsr.raw());
     }
 }
-
-#undef TAG

tests/cpu/arm/instruction.cc

@@ -1,7 +1,7 @@
 #include "cpu/arm/instruction.hh"
 #include <catch2/catch_test_macros.hpp>
 
-#define TAG "disassembler"
+static constexpr auto TAG = "[arm][disassembly]";
 
 using namespace matar;
 using namespace arm;
@@ -16,7 +16,9 @@ TEST_CASE("Branch and Exchange", TAG) {
 
     CHECK(bx->rn == 10);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "BXGT R10");
+#endif
 }
 
 TEST_CASE("Branch", TAG) {
@@ -33,10 +35,12 @@ TEST_CASE("Branch", TAG) {
     CHECK(b->offset == 0xFE15FF14);
     CHECK(b->link == true);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "BL 0xFE15FF14");
 
     b->link = false;
     CHECK(instruction.disassemble() == "B 0xFE15FF14");
+#endif
 }
 
 TEST_CASE("Multiply", TAG) {
@@ -54,11 +58,13 @@ TEST_CASE("Multiply", TAG) {
     CHECK(mul->acc == true);
     CHECK(mul->set == true);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "MLAEQS R10,R0,R15,R14");
 
     mul->acc = false;
     mul->set = false;
     CHECK(instruction.disassemble() == "MULEQ R10,R0,R15");
+#endif
 }
 
 TEST_CASE("Multiply Long", TAG) {
@@ -77,6 +83,7 @@ TEST_CASE("Multiply Long", TAG) {
     CHECK(mull->set == true);
     CHECK(mull->uns == true);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "UMULLNES R7,R14,R2,R6");
 
     mull->acc = true;
@@ -85,6 +92,7 @@ TEST_CASE("Multiply Long", TAG) {
     mull->uns = false;
     mull->set = false;
     CHECK(instruction.disassemble() == "SMLALNE R7,R14,R2,R6");
+#endif
 }
 
 TEST_CASE("Undefined", TAG) {
@@ -94,7 +102,10 @@ TEST_CASE("Undefined", TAG) {
     Instruction instruction(raw);
 
     CHECK(instruction.condition == Condition::AL);
+
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "UND");
+#endif
 }
 
 TEST_CASE("Single Data Swap", TAG) {
@@ -110,10 +121,12 @@ TEST_CASE("Single Data Swap", TAG) {
     CHECK(swp->rn == 9);
     CHECK(swp->byte == false);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "SWPGE R5,R6,[R9]");
 
     swp->byte = true;
     CHECK(instruction.disassemble() == "SWPGEB R5,R6,[R9]");
+#endif
 }
 
 TEST_CASE("Single Data Transfer", TAG) {
@@ -138,6 +151,7 @@ TEST_CASE("Single Data Transfer", TAG) {
     CHECK(ldr->up == true);
     CHECK(ldr->pre == true);
 
+#ifdef DISASSEMBLER
     ldr->load        = true;
     ldr->byte        = true;
     ldr->write       = false;
@@ -153,6 +167,7 @@ TEST_CASE("Single Data Transfer", TAG) {
 
     ldr->pre = true;
     CHECK(instruction.disassemble() == "LDRB R10,[R2,-#9023]");
+#endif
 }
 
 TEST_CASE("Halfword Transfer", TAG) {
@@ -176,6 +191,7 @@ TEST_CASE("Halfword Transfer", TAG) {
     CHECK(ldr->up == true);
     CHECK(ldr->pre == true);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "STRCCH R2,[R15,+R6]!");
 
     ldr->pre  = false;
@@ -193,6 +209,7 @@ TEST_CASE("Halfword Transfer", TAG) {
     ldr->imm    = 1;
     ldr->offset = 90;
     CHECK(instruction.disassemble() == "STRCCSB R2,[R15],-#90");
+#endif
 }
 
 TEST_CASE("Block Data Transfer", TAG) {
@@ -223,6 +240,7 @@ TEST_CASE("Block Data Transfer", TAG) {
     CHECK(ldm->up == false);
     CHECK(ldm->pre == true);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "LDMLSDB R7,{R0,R2,R3,R5,R6,R8,R14}^");
 
     ldm->write = true;
@@ -238,6 +256,7 @@ TEST_CASE("Block Data Transfer", TAG) {
     ldm->pre  = false;
 
     CHECK(instruction.disassemble() == "STMLSIA R7!,{R0,R2,R5,R14}");
+#endif
 }
 
 TEST_CASE("PSR Transfer", TAG) {
@@ -256,7 +275,9 @@ TEST_CASE("PSR Transfer", TAG) {
         CHECK(mrs->operand == 10);
         CHECK(mrs->spsr == true);
 
+#ifdef DISASSEMBLER
         CHECK(instruction.disassemble() == "MRSMI R10,SPSR_all");
+#endif
     }
 
     SECTION("MSR") {
@@ -272,7 +293,9 @@ TEST_CASE("PSR Transfer", TAG) {
         CHECK(msr->operand == 8);
         CHECK(msr->spsr == false);
 
+#ifdef DISASSEMBLER
         CHECK(instruction.disassemble() == "MSR CPSR_all,R8");
+#endif
     }
 
     SECTION("MSR_flg with register operand") {
@@ -287,7 +310,9 @@ TEST_CASE("PSR Transfer", TAG) {
         CHECK(msr->operand == 8);
         CHECK(msr->spsr == false);
 
+#ifdef DISASSEMBLER
         CHECK(instruction.disassemble() == "MSRVS CPSR_flg,R8");
+#endif
     }
 
     SECTION("MSR_flg with immediate operand") {
@@ -304,7 +329,9 @@ TEST_CASE("PSR Transfer", TAG) {
         CHECK(msr->operand == 27262976);
         CHECK(msr->spsr == true);
 
+#ifdef DISASSEMBLER
         CHECK(instruction.disassemble() == "MSR SPSR_flg,#27262976");
+#endif
     }
 }
 
@@ -331,6 +358,7 @@ TEST_CASE("Data Processing", TAG) {
     CHECK(alu->set == true);
     CHECK(alu->opcode == OpCode::AND);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "ANDS R7,R14,R1,ROR #22");
 
     shift->data.immediate = false;
@@ -392,6 +420,7 @@ TEST_CASE("Data Processing", TAG) {
         alu->opcode = OpCode::MVN;
         CHECK(instruction.disassemble() == "MVN R7,#3300012");
     }
+#endif
 }
 
 TEST_CASE("Coprocessor Data Transfer", TAG) {
@@ -412,6 +441,7 @@ TEST_CASE("Coprocessor Data Transfer", TAG) {
     CHECK(ldc->up == true);
     CHECK(ldc->pre == true);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "STCGE p1,c15,[R5,#70]!");
 
     ldc->load  = true;
@@ -420,6 +450,7 @@ TEST_CASE("Coprocessor Data Transfer", TAG) {
     ldc->len   = true;
 
     CHECK(instruction.disassemble() == "LDCGEL p1,c15,[R5],#70");
+#endif
 }
 
 TEST_CASE("Coprocessor Operand Operation", TAG) {
@@ -437,7 +468,9 @@ TEST_CASE("Coprocessor Operand Operation", TAG) {
     CHECK(cdp->crn == 5);
     CHECK(cdp->cp_opc == 10);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "CDP p1,10,c15,c5,c6,2");
+#endif
 }
 
 TEST_CASE("Coprocessor Register Transfer", TAG) {
@@ -457,7 +490,9 @@ TEST_CASE("Coprocessor Register Transfer", TAG) {
     CHECK(mrc->load == false);
     CHECK(mrc->cp_opc == 5);
 
+#ifdef DISASSEMBLER
     CHECK(instruction.disassemble() == "MCR p1,5,R15,c5,c6,2");
+#endif
 }
 
 TEST_CASE("Software Interrupt", TAG) {
@@ -465,7 +500,8 @@ TEST_CASE("Software Interrupt", TAG) {
     Instruction instruction(raw);
 
     CHECK(instruction.condition == Condition::EQ);
-    CHECK(instruction.disassemble() == "SWIEQ");
-}
 
-#undef TAG
+#ifdef DISASSEMBLER
+    CHECK(instruction.disassemble() == "SWIEQ");
+#endif
+}

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'
+)

tests/main.cc

@@ -0,0 +1,8 @@
+#include "util/loglevel.hh"
+#include <catch2/catch_session.hpp>
+
+int
+main(int argc, char* argv[]) {
+    matar::set_log_level(matar::LogLevel::Off);
+    return Catch::Session().run(argc, argv);
+}

tests/memory.cc

@@ -0,0 +1,121 @@
+#include "memory.hh"
+#include <catch2/catch_test_macros.hpp>
+
+static constexpr auto TAG = "[memory]";
+
+using namespace matar;
+
+class MemFixture {
+  public:
+    MemFixture()
+      : memory(std::array<uint8_t, Memory::BIOS_SIZE>(),
+               std::vector<uint8_t>(Header::HEADER_SIZE)) {}
+
+  protected:
+    Memory memory;
+};
+
+TEST_CASE_METHOD(MemFixture, "bios", TAG) {
+    memory.write(0, 0xAC);
+    CHECK(memory.read(0) == 0xAC);
+
+    memory.write(0x3FFF, 0x48);
+    CHECK(memory.read(0x3FFF) == 0x48);
+
+    memory.write(0x2A56, 0x10);
+    CHECK(memory.read(0x2A56) == 0x10);
+}
+
+TEST_CASE_METHOD(MemFixture, "board wram", TAG) {
+    memory.write(0x2000000, 0xAC);
+    CHECK(memory.read(0x2000000) == 0xAC);
+
+    memory.write(0x203FFFF, 0x48);
+    CHECK(memory.read(0x203FFFF) == 0x48);
+
+    memory.write(0x2022A56, 0x10);
+    CHECK(memory.read(0x2022A56) == 0x10);
+}
+
+TEST_CASE_METHOD(MemFixture, "chip wram", TAG) {
+    memory.write(0x3000000, 0xAC);
+    CHECK(memory.read(0x3000000) == 0xAC);
+
+    memory.write(0x3007FFF, 0x48);
+    CHECK(memory.read(0x3007FFF) == 0x48);
+
+    memory.write(0x3002A56, 0x10);
+    CHECK(memory.read(0x3002A56) == 0x10);
+}
+
+TEST_CASE_METHOD(MemFixture, "palette ram", TAG) {
+    memory.write(0x5000000, 0xAC);
+    CHECK(memory.read(0x5000000) == 0xAC);
+
+    memory.write(0x50003FF, 0x48);
+    CHECK(memory.read(0x50003FF) == 0x48);
+
+    memory.write(0x5000156, 0x10);
+    CHECK(memory.read(0x5000156) == 0x10);
+}
+
+TEST_CASE_METHOD(MemFixture, "video ram", TAG) {
+    memory.write(0x6000000, 0xAC);
+    CHECK(memory.read(0x6000000) == 0xAC);
+
+    memory.write(0x6017FFF, 0x48);
+    CHECK(memory.read(0x6017FFF) == 0x48);
+
+    memory.write(0x6012A56, 0x10);
+    CHECK(memory.read(0x6012A56) == 0x10);
+}
+
+TEST_CASE_METHOD(MemFixture, "oam obj ram", TAG) {
+    memory.write(0x7000000, 0xAC);
+    CHECK(memory.read(0x7000000) == 0xAC);
+
+    memory.write(0x70003FF, 0x48);
+    CHECK(memory.read(0x70003FF) == 0x48);
+
+    memory.write(0x7000156, 0x10);
+    CHECK(memory.read(0x7000156) == 0x10);
+}
+
+TEST_CASE("rom", TAG) {
+    // 32 megabyte ROM
+    Memory memory(std::array<uint8_t, Memory::BIOS_SIZE>(),
+                  std::vector<uint8_t>(32 * 1024 * 1024));
+
+    SECTION("ROM1") {
+        memory.write(0x8000000, 0xAC);
+        CHECK(memory.read(0x8000000) == 0xAC);
+
+        memory.write(0x9FFFFFF, 0x48);
+        CHECK(memory.read(0x9FFFFFF) == 0x48);
+
+        memory.write(0x8ef0256, 0x10);
+        CHECK(memory.read(0x8ef0256) == 0x10);
+    }
+
+    SECTION("ROM2") {
+        memory.write(0xA000000, 0xAC);
+        CHECK(memory.read(0xA000000) == 0xAC);
+
+        memory.write(0xBFFFFFF, 0x48);
+        CHECK(memory.read(0xBFFFFFF) == 0x48);
+
+        memory.write(0xAEF0256, 0x10);
+        CHECK(memory.read(0xAEF0256) == 0x10);
+    }
+
+    SECTION("ROM3") {
+        memory.write(0xC000000, 0xAC);
+        CHECK(memory.read(0xC000000) == 0xAC);
+
+        memory.write(0xDFFFFFF, 0x48);
+        CHECK(memory.read(0xDFFFFFF) == 0x48);
+
+        memory.write(0xCEF0256, 0x10);
+        CHECK(memory.read(0xCEF0256) == 0x10);
+    }
+}

tests/meson.build

@@ -4,11 +4,22 @@ tests_deps = [
 
 src = include_directories('../src')
 
-tests_sources = files()
+tests_sources = files(
+  'main.cc',
+  'bus.cc',
+  'memory.cc'
+)
 
 subdir('cpu')
+subdir('util')
 
-catch2 = dependency('catch2-with-main', version: '>=3.4.0', static: true)
+tests_cpp_args = []
+
+if get_option('disassembler')
+  tests_cpp_args += '-DDISASSEMBLER'
+endif
+
+catch2 = dependency('catch2', version: '>=3.4.0', static: true)
 catch2_tests = executable(
   'matar_tests',
   tests_sources,
@@ -16,6 +27,7 @@ catch2_tests = executable(
   link_with: tests_deps,
   include_directories: [inc, src],
   build_by_default: false,
+  cpp_args: tests_cpp_args
 )
 
 test('catch2 tests', catch2_tests)

tests/util/bits.cc

@@ -0,0 +1,106 @@
+#include "util/bits.hh"
+#include <catch2/catch_test_macros.hpp>
+
+static constexpr auto TAG = "[util][bits]";
+
+TEST_CASE("8 bits", TAG) {
+    uint8_t num = 45;
+
+    CHECK(get_bit(num, 0));
+    CHECK(!get_bit(num, 1));
+    CHECK(get_bit(num, 5));
+    CHECK(!get_bit(num, 6));
+    CHECK(!get_bit(num, 7));
+
+    set_bit(num, 6);
+    CHECK(get_bit(num, 6));
+
+    rst_bit(num, 6);
+    CHECK(!get_bit(num, 6));
+
+    chg_bit(num, 5, false);
+    CHECK(!get_bit(num, 5));
+
+    chg_bit(num, 5, true);
+    CHECK(get_bit(num, 5));
+
+    // 0b0110
+    CHECK(bit_range(num, 1, 4) == 6);
+}
+
+TEST_CASE("16 bits", TAG) {
+    uint16_t num = 34587;
+
+    CHECK(get_bit(num, 0));
+    CHECK(get_bit(num, 1));
+    CHECK(!get_bit(num, 5));
+    CHECK(!get_bit(num, 14));
+    CHECK(get_bit(num, 15));
+
+    set_bit(num, 14);
+    CHECK(get_bit(num, 14));
+
+    rst_bit(num, 14);
+    CHECK(!get_bit(num, 14));
+
+    chg_bit(num, 5, true);
+    CHECK(get_bit(num, 5));
+
+    // num = 45
+    chg_bit(num, 5, false);
+    CHECK(!get_bit(num, 5));
+
+    // 0b1000110
+    CHECK(bit_range(num, 2, 8) == 70);
+}
+
+TEST_CASE("32 bits", TAG) {
+    uint32_t num = 3194142523;
+
+    CHECK(get_bit(num, 0));
+    CHECK(get_bit(num, 1));
+    CHECK(get_bit(num, 12));
+    CHECK(get_bit(num, 29));
+    CHECK(!get_bit(num, 30));
+    CHECK(get_bit(num, 31));
+
+    set_bit(num, 30);
+    CHECK(get_bit(num, 30));
+
+    rst_bit(num, 30);
+    CHECK(!get_bit(num, 30));
+
+    chg_bit(num, 12, false);
+    CHECK(!get_bit(num, 12));
+
+    chg_bit(num, 12, true);
+    CHECK(get_bit(num, 12));
+
+    // 0b10011000101011111100111
+    CHECK(bit_range(num, 3, 25) == 5003239);
+}
+
+TEST_CASE("64 bits", TAG) {
+    uint64_t num = 58943208889991935;
+
+    CHECK(get_bit(num, 0));
+    CHECK(get_bit(num, 1));
+    CHECK(!get_bit(num, 10));
+    CHECK(get_bit(num, 55));
+    CHECK(!get_bit(num, 60));
+
+    set_bit(num, 63);
+    CHECK(get_bit(num, 63));
+
+    rst_bit(num, 63);
+    CHECK(!get_bit(num, 63));
+
+    chg_bit(num, 10, true);
+    CHECK(get_bit(num, 10));
+
+    chg_bit(num, 10, false);
+    CHECK(!get_bit(num, 10));
+
+    // 0b011010001
+    CHECK(bit_range(num, 39, 47) == 209);
+}

tests/util/crypto.cc

@@ -0,0 +1,21 @@
+#include "util/crypto.hh"
+#include <catch2/catch_test_macros.hpp>
+
+static constexpr auto TAG = "[util][crypto]";
+
+TEST_CASE("sha256 matar", TAG) {
+    std::array<uint8_t, 5> data = { 'm', 'a', 't', 'a', 'r' };
+
+    CHECK(crypto::sha256(data) ==
+          "3b02a908fd5743c0e868675bb6ae77d2a62b3b5f7637413238e2a1e0e94b6a53");
+}
+
+TEST_CASE("sha256 forgis", TAG) {
+    std::array<uint8_t, 32> data = { 'i', ' ', 'p', 'u', 't', ' ', 't', 'h',
+                                     'e', ' ', 'n', 'e', 'w', ' ', 'f', 'o',
+                                     'r', 'g', 'i', 's', ' ', 'o', 'n', ' ',
+                                     't', 'h', 'e', ' ', 'j', 'e', 'e', 'p' };
+
+    CHECK(crypto::sha256(data) ==
+          "cfddca2ce2673f355518cbe2df2a8522693c54723a469e8b36a4f68b90d2b759");
+}

tests/util/meson.build

@@ -0,0 +1,4 @@
+tests_sources += files(
+  'bits.cc',
+  'crypto.cc'
+)